Upstart Holdings Aktie

All right. We'll go ahead and get started here. Thank you very much for joining us this afternoon with Paul, Co-Founder and CEO of Upstart. I'm James Faucette, senior fintech analyst here at Morgan Stanley.

And before we get started with Paul, a couple of disclosures I need to read. First, our disclosure for Morgan Stanley. Please see the Morgan Stanley research disclosure website at morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative.

And then also for the -- for Upstart, just A reminder, today's discussion may contain forward-looking statements that relate to future results and events, which are based on Upstart's information available as of today and are subject to risks and uncertainties. Actual results may differ materially from those forward-looking statements. The discussion may also include non-GAAP financial measures, which are not a suitable -- not a substitute, excuse me, for GAAP results. Please refer to the company's filings with the SEC and its IR website for additional information, including GAAP to non-GAAP reconciliations, along with other disclosures.

So with that out of the way, Paul, thanks for being here. Appreciate it.

Thanks for having me.

Yes. So Just -- I wanted to kick off, I mean, you've obviously been at Upstart since the beginning, Co-founder, et cetera, stepping into a new role as CEO. Love to just get a minute or 2 of your reflections on that journey thus far with Upstart and then the transition and kind of the implications for you? And maybe what shareholders should be aware of?

Yes. Well, Dave and I started this company back in 2012, over 14 years ago. So we've been doing this for a while and working in close partnership for that entire time. I was very early in my career at that time, recent college dropout with Peter Thiel's 20 -- under 20 program. And so it was a real privilege for me to be able to learn from Dave all these years. And as we build the business together through so many stages, I think we chose a business that was probably unusually hard to build. I think it was -- a lot of the ways in which it was hard. We didn't even realize when we started just because essentially, we chose this market, that kind of made a lot of sense. This consumer lending market, we said, hey, you can go into this market. And what we're going to do is we've noticed that the sort of existing traditional players haven't really done much to adopt machine learning technologies or new sources of data. And if we use those things, then we're going to be able to pick up a bunch of advantage in being able to underwrite more people at much better rates, much faster than a traditional players can. It just seems like a straightforward application of technology to an industry where there's a lot of opportunity, let's just go do it. And what we learned after we started doing this is that, well, there's a bunch of reasons that no one wants to innovate in the space along these dimensions, and it comes down to the fact that there are a lot of a lot of sort of third parties that have to believe that your thing works. So if you want to be in this business, obviously, you have to be able to attract loan funding. That loan funding often depends on financing sources, which, in turn, often depend on ratings from third-party rating agencies. There is often a high degree of regulation. A lot of the institutions involved that are heavily regulated, financial institutions that have to show their regulators that what they're doing makes sense.

And on the technical side, maybe not so surprisingly, you can build these models and what happens is the first version of these models -- invariably, they're bad. They just -- they make bad loans and you lose money on the loans. And so now you're sort of in the situation where you're going to be going through this multiyear process of iterating through bad models in order to collect the training data you need to build good models, all while there are all of these third parties kind of scrutinizing whether your thing actually works. And so you're losing money and you're losing credibility all at the same time for a period of many years before you can actually get to the promised land. And so certainly, as a private company, we just went through a lot of years where it was very hard to raise any kind of money, get anyone to believe that these models work in part because maybe they weren't working that well in the very early years, and it took a long time before you could overcome all of that to get to the place that we are today. And one of the questions I get asked the most is like why I'm still doing this after 14 years, and so much of the answer has to do with the fact that, after those first few years, we kind of realized that on the one hand, the size of this problem and the opportunity were just enormous like, this is the sort of oldest industry in the world, like everyone needs credit, and it's probably among the very most important things you could work on to impact the greatest number of people in the greatest way. And also that this was going to take a really long time, like this isn't the sort of business that you're just going to build in 2 years and have it reached the potential it can reach. It was going to be something that was going to take multiple legs of the journey. And so one of the great advantages that Dave and I had as a team is our -- is that we could run this race in multiple legs. And so that's a little bit what's happened here is that we did this 14 years together. And then we think there's just so much more opportunity to do a second leg of this.

So yes, no, like it's an interesting perspective because as you said, it's like -- I think what I take from your comments is that you're probably blessed or lucky with your night of a [indiscernible] entered the market.

Some of that, yes.

But you also have adopted more of a -- like it's definitely like not just a marathon, but an ultramarathon type distance. And it's kind of really if you -- like we were talking the other night about Capital One and the development, for example, of that business, that's been a lifelong pursuit for some of those people, right? And I think that's probably true here as well. So a really interesting perspective.

So let's talk about a couple of the things you mentioned, outside capital as well as credit performance. So let's start with private credit. That's been kind of concern to your point of investors generally maybe doesn't have anything directly to do with you, but there's concerns in the market around availability of capital and some of what's happening from a redemption perspective. And some of the recent reports indicate actually the redemption pressure in semiliquid private credit is accelerating a little bit, like we saw that in some of the initial reports here in the second quarter. Against that backdrop, though, Upstart has just signed more than $4 billion of committed capital, including its first 24-month forward flow commitment and recently completed an oversubscribed securitization. How should investors monitor whether private credit redemptions could begin to impact your credit availability and/or the pricing of the capital? Like help us get comfortable that the things that -- the headlines maybe that we're seeing in private credit are not impacting Upstart adversely.

Yes. Fundamentally, capital is going to tend to flow to the places with the best risk-adjusted returns. And if you look at our results we have for a number of years here, delivered really consistently very high spreads against treasuries. I mean I think you'd be very hard-pressed to find spreads that are almost always like 400 basis points, averaging 600, 650 basis points above benchmark. I mean, there are just not a lot of at-scale places that you can consistently get returns like that. And so I think our capital partners that have done business with us have been very pleased by the returns, and you see that in the 100% renewal rate that we have with our partners that in spite of, for them, it being a relatively more challenging environment to maybe raise capital overall that they want to do bigger and longer deals with Upstart. That 24-month deal that you alluded to, that was a big focus of ours. When we think about these renewals, what are the deal terms we're really focused on. One of our top priorities has been getting the duration of these deals to be longer, periods to be longer because we know that capital markets are finicky. And even if we're in a place where the business is very strong, credit is strong. You can be in a place where the market has liquidity challenges for 3 or 6 months at a time. And we want to make sure our business is never in a position where a 3- or 6-month seize up in the markets can have a very significant negative effect on us. And so getting committed capital deals to be a year long, 18 months long, 24 months long, that is a major solve to that problem. And I think our ability to do that in spite of the challenging backdrop for a lot of these institutions is a testament to how strong performance has been for them.

So from your perspective, like at least right now in this most recent deal, like how should we think about like the sacrifices, if any, that you had to make from a potential profitability standpoint, spread perspective, anything like that? Because I mean, I think we're all kind of accustomed to obviously, longer duration tends to mean giving up some economics. On the other hand, as you said, it seems like there's a gravitation towards the kind of performance that Upstart is being able to deliver.

Yes. In recent deals, I mean, we've been very happy with the progression of our deal terms. Now this is a relatively new and innovative structure in this market. So I don't see a lot of it. And so a couple of years ago, when we started doing deals of this flavor, it is true that, I mean, you give up something in exchange for that duration of commitment. And generally, what we're giving up is we're seeing, we are going to put some skin in the game. And so we have a component of these deals, which is a risk capital that comes from Upstart, primarily funded by the contribution profits we make from loans that are getting originated. But we are putting a lot of those profits at risk in these deals. And so that's kind of like the give -- our give to it. And when we set up the structure, we kind of knew that was the fundamental trade here, and it was a good trade from our perspective. This is capital that is going to earn a very good return, including ours, and it buys us this sort of commitment that resolved one of the largest risks with the business, which is that we have no control over what's going on in the exterior markets, and we want that to be less of a risk for us. So there, of course, just like with every business, it's our objective that we can prove that everything works as expected. The first deal is generally not going to have your best terms, and we're going to expect to improve from there. And so that's been true for us, and it's continued to be true in this environment even as it's gotten more challenging, just because of how strong our results have been.

So let's talk about at least high level some of those results. Is there anything that you would call out in your own repayment data that would either validate or I guess, contradict the broader market concern that consumer or private credit performance is deteriorating. Are you seeing anything in your results or feedback in payment terms?

Yes. So we publish an index called the Upstart Macro Index, that's something that gets published, really updated every week, published every month. And it's pretty close to a real-time view of what we think is going on in the sort of consume -- overall macro consumer health. We think it's probably one of the fastest moving indicators that anybody has on this. And so that's pretty close to an encapsulation of our view on what's going on with the consumer. Broadly we've seen that over the last couple of years, we've been in this normalization post the sort of end of COVID stimulus and that big period of inflation. Notably, in the most recent handful of weeks and months here with the energy shock that we've experienced this year, we have seen some sort of pressure put on the consumer from that. UMI has gone up a little bit as a result. And we think that's probably a real effect. The only sort of silver lining to that is to say that over the last few years, if you look at the period of time when we had a 1.6, 1.7 type UMI, meaning consumers are 60% or 70% more likely to default than in pre-COVID normal times. Those times came in a world where inflation was near 10%. It was, I mean, a pretty dramatically different world than the one we live in today. And I think even though you've got headlines like today where you're saying, hey, inflation is at 4%, 4.5%, that's double the Fed's target. Yes, that's not ideal, but it's still -- there's a large space between 4% and 10%. And I think it would be very difficult for an energy shock alone to move us to a 10% like number just because energy is only a modest fraction of the overall sort of cost portfolio. And so at present, we don't see anything that suggests we're going to get back to a world that looks like that. And from our perspective, the most important thing is just like if macro is kind of like stable-ish, we're really happy because our business grows primarily from generating kind of compounding secular technology advances, things that improve conversion rates, make better underwriting, get the higher levels of automation, get better targeting, those things are kind of durable wins that are true in any macro climate. And generally, we're doing those at such a pace that some modest amount of macro fluctuation isn't really going to be a big problem for our business. But yes, like there ultimately is, of course, always some level of consumer deterioration, which would be a big problem for us and really probably for most businesses.

I want to come back to like as UMI moves and how that factors into your underwriting models, et cetera, because I think you guys are quite dogmatic about sticking with the models, et cetera. But before we go there, I want to return a little bit to funding and capital intensity, et cetera. And so just a moment ago, we touched on new capital commitments, longer duration, including a 24-month commitment. In the past, you've talked about 100% renewal rate since you did the first forward flow agreement in 2022. What level and duration of committed capital would make you comfortable that funding can support the 3-year growth plan even through a weaker credit cycle? Like how much cushion do you need?

Yes. We have very ambitious growth plans. We have guidance out there to grow the business at the top line at 35% a year for 3 years. And we think the TAMs across our various products are large enough to support a high growth rate for a long, long time to come, even beyond that. So we certainly expect that over time, the scale of this business could be one that is going to make it one of, if not the most important sort of source of yield for a lot of our partners. And so that means we're thinking pretty hard about like what the right structure, what the right blend is. And one of the things that we think is like this committed capital is really important for us because we want to make sure that any kind of market liquidity sees up, we can get through. But we also don't want to be so committed in a bidirectional way that there's no ability to flex up or down. And so we don't think 100% is the right number for this. We do want -- so it's just a question of, in the worst kind of market sees up where there maybe actually is something going on in the world, how much like downward and upward flex do you want? And so we want almost certainly a majority of our capital to be like long-dated and committed, but it's definitely not 100%. And so somewhere in there is the right number for us. And so we're working to make that the real mix for us. We are going to have to onboard a plenty of capital. And the good news is like we're doing that. I think in normal times, this actually really isn't a constraint on our growth because, again, to this point that capital is going to tend to flow to the places with the very best risk-adjusted returns. I mean, I think we've got it. And so as long as the markets are functioning properly, like that capital is going to flow in as we're able to underwrite and originate and acquire borrowers that meet the bar, and that is our limiting factor most of the time. And so it's a good way to run the business.

Got it. So with that being said, is, you are pursuing bank charter, right? And so you've emphasized that this is a regulatory and operating efficiency strategy rather than a balance sheet funding strategy. What are the most important milestones investors should track between application approval launch and then getting the benefits? And I guess that focus on it not being a balance sheet strategy, explain why that is? Because a lot of times, you see fintechs kind of a skew that strategy just because they're reticent to engage the incremental regulatory oversight, et cetera.

Yes. Maybe on the first, I would say, I don't actually think there are like that many milestones around this that are worth tracking. There is a regulatory process. We're in it, and I don't necessarily expect tons of kind of incremental updates along the way. I do think some of that time line is ultimately in the regulators' hands. I think the regulators have been really constructive with us and with other fintech companies that have been seeking bank charters. And so we're hopeful and optimistic about it. But ultimately, that time line is going to happen when it happens. That -- the benefits of the bank, we view as significant, but also just sort of part of the portfolio of bets that we've got over the next couple of years is sort of factored in how we think about guidance and where the business can get to already. And it's just -- these are just some of the bets in the mix and all of the bets will either play out or not, and they'll play out on a certain time line, and you need -- we need some of them to land in order to hit the guidance and don't need all of them to land. And I don't think if the bank is like qualitatively different than the other bets that we have going. So maybe that's sort of a high-level comment about how we think about that.

And then to the point about the purpose of the bank and why isn't it balance sheet strategy, the most fundamental reason is just that we have ambitions for the business to be extremely large. You just look at the sort of world of consumer credit, how much opportunity there is. You look across our different products in unsecured and auto and home and sort of the size of the business that we see over time is such that it would, it would be almost impossible in the short term for that business to be funded in a balance sheet centric way where, yes, a bank brings great super efficient cost of funding on the sort of financing piece. You still need to have equity. And one thing we care a lot about as a company is just being really, really efficient with equity. We want to maximize sort of returns on equity. We want to maximize returns for shareholders, we want to maximize the sort of any kind of measures of per share earnings or adjusted earnings or anything like that over time. And so that just means we want to be really, really careful about how we use equity capital. And at the same time, we look at the scale of the opportunity and our ambitions. And there's just really no way to square those 2 things unless you say, we're going to fund this with -- primarily with third-party capital. The only alternative is to grow slower, and we don't want to choose the grow lower path. So I think -- and I think actually, it does come down to the fact that you just don't see that many credit adjacent businesses that can sustain this level of growth at this scale for a long time, and we think that that's us, that can be done, and that comes along with this implication for the right funding strategy.

So talking just quickly on funding. I know that it's something that investors pay a lot of attention to, and that is the fund -- the amount of loans that are held on your own balance sheet, at least as of right now. And that was just over $1 billion at last quarter end. You've said that you expect some reduction over the remainder of the year. But what's the right steady-state balance sheet size relative to your originations, especially when you look at newer products like auto and HELOC and Cash Line, et cetera.

Yes. So yes, we have said that we do expect some reduction in the balance sheet. I would describe that as like tactical more than strategic. Like I don't really think that there is -- it's not that we think a number modestly lower than the current number is like more theoretically ideal than the number we're at day. I don't really think there's fundamentally anything wrong with the number that we're at today. I do think that it is to the point earlier about being really efficient about equity, like that to me is like really the limit factor. It's just like as long as we're operating within that constraint, that number is hopefully, and we expect going to grow over time, the amount of equity that is in the business just as a result of generating profits and retained earnings and like the business will have more equity to use. And so one of the things that it can be used on is R&D in the balance sheet. And that's, I think, a good sensible use of funds in certain cases. And so we'll kind of like opportunistically use it for that purpose. When we roll out new products, obviously, there's a lot of return and value in proving things out. Occasionally, -- and kind of like occasionally and doing some aggregations and stuff for sales when we think that those are good channels for us to fund loans, we'll use it. And then very occasionally, you might have some like money that is essentially just being parked and earnings some return, and that's not terrible either. But I view that as all kind of just tactical. The only really strategically interesting point for us is like we're going to be -- make sure to be super efficient about the total amount of equity that the business needs. And within that envelope, it will kind of go up and down, and I think as long as we manage that, it's not a major consideration in the business.

Got it. So let's talk about like actual underwriting in the models. First and foremost, talk a little bit about how you've construct the models and why be so -- like I used the word earlier, dogmatic about having models adjust to UMI and that kind of thing. Like how frequently should we expect model updates, like why take that approach? And what do you think comes next?

Yes. I mean, I think for us, credit is just nonnegotiable. And you never get credit perfect because the world is always changing a little bit. I think the best you can do is respond to the world as fast as you possibly can. And I think if you create processes that stand in the way of that, where you're saying, hey, I need to manage to this quarterly earnings number, so I better not -- I better close my eyes on like what's going on in credit. That I think, that's how you get into trouble. And we're here to build a business that's going to be here for a very, very long time through hopefully many, many, many credit cycles. And so I think if the best thing that we can do on credit is be the very best and first to respond and respond as precisely as possible. We're just going to do that. And if it comes at the cost of a little bit more volatility in the short term like -- so be it. I think that right investors and partners in this business are going to be people who like the fact that we take credit so seriously and respond to it as precisely as we possibly can and recognize that the real value from the business comes over multiple years as you compound these technology advantages, which are orthogonal to whether you're getting headwinds or tailwinds in the macro.

Got it. Just a quick question on UMI. Is there a point at which like resetting that makes sense just because you're -- we're kind of using pre-COVID as a reference time. And you're obviously a lot larger. The landscape is a fair amount different. Like would that ever make sense? Or how do you think about that? Not that it would make that much difference on the underwriting, but...

I think maybe. I do think that we believe that this whole period of time since COVID started has had unusual features. And so we don't want to get over-indexed to like any of those particular windows of time. But I mean, ultimately, it's just kind of quite, that's kind of just math moving the needle [indiscernible].

Right, right. That's right. Yes. Okay. So let's talk about the model, long runway for games there. And you've talked about lending model improvements as having, as I said, a very long runway for improvement. With traditional models, you've talked about having a 95% error rate and Upstart still leaving around roughly 86%, just contextualize for us what that error rate represents, and how you've driven variance versus the market? And what is the cadence for model improvement?

Yes. So we have a metric that we've developed internally that essentially, if you boil it down, it is looking at the difference between the net present value of the cash flows that you expected versus the net present value of cash flows that you got, and that is really ultimately what anyone should care about if you're in the business of credit is you're expecting a certain set of cash flow, you've got a different one. And how different that is when discounted to present, that's what matters. And so on that error metric, when we talk about 100% of a totally random model, it just means like if you just kind of randomly guess what cash flows you get from any given loan and compare them to the actual ones, 95% is kind of where traditional models sit. And then Upstart is at that 86% number. So call that almost 3x as smart or as accurate as traditional models. And that's measured as on sort of the 1 minus the amount of error that remains. And so that tells you 2 things. One is that Upstart has built a fairly significant advantage over the past decade plus, we've been doing this. And second, that there's still a lot of room to go. And I think one of the pretty remarkable things about the business is that contrary to, I think, maybe what many people would have thought. We didn't just like get this 0 to 1 moment where we found 1 nice variable or 1 sort of clever insight and use that to arbitrage the market. It was like, we have just continuously found ways to improve the model at a pretty consistent, almost linear pace over the years. as time has gone on, and we've continued to bring that number down, and we think that can go on for a long time just because you're only at 86% out of 100%.

Got it. So let's talk about the core business on personal loans. You called that your super power, if you will, and first priority as CEO from a product perspective. What are the specific indicators that investors should watch to see whether the ambition for reacceleration is coming from model gains, marketing efficiency, borrower demand or funding availability, like how do you rank order those, those drivers?

Yes. So for us, we do generally think of technology improvements as the primary driver of our business. That means you get better models. Those models can allow you to either approve more people, offer them better prices, give them a more automated instant experience or target the better. And that tends to be the single most important factor in the business. And it shows up in the form of efficiency. So these are improvements that are maybe different from improvements where you're just achieving them by virtue of spending more money on marketing exclusively, therefore, driving up your CAS. And I think what investors should look at to see if we're succeeding in this is just to see how much of our growth going forward starts showing up progressively as you move progressively down the income statement. And I think with less efficient types of growth, growth that it's like less technology-driven, you're just going to -- it's not going to show up in the way that you like. You see more of it on the top line, less of it as you move down. And I think throughout the course of this year, investors should look and hope to see that we're going to show not just what we've recently shown, which is that we have the ability to continually grow the top line really nicely, but that starts to sort of walk down the income statement as you go throughout the year.

So I want to talk about that. And I like how you characterize it and how to evaluate is like let's look at the P&L. And specifically, contribution margin was roughly 50% in Q1. And you've talked about that being the low point for the year, assuming no macro change. What level of contribution margin should investors think as being sustainable for particularly as auto, home, prime personal loans, Cash Line become a larger mix of the originations. Help us think through like the contribution margin differences in different products? And then what are going to be the drivers of improvement, at least assuming no macro change?

Yes. So a big part of the story of what's happened to our margins in recent periods does have to do with the mix across these products and segments. The go-forward story on these is pretty different, so I want to take a moment to break them out. Within personal loans, there's a really big difference between the margins we get on our core business and the margins we get on the super prime segment. Super prime, not surprisingly, is a very competitive market. It doesn't tend to come with very high margins. And frankly, it was -- if you look at the numbers, the place that, as a company, we put a lot of focus and got a lot of growth in over the last year. And going forward, you can expect that our focus is going to flip the other way, and we will care much more, invest much more, be focused much more on growing the core segment than super prime segment. We have, I think, achieved some great things with the super prime segment. 2 years ago, if you went to Upstart as a super prime person, you would have found a terribly noncompetitive rate. I think that translated into certain inability to use generalized marketing channels. It came with a certain brand reputation that didn't lend itself to being the very best. And I think today, we have established a product that any American can go to Upstart and find some of the very best rates that you can anywhere for anybody. And that's really powerful. It unlocks much more generalized forms of marketing, which are useful even in our core business. But from how much market share, how much growth do we need to do their perspective, I would call it mission accomplished, and that is no longer sort of going forward, our top priority. Now -- then there are these new products, in particular, auto purchase and HELOC and the sort of emerging Cash Line product. These products, we're really optimistic about. We think these products are going to be a really important part of the future. These products are still relatively new at different parts in the life cycle, but their contribution margins still have a lot of room to improve. And I want to maybe anchor everybody to the sort of point that our personal loan margins were improving for almost a full decade since we started the company, they were -- it's not like personal loans is a business that got to mature margins after 2 years and then just scale. It was scaling while it was improving it's margins. And that's because the amount of margin you can take is so deeply tied to the amount of tech differentiation that exists because that tech differentiation is what unlocks pricing power, right? It's like when our rate is so much better than anybody else's rate. That's why we can afford to take some of the highest take rates that exist in the industry on that product because the next best alternative is so much less good than what we offer now in personal loans. And so I expect that same dynamic to play out in these products. I think in the very near term, their margins will improve very rapidly because they're quite negative today, frankly, on these new products. And so they'll have a period of rapid improvement. And then they'll just continue to improve. And I think they'll probably should improve for years and years to come here. So I don't think we're going to reach like a mature margin anytime soon. I do think we're going to get a much higher margin soon and that's going to help out sort of the near-term financial business relatively quickly.

So help me in just the last minute balance this out. So I get like as you kind of go to the core part of your market that, that has better margins, and so that will be an uplift on a contribution margin versus where you have been. It sounds like you're going to see improvement on new products as well. But yet even as they grow as a percentage of revenue, they're still probably dilutive to overall contribution margin. So it's a little bit of a combination there where if you were just going -- changing the mix on personal loans, you probably could improve contribution margin faster, but then you're growing, and you're going to get improving margins. And so the amount of dilution will come down, but it's -- they're still going to be dilutive for a while. Is that how we should think about it?

Yes, that's true. And the sort of exact net effect does depend a little bit on how fast they're growing. But I mean there is definitely -- I mean, you do get you do get a benefit from moving from deeply negative upward exactly like that, that helps you a lot. So I do think there is a -- we have a lot of reasons that we feel good about where the lower half of the income statement is going this year, and that -- these are some of the reasons.

Got it. Well, we're out of time. Thank you so much, Paul. Really appreciate you being here and telling the Upstart story. It's been really fascinating. Thank you.

Great. Thanks.

All right. Hello, everyone. My name is [ Rod ] Dewan. I'm the Co-Head of our Fintech Investment Banking practice. I'm pleased to be joined by Paul Gu, who's the Co-Founder and CEO of Upstart. Looking forward to the discussion today.

To ensure that I don't get my wrist slapped, I'm going to read a disclaimer before we get into it. Today's discussion may contain forward-looking statements that relate to future results and events, which are based on Upstart's information available as of today and are subject to risks and uncertainties. Actual results may differ materially from these forward-looking statements. Our discussion may also include non-GAAP financial measures, which are not a substitute for GAAP results.

Please refer to the company's filings with the SEC and its IR website for additional information, including GAAP to non-GAAP reconciliations, along with other disclosures.

All right. Maybe with that, AI is the leading AI lending marketplace, connecting consumers to hundreds of partner banks and credit unions. Enabling an automated and streamlined borrowing experience, Upstart has expanded access to credit for millions of consumers and improved risk decisioning, leveraging its advanced AI models. Paul, you co-founded Upstart, what, 14 years ago now?

2012.

And you previously served as the company's CTO, where you helped build and scale the core technology and AI-driven credit platform. And congrats on your recent appointment to CEO, and appreciate you being here today.

So maybe with that, just take us to the beginning. You cofounded this company 14 years ago. You dropped out of VL as a Thiel fellow. What was the fundamental problem that you were trying to solve? And just think back over a decade, did you set out to build an AI lending marketplace? Or what was the vision back then? And how has that evolved over time?

Yes. So way back when I was 20 years old, as you said, I dropped out of college, I did the Thiel fellowship. And before -- this was before Dave and I had met Dave my co-founder, who was CEO until very recently.

And from my side of it, I had spent a summer after dropping out of college at D.E. Shaw and generally was very interested in playing around in the world of quant finance. And my observation there was that you hired very smart people using really sort of cutting-edge computing techniques to go after a problem that I thought was a very narrowly defined problem that was like how do you arbitrage securities prices to be a little bit more efficient, but to go after that problem with extreme amounts of accuracy, extreme amounts of sort of intelligence.

And it occurred to me that if you could just take these same techniques and apply them to a problem that would impact a lot more people, you could solve something that would be fundamentally very important for the world. And so naturally, you take just one step from the world of kind of maybe securities finance to the world of consumer finance and you say, okay, are there problems that you could solve here by using these techniques?

And I thought that if we could use very similar techniques as are used in sort of the world of quantitative finance, but apply it to this problem of how you understand consumer risk, you could get a lot more people access to credit that maybe they didn't have or didn't have good enough of.

And so that's what I wanted to do. I met my co-founder, Dave, while thinking through this idea and working on the very first versions of it. We got together because we both thought that the right packaging of the solution would be something called an income share agreement, which is basically an idea that you give someone money in exchange for a small fraction of their future income. So kind of like a loan, but not really.

But what it had in common with the thing that the business later did was that, one, we were solving a credit access problem; and two, we would solve it by building models that could understand the financial trajectory of someone who didn't necessarily have a lot of conventional credit history. And pretty quickly, what we realized once we got into this business was that this was exactly the right problem to be solving. Tons of people had it. It was a huge market opportunity and the incumbents just weren't going after it.

And that also we could solve this problem with this particular technology approach of building machine learning algorithm-based models on alternative data for consumer underwriting. And so we started to do that, and we discovered that in the sort of conventional loan format where people understood what the APR was, what the repayment terms were, you didn't have to explain to someone like what an income share agreement was or why it wasn't indentured servitude. It just worked much better.

And so we did more and more in that business. We quickly discovered that it wasn't just people who were new to credit or didn't have a lot of credit history who had this problem. It's really people across the entire economic spectrum that are underserved in terms of credit.

Either they don't get approved enough, they don't get the right APR for them or they get served an offer of credit that comes with a huge number of procedural steps and verification required. And all of these people could be much better served by an AI-first way to do credit. And so that's what we started to invest all of our energy and attention into, and that's how we built the business that we have today.

I like how you frame the problem. Just again, think back 14 years ago. How is the problem similar versus different today than it was a decade ago?

Surprisingly, very similar. So one of our favorite metrics is a metric about the amount of inaccuracy in lending decisions. And so if you said 100% is a metric where you have just a totally random model, you have no intelligence, you're just making lending decisions at random, 0% means that you have gotten rid of all of the inaccuracy and now you have a totally perfect model that can get everything right. A sort of textbook traditional model lands you at about 95%. That's how much of the error a traditional textbook model leaves on the table.

Our model after 14 years of optimizing and improving at a really -- what we think is really a decent pace, has us down at 86%, which is to say that 86% of all the error in lending remains to be solved. And that means that if you were just to start from where we are today and look out, you would say that there is an enormous amount of opportunity for models and lending to get more accurate to understand the consumer risk better. There's more opportunity to get more data into the models, more opportunity to build more sophisticated algorithms that understand the signals from that data better.

And frankly, the difference between 86 and 95 like zoomed out doesn't look all that big. So I think if I were starting a company today, it probably wouldn't be much different than the one that we started 14 years ago.

Now let's maybe shift gears a little bit on your role. So you're now the CEO, you were formerly the CTO. If I just think about that dynamic, Satya and Sundar both had kind of technical backgrounds and experience, then became CEO. What's going to stay the same? What's going to change from your vantage point?

We always had a strong conviction that the answer to making consumer lending fundamentally better and different was going to come from technology. I think that's probably always been our most contrarian take on this market is that if you look at most players in this business, they don't have a technology-oriented thesis. They have either a capital markets or funding-oriented thesis, like they have some sort of better way to get a lower cost of funding or they have a sort of marketing-oriented thesis.

They have a better way to sort of get the consumer into their ecosystem to have customer loyalty or to get customer acquisition. And our thesis from the very beginning was that those things are nice, but the most transformative thing that you can do in lending is if you can actually change your understanding of the risk, so you can get dramatically better separation of risk. That's what allows you to approve people that otherwise couldn't get approved or approve them at much lower APRs and therefore, have the pricing power to generate unusually high margins and build a really fantastic business.

And so I think that comes from technology. You have to be a really good technology company. And when I say technology, I don't mean in the sense of just building a website or writing a bunch of code that has features, really like technology in the sense of, I think, being a research-first company that fundamentally is about building models and building more accurate models is sort of much deeper type of technology than I think your -- maybe your typical software company.

And I think the DNA for doing that as a company is fundamentally a very technical thing. So I think in that sense, we've always had that conviction. I think that's even more true for me coming from the seat that I sat in before. And I'm going to be very focused on making sure that we extend that lead, grow that lead and keep investing in it going forward.

That's great. So we've talked about the past. Let's flash forward 10 years from now. AI permeates the consumer finance space. How does the experience change for the average consumer? That's part one.

And then part two, just as you think about AI for your business specifically, are there certain things that you're seeing an acceleration in with respect to operational performance, costs, et cetera? Or is it more driven kind of top line basis?

Yes. Today, I think we live in a world where if you want credit, a bunch of things have to happen. One is that you have to know that you want or need credit. You have to know which type of credit is the right one for you. That could be a HELOC, it could be an auto loan, it could be a refinance loan. It could be an unsecured loan, it could be another kind of revolving loan. You need a lot of knowledge and sophistication as a consumer to know which one is the right one for you.

And then you have to have the willingness to actually do the work required to go and get that loan. And when you put all those steps together, it's no surprise that the vast majority of people just don't do it. They don't get the right form of credit at the very lowest price at the right time for them. Instead, what you see is you see a lot of people sit on really large amounts of credit card debt for a large number of years accruing at a very high interest rate, not because that is the very best price for them in the market or the very best thing for their financial lives, just because of all of those steps that you have to go through.

And I think instead, we can get to this world where credit can be what we call always on, just there for you, where you can have sort of AI that helps people figure out what the right form of credit for them is, and then you can make it effortless and instant as the vast, vast majority of our loans are today to actually get that credit so that there's no friction between a consumer and doing the thing that's best for their long-term financial health.

And as you think about that always on credit, which I think is the right kind of frame, what are the different ways to potentially monetize that over time?

Well, the beautiful thing about credit is that if your credit is actually differentiated, it's just a fantastic business, and we see this with our core personal loan business. When we serve a personal loan to a consumer that's not rated conventionally prime in the market, this is a consumer that gets a huge amount of value from our loan.

And as a result, we have a lot of pricing power in that product. We're able to earn very good margins in that product, and that basically happens because it's such a differentiated product and has a ton of value creation for the customer. And so I think in a similar way, when you think about something like the always-on credit, it's -- first of all, it's credit.

And if that credit has the same great underwriting attributes, the same level of technology differentiation as all of our credit products today do, it's going to have the same margin profile, same margin tailwinds. But then it gets an extra benefit, which is that today, especially in our business, we continually have to acquire customers for each loan as if it's a new transaction, a new customer, and that comes, of course, with customer acquisition cost.

I think in a world where you have people that are living within the ecosystem and the credit is coming automatically as just a part of being in the ecosystem, then obviously, you're going to be able to amortize that acquisition cost over a much longer customer lifetime.

And as you think about the products, and you talked about the customer acquisition cost, top of funnel, do you lean towards one specific type of product or type of customer because you've seen the customer acquisition cost for that being lower, all else equal, and as they move down the funnel and become multiproduct customers? The overall opportunity is greater?

Yes. I'll say 2 things. First is we're very committed to having the best product for Americans of every category, people up and down the economic spectrum. Anyone should be able to come to Upstart. And for any use case of credit, obviously, we're not quite there.

We don't have all the products yet. But for any use case of credit, eventually, you should be able to go to Upstart and just confidently know you're going to get the very best credit product. Now having said that, the second thing is also true, which is that some customers are much less well served by existing credit markets than others.

And those consumers that are not conventionally rated super prime tend to have many fewer good options in the market. A conventional issuer will rate their risk of defaulting much higher. And so the space to reduce their prices, improve the speed of their process, improve the sort of size of credit lines that they have access to is much greater.

And therefore, our ability to built a great business there is much greater. And so we will increasingly going forward, be focused on how we can grow the most in the segments where we have the most differentiation while maintaining the baseline fact that anyone can come to Upstart and get a very good product.

You talked about a few of your different products. Let's double-click on just the product road map and that journey. So you started out as a single product company way back when, right, unsecured consumer lending.

It's now a multiproduct platform that's had a great amount of success spanning auto, HELOC and now revolving credit. Now as you reflect on the journey, is that kind of how you imagine things to happen sequentially? And if not, maybe just share a little bit around the dynamics of how you decided to prioritize expansion into one product over another.

Yes. I would really say, actually, this is a pretty new thing for us. Until very recently, we effectively were a one product company. We had a personal loans business and just a bunch of exploratory bets.

I think really as of the last quarter, I would say that we are really have grown into our own right as being a multiproduct company. We have shown that we can deliver real growth in distribution in our auto purchase product and our HELOC product.

And those products are, I think, now well past the exploration phase. They've sort of made it on their own, and I think are going to be very good businesses for us. I think in terms of the order, probably if I could do it again, I would do the order a little differently.

We just announced a new product called Cash line. It's actually a natural adjacency to our core personal loan customer segment and core personal loan product. And if you look at the success that a lot of players have had in that market, it just was a really natural product for us to be in.

Probably the lift to get there was lower than the lift to do something like an auto or a home. I think those are really, really good markets to be in. They have enormous TAM. We want to be growing in them for a really long time to come. But the lift to get from 0 to 1 on those products, frankly, took longer than we would have thought.

I think especially in auto, we've been humbled a little bit. I think if you asked us 3 or 4 years ago, how long it would take for us to get set up in auto, we would have just been too optimistic about the time lines. And so we're really, really happy to have the auto business working now, but it was hard work to get there.

We had to really understand the dealer as a customer, and that's something that wasn't natural to us coming from kind of a direct-to-consumer background as a business. We just weren't naturally attuned to thinking in terms of the end borrower.

And we've had to learn kind of all the idiosyncrasies of how car dealerships work, what they need to be successful, how to think about them as really our customer. And that's a capability that we've developed over the last couple of years that's enabled us to be very successful in auto, but it took work.

Maybe just while we're on that topic, 1 or 2 lessons learned that you can apply to the rest of your product road map? And then maybe just spend a little bit of time on like how do you think about conceptually, like where do you want to go next?

Yes. Just working backwards, ultimately, we want to have the very best credit product for every consumer credit need. And I think at this point, we're actually not that far away from that. We're probably just a few major products short. So there's probably not that many degrees of freedom in how we go from here to there. We're going to build all of those products.

We're going to make it so that if you are an American consumer and you need credit, Upstart is the place to go. And whatever stands between here and there, we're going to build those things. And so I think there probably were more degrees of freedom in the past when we needed to build everything than now when we've built many of the things.

So I think we are well on our way. Having said that, I think we just now have a much better understanding of the different pieces of standing up a new product in terms of managing the R&D on the balance sheet, managing the sort of credit models and how you want to ramp those, thinking about when you're first scaling distribution channels, having realistic expectations about getting the -- about when those channels are going to scale at what kinds of -- and there's a lot of learnings in there.

I think we kind of understand the life cycle of a new product now, and we'll finish the consumer credit suite soon.

And while we're on the topic, just build versus buy a framework, how do you think about that given the journey that you've been on?

Yes. I always think that the things that you want to be, best-in-class at that you want to be really differentiated at that you really are going to be your source of alpha in the market, those are things you have to build. You need to own them.

And if you're going to make them better than what else is available in the market, then, of course, you can't just buy elsewhere in the market because by definition, that won't be differentiated, won't be the very best.

And then I kind of think for everything else that you don't care to be the very best at, there's no reason to spend your own effort building it. So when people talk about, oh, everybody is going to build their own sales force or something and there's going to be no more vertical SaaS, mostly, I think maybe some company will decide to do that.

But for a business like ours, where the growth rates are very high and the TAM is very large, and you can sort of do that growth rate and compound for a long, long time. It's just very hard to justify spending your resources anywhere other than investing back in the things that are going to make you special and different.

But for us, all of lending is that. We want to be differentiated in auto, home, unsecured, every major category of consumer credit, we want to have the very best product in. So all of those are going to be built for us.

Okay. Maybe just shifting gears to financial profile. So there's a lot of debate in the space around growth versus profitability. How do you think about that today? How has that shifted over time? And just in your new seat, just any dynamics relative to just the way that the business has operated historically?

Yes. We've always -- taking sort of a 10,000-foot view first, we've always felt that consumer credit in general, is insanely competitive market. There's a countless number of players that want to be in loans and have been since forever. So there's nothing inherently interesting or special about just being another player in credit.

What's interesting is where you can find places in the market that you can be really different and have a lot of pricing power. And we've built that and established that in what we call the core personal loan segment, core personal loan business for us. That is a fantastic business.

If you were to look at that on a stand-alone basis, it's a very high-margin business, and it's done really well for us over the years. It's also a business that we think can continue to grow at a very high rate because its fundamental driver of growth is just better technology. We make the models better.

We can approve more people at lower prices with more automation. That leads to higher conversion rates, that leads to more of the market being addressable. The growth engine is one that we're really good at. We've really well understood for a long time.

And we think that product can just keep growing. Now it has not been our focus to maximize growth in that product because we've been focused on so many sort of adjacent priorities around rounding out the product suite. But this year, in particular, we expect to see growth return to this core personal loan segment as we've put the focus of our product, technology and marketing teams back on to it.

And it's actually the natural thing for us to do. It's probably easier for us to do that than all of these sort of new and different things that came less naturally to us. And this product naturally, as it grows, it's very nice because it both comes with growth and it comes with profitability.

And so it moves things on both fronts. And so as that happens, I think naturally, you're going to see that sort of impact the income statement up and down. I think over the course of the year, gradually, we will see a sort of re-expansion of some of our contribution margins from local minimums that were just the result of sort of some of these shifting mixes.

And as that sort of shifts back towards growth in core personal loans, we will gradually see it go back the other way.

Let's go a little deeper on the growth side of things. So if you think about just new customers, expansion within your customer base and then things that are kind of pricing related, how do you think about that dynamic between those as you think about just driving your future growth?

Well, our growth always -- we always think about growth as -- in a mature product for us, the most important source of growth is going to be higher conversion through better technology. And that will always be the sort of #1 thing we come back to. We think there's a lot of that left to do. We can just keep delivering 1%, 2%, 3% type wins, and we have a bunch of teams that's mandated to do exactly that. So we will keep on doing that.

Then there's our new products. New products have a slightly different dynamic where sometimes they have product-specific marketing channels that need to be experimented with and activated and understood. So if you think about something like the home loans market in HELOC, for example, there's a lot of home adjacent businesses that you can have partnerships with.

Those are like very specific channels that you have to come to learn how to work with and understand. Obviously, our auto purchase business, as we talked about earlier, that business has very specific dynamics that's distributed through car dealerships. So you have to understand the car dealership as a customer.

And so that's a distinct channel that you have to come to learn. So new products have some of their own dynamics in terms of mastering channels that has sort of a 0 to 1 dynamic and 0 to 1 motion associated with it. But once mature, then it really becomes a technology compounding game, and that's the game that we play in personal loans and one that we've really built the whole business around being good at doing.

That's great. I have one more, and then I'm going to open it up to the audience. I have a few more topics that I'd love to cover. You reported Q1 recently, strong results. How do you bridge kind of Q2 through Q4 of the business relative to how Q1 performed? And is there anything that you think that the Street is missing?

Yes. So our Q1 results, I think the real sort of fundamentals of them, the actual growth in the underlying business, the sort of proof points on new products, we think, in a long-run sense of the business are really, really good facts. It's like we have a core business that is doing really well, can continue to grow for a long time, it's very profitable.

And then we have these new segments in home and auto that have enormous relative to personal loans, almost like unlimited sized TAMs that you can just grow in and compound in for many, many years to come. And that's a really positive set of things for the business. Now I think the market was pretty surprised by the bottom line results in Q1. We didn't have very high EBITDA margin.

But we did reaffirm the full year guidance. We expect that to ramp back up over the course of the year so that we still get back to the full year guide on that. And that is going to be pretty heavily backloaded in the year.

And that's a function of a few different things. Some of those are kind of like not strategically interesting, just have to do with like the timing of when certain expenses are growing in the year. But basically, we expect that fixed expense growth is sort of like disproportionately growing at the very beginning of the year, basically will moderate in terms of its growth rate from here.

It's not like it's going to like a big step down or anything, so nothing terribly unnatural, but just like the relative growth rate was very high in Q1 and then just going to be very modest after that. And then there's going to be -- then there's basically back to like the actual business and what's growing in it, we have a real focus now on growing in the core personal loan segment, and this is a segment that has much higher margins.

And so as that increases its sort of percentage of all the growth that's happening throughout the year compared to the last few quarters, that naturally is going to have an effect that pulls margins up.

And that's because it's a mix type dynamic, it's not something that happens overnight. It is an effect that happens gradually. But you'll see it in the contribution margin line, you'll see it in the sort of EBITDA line.

And you'll see it sort of up and down, but it will happen gradually as that sort of mix of -- as a percentage of the growth comes in and therefore, affects the overall mix of products and segments that we're originating.

I have a few more topics I want to hit, but maybe I'll just open it up to see if anyone in the room has questions. So maybe if not, let's talk bank charter. So there's multiple companies, including yourself that have announced moves towards a bank charter. Just walk us through the evaluation of that decision and maybe what you see as the tangible benefits for that near term, medium term and longer term?

Yes. Our rationale for getting a bank charter are really different than I think many players in the market. We have long held that we want to be very capital efficient. We don't want the growth of the business to be tied to growth in the amount of equity capital that the business has. And one of the challenges with being -- running a bank business model is that those things can get mixed up.

We have no intention of doing that. So while we will hope to have a bank charter in the near future, we don't intend to use a bank business model to fund our loans. We will still be predominantly third-party funded and run in a really capital-efficient manner. That is super important to us.

However, what a bank charter will do for us is it will give us much clearer and faster regulatory ability to lend and lend where we want to across the country. Today, in our current business model of lending across a network of partner banks that do the actual origination from a legal perspective, there are a lot of states where you have a restricted ability to operate.

You can't quite reach all customers in all 50 states. You also have a lot of costs involved in using the structure that because you have other partners that are taking pieces of revenue along the way before it gets to us. And so there's just more efficiency there.

And then finally, I think in the most fundamental sense, today, we have a lot of costs and process associated with maintaining regulatory relationships via a large number of intermediaries. And I think as AI becomes a more central topic, it just is natural that as the AI lending company that we establish a first-hand direct relationship with the regulator to represent how AI is going to help the American consumer, and we're excited to do that.

That's great. While we're on the AI topic, just you've been an AI forward company for a period of time. How do you think about just the end state and working backwards from where you're at today? What do you want to see achieved over the coming quarters with respect to AI being infused further in your business?

Well, we've been doing AI and lending for a really long time. And the thing about what we do is that it's a very different kind of AI than, say, when you're looking at LLM models or foundation models that are now very popularly used in a bunch of applications. Broadly, I think the problem space divides into problems that AI is better at people and problems that AI sort of started out being worse than humans.

Like are humans good at this problem? Are they bad at this problem? And a lot of the problems that we deal with are like giant math problems. They're like problems of the form. You have a ton of data about a person. You can know thousands upon thousands of different facts about a person. And then you have to crunch that through a bunch of historical data to decide based on the patterns if you think this person is going to pay back a loan. And that is sort of like a giant math problem.

And it's actually a problem that humans have been terrible at forever. It was never a problem you want to hand even to a really smart person to decide how to underwrite a particular applicant for credit. And so it always required a very different kind of model than what you use, say, in Quad or ChatGPT or something like that.

It was always something that was going to be much more heavily numeric and fairly proprietary to the type of problem that is present in lending. And so we spent the better part of the last 12 years or so building models that are optimized for this use case. There's sort of a lot of Upstart-specific innovations that have made that successful. And then there's a classic problem that humans are actually pretty good at and are more like multimodal sort of general intelligence problems. And some of those problems, I think, are now newly solvable because of what's changed and advanced in AI.

So if you think about some of the problems associated with, say, verification of a home loan, you're looking at property records, they're coming from a fairly offline county system. They might involve like hand sketches of property borders and you have to like decipher that, interpret and decide whether that matches up with the property that you're trying to verify and place a lean on. Those are problems that multimodal sort of foundation models are very good at dealing with and are suddenly possible now.

And I think that there's something really powerful about the idea of bringing these 2 types of models together, so you have the sort of like crunches large amounts of data to do a giant math problem and get really high numeric predictive accuracy, but something that a human would just be terrible at is not sort of a general intelligence AI problem is instead a giant math problem.

You combine that with those types of problems like the sort of home loan verification problem that is much more of a general intelligence multimodal AI problem. And now you can really get much closer to this world that we talked about at the beginning, where you can have this kind of always on credit that gives people much better rates, much better access and does it effortlessly from their perspective.

And always on credit, how far along on that journey are you right now?

We're early. We're early. So we recently launched a product called Cash line. It's our first product that really is always on in the sense that Cash line is a product where someone can be a regular subscriber to the product and the credit will always be there for them no matter what happens as long as they're meeting their commitments to us, our product is going to be there for them.

And that is, in some sense, a very small first step. You can imagine that within the same kind of subscription membership ecosystem, there will be other types of credit that we can start sort of bringing into that and making similarly always on and available to the customer, having the sort of like always-on underwriting, always on sort of data access to what's changing in the consumer's financial life.

Those are the other pieces of it that are getting tied together. And then, of course, the last step will be bringing the really big kind of secured credit products into that ecosystem. And then I think you have something that is really different and powerful and suddenly looks very different than the way that people normally had -- traditionally had to get credit where they had to go looking for it and decide that now is the time to refinance particular debt or other.

We're coming up on time. Just one last question for me. What's one thing that's either been misunderstood or underappreciated about Upstart?

I think people always have always thought that Upstart was a business where you had one clever mousetrap, you found maybe one little arbitrage opportunity in a model and it was over. In the earliest days, that took the form of people asking us, what's the one secret sauce variable that explains your success? And we try to tell them there's not just one. There's a bunch of variables.

And then nowadays, I think it's reflected in the perspective that I hear a lot, which basically boils down to when you look at how people think about the growth of this business, they think, okay, well, you found this mousetrap, you're basically going to arbit out in the next year or 2. And then I guess from here, you'll just cash flow. And it doesn't look like you have that much cash flow. So it must not be that good of a mousetrap.

And the very strange thing about this is if you actually look at what's happening in the business, it's like we have a business that is growing at an incredibly fast rate into a market with an extremely large TAM, so large that we're just a rounding error in the size of it. And so by every -- like every measure, this is a business that I believe can compound at an extremely high rate for many years to come. And so it's just going to be much more sensible.

The math is very clear that the thing you should do in a business like this is you should make investments back into the business, make sure that you can maximize how much of this TAM you can address, how quickly you can go after it. And that's what we're setting ourselves up to do.

So we certainly hope to be compounding in this business for a very long time to come. And I think we're going to prove a lot of models were too shortsighted in how long they thought this could go.

This is great. Thanks for being here, Paul. And congrats again on the new role.

Great. Thank you.

Good afternoon, and welcome to the Upstart First Quarter 2026 Earnings Call. [Operator Instructions] As a reminder, this conference call is being recorded.

I would now like to turn the call over to Sonya Banerjee, Head of Investor Relations. Sonya, please go ahead.

Thank you. Welcome to the Upstart Earnings Call for the First Quarter of 2026. Joining me today are Paul Gu, our Co-Founder and CEO; and Andrea Blankmeyer, our CFO.

During today's call, we will make forward-looking statements, which include statements about our outlook and business strategy. These statements are based on our expectations and beliefs as of today, which are subject to a variety of risks, uncertainties and assumptions and should not be viewed as a guarantee of future performance. Actual results may differ materially as a result of various risk factors that have been described in our SEC filings. We assume no obligation to update any forward-looking statements as a result of new information or future events, except as required by law. Our discussion will include non-GAAP financial measures, which are not a substitute for our GAAP results. Reconciliations of our historical GAAP to non-GAAP results can be found in our earnings materials, which are available on our IR website.

With that, Paul, over to you.

Thank you, Sonya, and thank you, everyone, for joining us today. I want to start my first official earnings call as CEO by stating simply that the Upstart leadership team and I are here to build a high-growth and high-return business. I'm the founder at heart. I dropped out of college at 20 to start something. And now after 14 years, I wouldn't be doing this if I didn't believe the upside ahead for Upstart was as good as that of any startup.

In recent decades, there's been a growing trend for the fastest-growing companies to stay private. And as a result, public companies are typically past their high-growth years. We believe Upstart is not. As reflected in our 3-year outlook of 35% annualized revenue growth, we expect to be one of the fastest multiyear compounders at our scale. Consumer credit is arguably the oldest, most economically foundational business there is, and today is the perfect time to reimagine it. Unlike in some areas, the application of AI to credit is an unambiguous good for the consumer, saving them time and money to use on the parts of life that really matter.

For lenders, AI will transform credit from a structurally commodity-like business to one where the player who wins the technology and modeling race wins the market. With a decade-long head start, we believe that race is ours to loose. Capitalizing on this enormous market opportunity will require some investment. Fortunately, we have just the right business to fund it, core personal loans, unsecured installment loans to consumers not conventionally considered super prime. Our significant and growing lead in technology built up over that decade plus gives our product there the best rates and best process in the market, making room for unusually high margins while still delivering the best product to the customer. You're going to hear me talk a lot more about this as CEO. Our core personal loan business makes a lot of money and my first priority is to do a lot more of it.

Businesses in today's world, especially in lending, can too easily put up big numbers that depend on even bigger equity bases. At Upstart, we have always treated equity as a real cost, and I intend to double down on that rigor. Our operating strategy is to reinvest the profits from core personal loans into building the best product and most trusted brands across every category of consumer credit. This approach allows us to simultaneously maximize earnings over the long run while running an extremely capital-efficient business. Similarly, our funding strategy for loans will continue to be one that relies primarily on third-party capital. As they say, the market is a weighing machine in the long run, and my bet is that the businesses with the most profits and the least dilution will weigh the most.

Now I'd like to turn to Q1 and where we stand today. Originations grew 61% year-over-year and revenue grew 44%, while profit declined marginally. These are strong results and put us comfortably on track to meet our full year guidance on both the top and bottom lines. These numbers reflect a mix of 4 factors: secular improvements to technology and marketing, strong momentum in newer products in the super prime segment, the usual Q1 seasonal headwinds in borrower demand and annual employee-related expenses and some planned investments. I'll focus on our platform and product strategy, and Andrea will walk through the numbers.

As always, our most important growth lever is improving our underwriting model. In Q1, we increased the accuracy lead of our personal loans model over benchmark by 1.4 percentage points. Our models to manage now stands at 173.6%, while 87.4% of the total inaccuracy remains to be solved. This quarter, we extended the scope of our models to predict post-default recoveries, replacing the assumptions we've used historically with the full strength of our AI models. This fuller view of loan economics lets us serve more creditworthy borrowers, which drove approximately 3.5% more originations at equivalent risk levels relative to our prior model. Simply put, our lead over traditional credit scoring continues to grow.

We're also moving quickly to maximize use of AI across every part of the business. In servicing and collections, we doubled daily AI-assisted borrower conversation volume, brought that capability to our mobile app and expanded our AI-powered payment features. We also deployed AI-driven quality assurance tools to review customer service calls, giving us a scalable, consistent way to continuously improve the borrower experience. Across our platform, we originated more than 425,000 loans in Q1. We believe more Americans are choosing to borrow from us than any pure fintech platform. With well over 20 million unique consumers having created accounts to check their rate with Upstart, we are rapidly building towards being the most trusted brand in consumer credit.

In auto, originations grew more than 300% year-over-year and 30% sequentially. Auto retail was a standout with originations up roughly 13x year-over-year and nearly doubling sequentially, driven by a rapidly expanding active dealer network. Our work to reduce friction for dealers is paying off. About 1/4 of retail transactions in Q1 use the remote signature capability we launched late last year. We also rolled out a new feature that lets dealers generate firm AI-powered offers across multiple vehicles from a single customer application. And we deepened integrations with dealers' existing compliance and CRM tools, embedding Upstart more naturally into how they already work.

Home originations grew approximately 250% year-over-year and 16% sequentially, driven by better marketing reach and efficiency. In Q1, more than 1/4 of these loans were fully automated, and we achieved an average time to close of just 6 days from application to signing, a new record for us and a fraction of the industry average of roughly 40 days. In early April, we also added richer bank account data to our HELOC income verification process, improving accuracy and the salability of these loans to capital market partners. This progress in auto and home has set us well on our way to serving the full range of consumer credit needs. With growth strong and technology advancing rapidly, the time is now right for both products to begin shifting some of their focus from pure growth to unit economics.

Last month, we also launched cash line, our first unsecured revolving credit product. This is an important step toward our vision of always on credit for every borrower, and we're thrilled by the early results. Looking forward, the next area we're focusing our product and growth efforts on is none other than core personal loans. I said earlier that the profits from this business are central to our strategy, and we have already begun taking action to grow it. While we would normally expect originations to decline sequentially in Q1, core personal loans were flat to Q4. That stronger than seasonal performance signals the early stages of the reacceleration we expect to continue through the rest of the year.

Now I want to turn to the capital side of the business. Funding supply for loans is strong. Thanks to the pioneering work Sanjay and the capital team have done, well over half of our capital is committed. Year-to-date, we've expanded and deepened our forward flow relationships, securing over $4 billion in new committed capital. That includes about $2 billion in new commitments from [ Altura ], Centerbridge and [ Wafra ], alongside renewals from Fortress and Blue Owl. Notably, we've closed a 24-month commitment, which is our longest deal term yet, designed to provide durable capital through market cycles. I'm also proud to share that this continues our track record of a 100% renewal rate with every partner since our first deal in 2022.

Additionally, our recent securitizations totaling approximately $1 billion were multiple times oversubscribed with the most recent transaction upsized. This reflects strong secondary liquidity for our loans even in the broader market volatility. We also included auto secured personal loans in a securitization for the first time, an important milestone when it comes to new product funding. These results happening against the backdrop of market volatility in other areas of credit are a clear vote of confidence in our platform. We take the trust our capital partners have given us seriously and always treat credit performance as an uncompromising first priority. The average return of our last 12 quarterly vintages of loans exceeds U.S. treasuries by 651 basis points, with every individual vintage exceeding treasuries by at least 385 basis points.

Finally, the bank charter. In March, we announced our application for a national bank charter. As I said earlier, our strategy for funding loans is to rely primarily on third-party capital and the bank charter doesn't change that. We expect banks, credit unions and institutional investors to continue to purchase the vast majority of loans originated on our platform. Bank charter will, however, bring significant regulatory benefits to Upstart, including by expanding our addressable market across all 50 states, reducing the operational and financial cost of originating loans and accelerating our technology velocity by enabling us to interface with regulators directly. These benefits directly support our growth and profit goals and will show up over the next few years.

Now I want to close by welcoming Andrea, who joined us as CFO in March. Andrea is an incredibly talented finance leader with a background in complex novel business models. She's learning the ropes here faster than I could have hoped for and is already making an impact on how we plan, prioritize and execute. It is now my great privilege to turn the call over to her for a discussion of our financial results.

Thank you, Paul. I appreciate the warm welcome, and good afternoon, everyone. I look forward to spending more time with many of you in the coming weeks and months. It is a privilege to take my first earnings call as CFO. The Upstart team has built a highly differentiated AI-powered credit platform and the runway in front of us is enormous. I take seriously my responsibility as the financial steward of this platform, including the discipline Paul described around treating equity as a real cost and running a capital-efficient business. I spent my first weeks here digging into the business and the plan, and Paul and I are fully aligned on our financial priorities. I look forward to updating you on our progress each quarter.

Turning to Q1. Before I review the numbers, I'll provide some color around some of the factors Paul mentioned that were specific to the quarter and an expected part of our trajectory for the year. Starting with newer products. We continue to make progress growing our auto and home businesses and saw strong growth in super prime personal loan originations as well. This drove a sequential dip in our overall take rate and our contribution margin.

Next is seasonality. At the top of the funnel, we typically see consumer demand for personal loans soften in Q1 as tax refunds reduced borrowing needs. This soft demand typically translates into lower conversion and a modest step down in contribution margin in Q1 versus Q4. Additionally, our business has OpEx seasonality in the first quarter of the year with a step-up in corporate costs associated with our compensation and benefit cycle and the timing of our annual company-wide gathering.

Finally, we made deliberate investments in talent in Q1. This sets us up to achieve our objectives for 2026 and beyond. Each of these factors, mix, seasonality and investments, in addition to the platform and product gains Paul mentioned, was contemplated in the team's planning for the year. We are on track to deliver on our full year guidance.

Now I'll walk through our Q1 results. Originations were $3.4 billion, up 61% from the prior year and 8% sequentially. Within this, total personal loan originations grew 6% relative to Q4, reflecting 26% sequential growth in super prime and better than typical seasonal performance in our core business, which was roughly flat sequentially relative to the historical Q1 step down. Our newer secured products continue to scale with auto originations up 32% sequentially and home up 16%. Taken together, these results demonstrate the strength of our core business and the growing contribution of our newer products to overall platform growth.

Total revenue came in at approximately $308 million, up 44% year-on-year and 4% sequentially. This included revenue from fees of roughly $277 million, up 49% year-on-year and 4% sequentially, driven by growth in platform originations. Within fee revenues, servicing revenue continued to show solid growth, up 52% year-over-year and 22% sequentially, driven primarily by higher origination volumes, along with an increase in fees connected with the sale of loans. Net interest income and fair value adjustments totaled approximately $31 million, up modestly year-on-year and roughly flat with Q4.

Our contribution profit, a non-GAAP metric defined as revenue from fees minus variable costs for borrower acquisition, verification and servicing, was $137 million, up 34% year-over-year, but down 2% sequentially, primarily as a result of increased marketing investment to optimize digital channels and support new product growth. Contribution margin came in at 50%, down 3 percentage points from the prior quarter, reflecting the mix, seasonality and marketing investment dynamics. We expect contribution margin in Q1 will be the low point for the year, barring any changes in the macroeconomic environment.

In total, GAAP operating expenses were [ $16 million ] in Q1, up 45% year-on-year and 14% sequentially. Variable expenses, borrower acquisition, verification and servicing rose 68% year-on-year and 12% sequentially with the step-up versus Q4 reflecting marketing investments. Fixed expenses were up 31% year-over-year and 15% sequentially, reflecting the beginning of year investment and seasonal step-up in corporate costs I discussed earlier. I'll note that our fixed cost investments for the year were front-loaded into Q1, and we expect more modest sequential growth for the remainder of 2026. This sets us up for the adjusted EBITDA margin acceleration we've guided to as the year progresses.

In Q1, we had a net loss of approximately $7 million. GAAP earnings per share was negative $0.07 based on a diluted weighted average share count of 97 million. Adjusted EBITDA was roughly $40 million with a margin of 13%. With this quarter's results, we are on track to deliver on our adjusted EBITDA outlook of $294 million for the year and to be solidly profitable on a GAAP basis. We ended Q1 with just over $1 billion in loans held on our balance sheet, up approximately $30 million from Q4. It continues to be our strategy to primarily rely on third-party capital to fund our growing originations. Notably, our secured products and other R&D loan balance declined modestly quarter-over-quarter, even as auto and home originations accelerated.

More broadly speaking, as Paul mentioned, and supported by consistent credit performance, we've continued to enhance our capital platform. So far this year, we've signed more than $4 billion in committed capital partnerships, completed 2 securitizations for $1 billion in total collateral and increased the proportion of home and auto loans funded by a third party. Additionally, in February, we bought back 3.2 million shares of Upstart stock for $100 million, and we have about $122 million remaining under our current authorization.

Looking ahead, we are reiterating our full year guidance. This means that for full year 2026, we continue to expect total revenues of approximately $1.4 billion, revenue from fees of approximately $1.3 billion, adjusted EBITDA of approximately $294 million, which equates to approximately 21% of total revenues and consistent with our prior guidance. Our guidance assumes a stable macroeconomic backdrop.

Additionally, I'll share some color on the drivers and the shape of the year. First, for the full year, we continue to expect growth in absolute contribution profit dollars to be within at least 5 percentage points of fee revenue growth. We plan to deliver this profit growth using 2 complementary levers: growing our core personal loan business where margins are already strong and continuing to improve the unit economics on auto and home as they scale. Second, marketing and OpEx growth should moderate in the remainder of the year relative to what we saw sequentially in Q1. Third, we continue to expect adjusted EBITDA to be weighted towards the second half of the year, driven by originations growth, improved contribution margin and OpEx leverage as we progress through the year.

To close, our performance in Q1 was right on track. We entered Q2 with momentum across our core business and our newer products with consistent credit performance and with a reinforced capital base. I also want to thank Paul, Sanjay and the whole team for their support and partnership as I come up to speed.

With that, I'd like to turn it over to the operator to begin Q&A.

[Operator Instructions] Our first question will come from Mihir Bhatia with Bank of America.

Congratulations to both of you on the new roles. So just I guess in 1Q, you guys called this out a little bit. Originations were up 60%, revenues up 40%, but profitability declined. You highlighted some of the OpEx headwinds during the quarter. You're reinforcing the full year guide. So maybe just take a step back since you are both newer to the roles and just elaborate on how you're thinking about balancing near-term profitability versus reinvestment and growth. I guess, specifically, is there a framework guiding that trade-off? Or are you just comfortable prioritizing reinvestment at the expense of margins given the opportunity that you're pursuing?

Yes. I think the first thing I would say is that we are very much on track for our full year guidance. We are reiterating the guidance on the year. And so a lot of what you're seeing in Q1 is very specific to Q1, which are some seasonal effects and some sort of front-loaded investment effects. But I wouldn't say necessarily that as a matter of strategy, we've decided to make any changes to how profitable the business is going to be. We're very much on track for the annual guide, which does have $294 million of adjusted EBITDA there and net income positive.

So we feel really good about that. I would say taking a step back and just thinking about the business on a go-forward basis, we do think there's an enormous amount of growth ahead of us. We think we can keep compounding revenue at a high rate for a long time here, and we certainly intend to do that. I think the math is just really clear that in the long run, this business is going to throw off the most profits and be worth the most if we capture that growth in that market share. And so we definitely want to be reinvesting.

But as you heard me say earlier, we want to reinvest in a way that's capital efficient. And so a big part of the way we intend to do that is by growing in segments where we have high contribution margins and take those margins and reinvest those into growing the business. Some of that is from an accounting perspective, is going to show up as costs and some of that is going to show up in ways that will decrease the near-term profitability, but they are sort of investment choices that we think we can make in a capital-efficient way to grow the sort of total long-term profits of the business.

And then if I could just ask a follow-up on the revolving product some more. Talk about the availability of that? Is it broadly availability? And also just discuss the economics, funding partners, how you're thinking about that product evolving? It seems like a little bit of a credit card replacement. Is that the right way to think about it? I just -- yes, anything more on the revolving product?

We don't have too much to share on that product other than, yes, it's out broadly. It's called Cash line. We're really excited about it. It had probably the best first day we've ever had for a new product launch. I think there's an enormous opportunity, enormous need in the market. It is a revolving-like product. And as it grows and reaches scale, we'll naturally, just like with our other products, figure out the right third-party partners to work with on the capital side. But for now, it's early days. We're getting the product right, and we're very optimistic about it.

We'll now take our next question from Kyle Peterson with Needham.

I wanted to follow up on the prior questions on expenses. I guess, both in sales and marketing and G&A were a little higher than we expected. I know you guys called out some of the seasonality. So I guess I just wanted to see geographically, where did some of the seasonal expenses fall? And were there any costs related to the announcement that you guys are pursuing a bank application? And I guess, how should we think about that expense load moving forward as you guys go through the process?

Great. Thanks for the question, Kyle. So the expenses from a seasonality perspective are largely showing up in 2 places. So one is kind of on the payroll expense and with respect to our people that's really related to the annual reset of our compensation and our benefit cycles. And then additionally, I'd say primarily in G&A is where we see costs associated with our annual company-wide gathering, which happens here in Q1 and represents a seasonal uptick. And so those are the 2 largely the seasonal elements that we see here in the P&L. As we mentioned on the call, as we look out over the remainder of the year, our expectation is both on the marketing and the more fixed OpEx side for the business that we'll see more moderated sequential quarter-on-quarter OpEx growth relative to what we saw here in Q1.

Okay. I guess were there any material expenses from the bank application this quarter or nothing worth calling out?

That's a great point. There's nothing material this quarter. We are contemplating the expenses associated with implementing and sort of launching the bank in that application throughout the remainder of the year in the guide.

Okay. And then I guess just a follow-up on funding, it's been great to see the volume hold up. I know there's been a lot of concerns about private credit. I guess, could you guys give at least at a high-level update like where kind of your funding comes from in terms of what is maybe stickier institutional money with more permanent forms of capital versus some of these either interval funds or BDCs that probably are suffering a little more with redemptions and such. I guess just if you could size up the relative nature of those and I guess, how you guys feel about and feedback you guys are getting from your partners, I think that would be really helpful.

Yes. Great question and what I'm excited to talk about. Funding has been a real area of strength for us. It's been an area of strength because I think fundamentally, in the markets, the capital is going to flow to the places with the best performance. And the performance that we've had on the credit side and that our partners have been able to see over the last couple of years has been exceedingly strong. Earlier in the prepared remarks, I shared about the performance over the last 12 quarters and the spread to treasuries being very consistently high. And our partners have been able to see that. And as a result, year-to-date, we've been able to add over $4 billion of additional capacity. And most of that capacity is coming in the form of committed capital deals. So these are deals that have commitments over an extended period of time.

I mentioned that we now have our first 24-month deal. That's a new high for us in terms of how long these deals are committed. And that's so important to us because fundamentally, we can -- we have a lot of confidence in how our credit is going to perform, but we don't have a lot of control over what happens in the outside world with market volatility, market perception, what's going on in other categories of credit, whether it's software or something else. And so from our perspective, it's a huge derisking to be able to do deals with partners that believe in the credit and want to sign up for a commitment over an extended period of time that can get us through any potential market cycle that arises. And so we think of that as a real win, and we've been able to get a lot of partnerships going that have that extended commitment in place. And so we feel very good about where we are on loan funding.

We'll take our next question from Simon Clinch with Rothschild & Company Redburn.

I was wondering just on the point about funding and the long-term capital commitments, Paul. And when we think about the signings you've had recently, could you perhaps just describe whether the economics of the sharing of risk have changed in any way or how that's sort of evolving as these newer deals are being renewed?

Yes, sure. So we do have risk sharing as a component in many of these deals, and you can -- investors can see some of the details about that in our earnings presentation. I would say those deal terms have largely stayed consistent to improving over time. And certainly, like anything as we do more of these and we do them at greater scale and we sort of prove how they work, we expect over the long run that the terms will work better and better for us. But certainly, they've been consistent and consistent and improving.

And so there is a risk-sharing piece of these deals that is capital that we think is well spent. It's a very small percentage of all the capital that funds these loans. And increasingly, that is capital that, from our perspective, is expected to earn a quite healthy, strong return. So capital well spent and certainly fits within the framework of running a capital-efficient business that we talked about.

Okay. Great. And just as a follow-up, just going back to the balance sheet. They were up marginally sequentially. I think usually, when we see a lot of new funding commitments or agreements coming through, sometimes they come with upfront purchases of loans off the balance sheet. And it doesn't look like we've seen that. I was wondering, could you talk about the dynamics of that, please? And is $1 billion really the kind of the level that we should be expecting on a go-forward basis?

Sure, Simon. Yes, I'm happy to take that one. So the -- really, what we see, we expect to see some degree of normal course fluctuation on the balance sheet side on a quarter-on-quarter basis, just driven by the timing of sales, and that's largely what we saw here in Q2. It is our expectation that as we look out over the remainder of the year to see some step down in that overall balance in the rest of the year relative to Q1.

With respect to the dynamic of back book sales versus forward flow, we saw both, a mix of both on the balance sheet in the quarter. I think what was very good to see when it comes to the strength of the capital platform is in Q1, we did see a higher proportion of our originations on auto and home sold directly through to third parties versus what we saw in Q4 and in 2025. So making very good progress on that front on the secured products. And then otherwise, just have this timing dynamic on the balance sheet that we do expect to see from time to time.

We'll now take a question from Dan Dolev from Mizuho.

Congrats on the quarter. I got 2 quick questions. I'm looking at, I think it's Slide 22, and that's the expected cash flow versus the upside downside. Can -- I mean, it looks like the trend is widening there. Maybe can you explain some of the dynamics? And then I have a quick follow-up.

Sure. I'm happy to speak to that, and then Sanjay can chime in here as helpful. So basically, we sort of see it widening out, that's really reflective of just the increased capital co-investment amount. And so you sort of see the max upside, max downside is really just reflective of the total dollars that are at stake and co-invested here. Does that answer your question?

Yes, it does. It does. I do have a quick follow-up. Can you just give some comments about the overall health of the consumer that you're seeing? I think it will be helpful for investors and congrats again.

Yes, certainly. We see the consumer -- the American consumer as largely stable over the period. In fact, we've seen the consumer largely stable really since late last year, and we've been kind of in a pretty tight range of what we call the Upstart macro index. And from our perspective, stability is a really good thing. That's all we ever ask. Certainly, an improving consumer could be a tailwind, but a stable consumer is a good one from our perspective. And that's what we've seen -- we are -- like everybody else, we watch developments, but probably where we're unique is that we are very committed to being a model-first, model-led company. And so we let our models detect all that's going on in terms of consumer repayment patterns, both in the aggregate and at the segment level. And we just -- we haven't seen any of the factors in the news come into play in a significant way yet. And so consumer is stable.

Our next question will come from Peter Christiansen with Citi.

Congrats on the committed capital level. That's great to hear. Just would love to hear your take, at least on demand that you hear more on the [ at-weel ] side from some of your bank partners there? And then I have a follow-up.

Yes. It's -- and feel free to follow up if this isn't what you're asking about. But we have been announcing and signing deals with partners, both that are sort of traditional financial institutions, banks and credit unions as well as with institutional investors that are private credit or another form of institutional capital. So both sides of the demand for loans has been very healthy and growing. That's, again, against the market backdrop where there have been concerns in other categories of credit, in particular, in software and some other areas. But for us, because of the strong performance, the demand from both types of institutions have been very strong, and we've been doing deals in both places.

That's helpful. And I just want to dig into conversion rate seasonality a little bit here. You did have a little bit of a step down last year in 1Q as well. Maybe it was a little bit more profound this year around. But generally, as we look at the conversion rate and how it progressed last year, kind of peaked in 2Q and then like leveled off and stayed fairly flattish in the second half of '25. Should we think about that progression being the same or at least expectations right now for the remainder of '26? And at least on the 1Q sequential step down, which is seasonal, it seems like it was a bit more than in previous years. Just wondering if you had any additional comments on that.

It's a good observation. You're right about that. So there's 2 different effects going on with conversion rate. The first is a seasonal effect, and that happens every Q1. Every Q1, there was a noticeable reduction in borrower demand for loans related to tax season and tax refunds. And that happened this year just like it happens every year and is an important part of the story. But with respect to the conversion rate metric specifically, this particular metric has a lot going on in it. It used to be a much simpler metric when we really just had one product, one segment, what we now call core personal loans. But now because we've got a mix of products in there that serve consumers up and down the spectrum, there's pretty significant mix effects going on here. And this metric in recent quarters has become increasingly affected by our small dollar product, which is still a relatively new, not totally mature product.

But the small dollar product, because they're very small loans, don't have a big impact on the sort of origination dollars or the financials of the business, but they do have a quite outsized impact on the unit count conversion rate, just like how many loans in the numerator converted. And so we did have a decline in the small dollar product in Q1, and that had an outsized impact on the conversion rate metric that we cite. So this is something we'll think about how to improve going forward. But just from a metric perspective, I mean, but from a core personal loan perspective, that business actually had very stable conversion rates, unseasonally strong conversion rates and unseasonally strong volumes as we talked about earlier.

That is super helpful, Paul. I appreciate it. Sorry, just one follow-up on that. considering Upstart macro index is doing marginally better at least on a trailing basis. Should we expect some of that small dollar mix shift impact perhaps bleeding a little bit into 2Q?

We don't have any specific guidance on the small dollar product volumes at this time. I would say that we -- the seasonal effects, of course, will run off as we get further into Q2 just as we get past tax season, obviously. So that will no longer be a factor. But you are right that there is an effect where small dollar sort of sometimes can move inversely with core personal loans just because of -- because it sits after core personal loans most of the time in our approval funnel. So that can be a dynamic, but we don't have any specific guidance on the small dollar numbers. They're not a very large part of the overall financials right now.

Our next question will come from James Faucette with Morgan Stanley.

Just a couple of -- maybe a follow-up on forward flow agreements and then just a more strategic question. So on the forward flow details, obviously, you've been really active there. But just wondering on top of the ability to sign those, are you seeing any change from those partners with respect to target gross yields or return on equity? Any internal metrics that are changing at all, especially given kind of the environment that people have pointed to?

Well, as I said earlier, our -- we've been able to do these deals against a challenging macro backdrop, and we've been able to do them with largely consistent to improving deal terms, and that includes the kinds of spreads that people are looking for above benchmark rates. So -- and that's -- ultimately, it's all downstream of credit performance. If credit performance wasn't good, then that wouldn't be true. We might not even be doing some of these deals. But because credit performance is strong, ultimately, everything else is downstream of that. The amount of spread you need is a function of how much risk you perceive there to be and underperformance and all of that. So we've been really happy with the way we've been able to do these deals, and we expect to continue doing them.

Got it. And then I wanted to ask about the HELOC product. Obviously, really good growth and the highlight of being able to look at a 6-day or so process versus up to 40 days as being the industry norm. just talk about where you're seeing, at least in these early days, that speed advantage show up. I mean I've got to imagine maybe it's everything, but wondering if you're seeing outsized benefit in higher conversion, lower CAC, loss selection, partner appetite take rates, anything like that? And where is that mix coming from or what's driving that? Is it cross-sell from personal loans or direct-to-consumer? Just trying to get a little flavor of the type of customers you're seeing in HELOC, what they're responding to and where they're coming from.

Yes, you're absolutely right that being able to run a 6-day process is huge. It's a huge advantage in the HELOC product. And -- it does manifest in all of those places that you listed. You get better conversion, better conversion necessarily means lower CAC. The other place that is really, really directly impactful for the HELOC product is in the operational cost of originating one of these products. Every time you can move a loan from one that has a heavy dose of sort of manual work involved to one that can either be fully automated or just require a tiny bit of manual work, there's very significant op savings there.

I talked earlier about how we're now turning our attention towards optimizing the margin profile on these new products. And a big part of that is right in here of getting the process right and getting the cost down associated with that. So technology and getting that 6-day process are just a huge part of making that happen. From how do we find customers for a HELOC perspective, it's also -- we do a wide range of things, but probably compared to our personal loans product, we do have a heavier dosage of cross-sell from the existing customer base, and that will be an increasingly important part of our strategy going forward just because as I talked about, we have a growing and very large population of people that have established a relationship with Upstart.

Over 20 million people have created accounts at some point to check their rate. That means we've got a lot of information on these people. We've got a relationship with them. They think of us as a place for credit. And as we get more offerings that can serve Americans across the full economic spectrum across all types of credit needs, that's going to be a really powerful thing, and it's already showing up in HELOC and our ability to cross-sell.

We'll now take a question from Will Nance with Goldman Sachs.

I just wanted to follow up on the commentary around the 5% spread between CP and revenue from fees. And I was wondering if you could talk about kind of the framework for thinking about that, particularly in light of the ramp of new products, which I understand put some pressure on that spread, but maybe isn't something you want to artificially throttle. So can you just talk about the puts and takes there and what would cause you to come in above or below that level?

Thanks, Will. And yes, you're sort of hitting the nail on the head there. And so we're looking to grow contribution profit within 5 points of fee revenue growth this year. That is sort of the lag on contribution profit dollar growth relative to revenue is primarily driven by mix and the strong growth in our newer secured products as well as in prime. And it still represents very substantial sort of contribution profit dollar growth against the platform throughout the year.

And so if you look at the things that are going to drive that contribution dollar growth on the platform this year, it's really the 2 things we alluded to in our comments earlier. First is continuing to lean into the strength on our core personal loan product and to drive the growth of originations there, which are quite accretive from a margin perspective. And second is going to be driving the continued growth of the secured products alongside the continued improvement on the unit economic performance of those products. So as those products grow and scale, that will contribute to improved contribution margin as we continue to drive more automation in our process and reduce friction in the process that will help improve the unit economics there. And as we continue to improve and increase the sell-through of the product off the balance sheet, that will also help drive the unit economic performance of those products, representing sort of a tailwind to the contribution profit dollar expansion throughout the course of this year.

So that's really how we're thinking about it. We are looking to continue to sort of hit that number on a contribution profit dollar basis, very important from a platform perspective. And those 2 levers are the things that are going to drive it.

Got it. I appreciate that. And then if you could just talk about -- I think you had a comment around seeing the early signs of acceleration. You mentioned unseasonally stronger contribution margins and better than seasonality performance in the core personal products. So just maybe talk about what you're seeing that is allowing you to outperform seasonality and make the comment that you expect to accelerate that over the course of the year?

Yes. So these comments were in reference to the core personal loan business. Core personal loans, you can think of this as our historic personal loan product offered to consumers that are not conventionally defined as super prime. And so these borrowers have long been the place that our business has had the largest competitive advantage. We have a very large amount of differentiation in our ability to underwrite these borrowers compared to what the market offers. And as a result of that, we've had very strong pricing power historically in this segment.

Over the last year, I would say that we've been very focused on growing -- establishing our foothold in new products, especially in home and auto and also balancing out the platform by getting very competitive and having a best rates -- a set of great rates to offer very prime customers. And so we've had maybe relatively less focus on this core personal loan segment. And so we've -- now with the success we've had in home and auto, we've been able to redirect more focus back to the core. And so we just see that growth is basically driven by investments in technology, improvements in that funnel, improvements in marketing directed towards that customer. And so we've just been doing those things. Those are all kind of durable improvements that we make, and we make those week after week, month after month, and they compound, they add up.

And so we're starting to just see some of those benefits come into the Q1 results. That's why it was able to beat its sort of seasonal expectations and run ahead of those, and we expect that to continue through the year where we keep reinvesting back into this product and make -- widen its technology lead, improve its sort of marketing, get -- reach more people. And by doing those things, we will see this product grow more, which in turn will generate more contribution profits for the rest of the business to use and reinvest.

And just put some numbers on what we're seeing here in Q1 versus previous year. So in Q1 '25 and Q1 '24, core personal loans saw about a 10% quarter-on-quarter decline. And so this year, we're seeing flat originations. And that's sort of what we're looking to additionally from a numbers perspective to speak to that better than seasonal performance.

Our next question will come from John Hecht with Jefferies.

You guys talked about some of the seasonal factors and product shift changes with customer acquisition costs. But I'm wondering, is there anything going on at the unit level? Like have you seen any changes to origination fee structures in various products? And/or are you exploring different channels of customer acquisition? Anything going on there just to talk about that piece of the business?

No large fundamental changes there. I would say we still use the sort of a full range of marketing channels that we used before. We've made improvements across many of those leading to some of the wins and better than seasonal numbers that we've had. We have largely -- we talked last quarter a lot about our intentional strategy not to not to max out on take rates from borrowers, and we've stuck to our strategy there. So we are very intentionally not maximizing profitability of the business. If that was our North Star short-term profitability, I think we could have a lot more of it by squeezing more out of take rates and fees, especially in certain segments. But that's not our strategy because we don't think that is the best way to maximize the long-term value here.

We think there's just incredible value associated with winning over customers and building relationships with them and leaving a little bit of extra on the other side of the table. So we've been doing that. We've stuck to that strategy with respect to how we think about origination fees. And when we go out and do marketing, we certainly keep in mind that it's valuable to win over a customer, and it's not all about maximizing the profit on day 1.

And then with that in mind, any comments on whether it's recurring customer activity or maybe even think about as direct-to-customer activity or cross-sell? Any kind of signals you're seeing there?

Yes. We certainly do both. We think it's really important to have to have a lot of repeat customer activity. We are increasingly focused on what we think of returning user activity. So these aren't necessarily people that got loans with us before, but you look at that $20-plus million number, those are all people that have decided to come check their rate from Upstart at some point and maybe they couldn't get approved the first time, maybe they couldn't get the sort of type of loan or the dollars of loans that they were looking for the first time so they didn't accept. These are all perfect candidates for people as we've got more and better products to go back to.

And so certainly, we do more and more of that, and that's something that we want to maximize. But we're also still so early in our growth journey, 20 million is just a fraction of the U.S. population. And as the player in the market that uniquely, we think can serve the entire spectrum from having great rates for people on the very prime end to having great offers for people at the other end of the spectrum, we think we can be sort of a full spectrum offering. And so we think our addressable market is a lot more than 20 million Americans. And so we want to keep adding new people into the database. And so we're going to keep marketing to do that. And that -- you put those together, and I think in the long run, you're going to have a very, very valuable business.

Our next question will be from Patrick Moley with Piper Sandler.

Just wanted to go back to the Bank Charter. Could you walk us through maybe some of the key regulatory milestones ahead there? And what's the expected time line before you start realizing some of these operational and financial benefits that you talked about?

Yes. We're excited about the Bank Charter. As you mentioned, the benefits are primarily regulatory. So just to clarify in case that was lost on anybody, but we expect to get nice improvements that are both operational and financial out of doing the bank, but it is not a balance sheet strategy. It's not something that we're doing to change how we fund loans or how much capital the business needs to operate.

In terms of process from here, we have submitted our application with the OCC. It's for a national bank charter, and we're working with the OCC on pieces of that application. We don't have any specific guidance on the exact timing of this. That's going to come in our work with the regulator, but we're very motivated, and I think the regulators have been really constructive in how they've worked with us and other companies in this current situation.

Okay. Great. Then maybe just a quick one. You bought back $100 million of stock in the first quarter. I think you have like a little over $100 million left on the authorization. How are you thinking about the pace of buybacks, I guess, throughout the rest of the year? And how do you balance that kind of with some of the balance sheet co-investment and new product funding needs?

Yes. Well, I'll go back to saying that we think capital efficiency is really important to us. We want to think of equity capital as a real cost. We want to think about metrics in per share terms as often as we can. And so in the long run, we're just going to be thinking about how do we maximize the earnings of the business and how do we minimize the amount of dilution in the business. And so whenever there are opportunities afforded by a combination of considerations of available cash and liquidity and financial outlook and the price is right, we're going to be looking at an opportunity to use those stock buyback dollars.

Now having said that, the reason we aren't like always buying back all the time is that we just have so much growth ahead of us that the threshold for doing that is really, really high. We know that there are so many growth opportunities that we can invest in operationally that our threshold for using cash for any other purpose is going to be really high. But yes, once in a while, I think we'll have that opportunity in the market. And whenever that is, we will certainly consider doing it.

We'll now move to Vincent Caintic with BTIG.

I wanted to follow up on the bank question. And I actually wanted to focus on the economic implications of becoming a bank. I thought it was interesting on Upstart's blog post when you highlighted about $200 million of annual frictional costs and also the lack of being able to be in certain geographies or serve certain customers. I wondered if you could elaborate on that. How difficult is it? And could you really remove $200 million of annual frictional costs that would be big versus your EBITDA guidance of the $294 million. So I'm just kind of wondering like how difficult is it to achieve those things? Do we see that in economics in terms of EBITDA? Or does it come from higher transaction volumes or some combination?

Yes. So a lot of that 200 number is really a missed opportunity on the revenue line. So I think it might show up in a different place than you would think if you were just thinking about these as true kind of like friction costs. But the way it manifests is a few different things. The first is that we have a number of states and segments of the market where we can't operate or we're limited in how much we can operate because of state-level regulatory issues and having a national bank charter to operate through resolves most of those and gives us access to sort of the full market up to the 36% rate limit. So that's a big deal. I mean that's just TAM that we're missing out on today that directly gets solved by having access to the National Bank charter.

The second issue are direct kind of operational and financial costs associated with the way we operate today, and that is we originate loans through a large network of many financial institutions, and that comes with both direct financial costs that are paid or earned by the financial institutions instead of us or the cost of friction in managing that sort of complex system of many players originating.

And then the last is -- so those, I would say, are the kind of really concrete costs that go into that $200 million number. And then there's a separate thing, which is more of an intangible. It doesn't go in that number, but I think it's just as important. And that is that it's obviously, we think, a decade where there's going to be substantial advances in AI that AI is going to do a lot to transform consumer credit. And I think every regulator is going to naturally be asking questions about that, wanting to understand that, wanting to work with the leading frontier companies in doing that. And we just felt that from our position as the company that's been doing this longest, we should have a direct relationship with the regulators in helping them understand what it means to apply AI in the context of lending, help them get that right and do that directly as opposed to trying to do it through a large number of intermediary financial institutions.

Okay. Great. That's super helpful. Switching topics, if you could maybe kind of going back to the topic about take rate. Seeing the transaction volumes grew 61% year-over-year and your overall revenues grew 44% year-over-year. I'm wondering if you could kind of talk about how we should expect that dynamic going forward. And when you talk about, say, improving the unit economics of Auto and Home, I know you said that you didn't want to maximize profit, but does part of improving the unit economics involve the revenue side? Or is that primarily on the cost side?

Great. Thanks for the question, Vincent. So with respect to the take rates we're seeing here in Q1, those are largely a reflection of the dynamics we've spoken about previously. So it's about the growth of our newer secured products as well as mix shift to prime in personal loans, all of those have been growing very well. And so those have largely manifested in take rate that has come down year-on-year as expected, right? And that's something we were expecting here in Q1 in addition to the seasonality dynamic we spoke about before.

So we typically do see some softness in take rate in Q1 relative to Q4 kind of associated with the softness in demand, all of which is quite expected.

As I sort of step back and think about the remainder of the year, a couple of things. One, take rate is not a metric that we -- it's really an output metric for us, not kind of like an input metric, if that makes sense. And so largely, as take rate moves in the business, it's going to be reflective of kind of changing mix on the products. That being said, as we alluded to before, a key driver of kind of the ability of the platform to deliver contribution profit dollar growth this year is going to be the improving unit economics on our auto and home products throughout the course of the year, where we've already seen meaningful progress on that over the course of the last year, but we expect that progress to continue.

And to your question, that should show up sort of up and down the P&L. So it will come from both improvement in efficiency on the cost side, driven by increasing automation as well as just the benefits of scale and kind of having more loans kind of moving through the platform. And we do expect it to show up in improving take rates on average across the sort of secured product set as we look throughout the remainder of the year, driven primarily by an expectation of increased sell-through of loans off the balance sheet to third parties.

Our next question will be from Giuliano Bologna with Compass Point.

Congrats on the results. As a first question, last quarter, you mapped out an expectation of around $100 million of revenue from HELOC and auto between a ballpark 4% upfront take rate and 2% servicing. Is that still the rough expectation? And then when I think about that, one thing that I noticed this quarter is that within the servicing line item, there was a step-up in kind of the other fees, which has historically been de minimis it's closer to $3.9 million. Is that anything related to servicing some of those HELOC or auto loans? And how should we think about that going forward?

Yes. Thanks for the questions, Giuliano. So on the first point on the $100 million or so fee revenue from auto unsecured, that continues to be sort of in the right ballpark in terms of what we expect. The sort of the take plus servicing that Sanjay spoke to last quarter represents more of, I would say, sort of the medium-term take rate for that product set in aggregate. So may or may not fully achieve that inside of 2026, but sort of more over a 1- to 2-year time horizon as those products grow and scale.

Go ahead, sorry.

On your second question on the other fees and servicing, I might need to follow up with you on that one, but I can certainly do that.

That's very helpful. And then thinking from an execution perspective, as you sell more of those loans you flow more through, should we expect servicing fee revenue to be an incremental driver, especially on the margin side because you're probably spending disproportionately on marketing and other expenses on the front end, but then a lot of the revenue from the growth in those new products is really deferred and realized over time?

I -- I think I'm understanding your question. And yes, that would be the case. So as we sell these loans through, it will -- in terms of how it impacts revenue, will allow us both to recognize kind of fee recognition or take rate upfront upon the sales of those loans. But especially on our auto product, where we expect to have a higher proportion of the compensation come from servicing, we will also be generating a servicing revenue that we will recognize over time and will offset those servicing costs that we bear.

We'll now take a question from Rob Wildhack with Autonomous Research.

I just wanted to follow up on that comment around the medium-term take rate and servicing rate for the HELOC and auto. In your experience, do you have any sense for how long like a new product takes to reach like mature unit economics? Andrea, you mentioned like maybe not 12 months, but 12 to 24 months. Will they be mature by then? Or is there scope to improve beyond sort of that 1- to 2-year time frame?

Yes. Great question. The beauty of our business is that there's not really any moment where we deem a product mature because the margins ultimately are driven by the level of differentiation that have been created by our technology. And so it was a lot of years before our core personal loans product reached its current level of take rates and its current contribution margins. We were doing that for probably 7, 8 years before we got to levels like the ones that we have today. And that's basically because year after year, the models kept getting better, the level of automation kept getting better. And that's really the thing that allows you to be able to get pricing power in that market.

When the next best offer is so far away that you can increase the take rates there and still have the best product. That's where it makes sense to start. ramping that up. So I would say we are very much in a position today where these products have found a great fit in the market. We are ready to start optimizing their margin profiles. But it's not like they're going to be done optimizing this year or even next year or probably even the year after that. They're going to just keep getting better as we create more space for differentiation and therefore, more space for pricing power. And that's -- we view that as a fundamentally very good dynamic in our business where we can keep on improving the technology that keeps on creating more differentiation, which then allows us to have more pricing power. But we do expect these products to get meaningfully better in their margin profiles in the near term.

Okay. And I just wanted to ask too about the cash line product. Can you talk a little bit more about who the target user is for that? Is it something you see as a complement to personal loans or maybe more of a substitute for someone who doesn't need or want $15,000, $20,000. And then what about the economics there, like origination fees? And do you expect that those are balances and loans that will sit on your own balance sheet? Just any color you could provide there would be interesting.

Yes. Cash line is a product designed for someone who has generally a need for a smaller amount of credit, but on a more regular basis. So if you think of personal loans, there's a one-shot $10,000 loan. A cash line is -- we're talking about hundreds of dollars, but something that you might access multiple times a month, multiple times a quarter. And so the design of the product needs to be very different. But it's actually still a really valuable loan, a really valuable customer because it's accessed so frequently over a period of time. And so that's who it's designed for. In terms of its funding, yes, in the -- in the medium term, as that product grows, it's going to have its own sort of dedicated funding partners that are the right home for it. But because the product is so short term, the loans are so small, it's not a major factor in the balance sheet today, and it's not something that we're particularly worried about.

And we'll now take our next question from David Scharf with Citizens Capital Markets.

I had a couple of questions, but both related to how we ought to be thinking about the impact of kind of the shifting origination mix, specifically more prime personal loans, HELOC by definition, is going to be more of a prime borrower. And first, just focusing on capital. It looks like 1/3 of your retained risk on the balance sheet is the beneficial interest, sort of the co-investing. Should that relationship or percentage drop over time as a bigger part of your business becomes prime focused? Because it seems like the investor part of the marketplace is going to require risk retention to a greater extent, the lower rate of the credit. Can you just kind of walk us through if the balance sheet is going to change as the mix shift of your originations change?

Yes, it's a good question. I think I maybe want to start by just correcting a misconception about the required level of risk sharing. I mean, really, when we thought about -- when we looked at our business over the last few years, strategically, one of the most fundamental convictions we've had was that we had confidence in our credit, but we didn't have control over what happens in the outside market. And so there would be market environments where funding was less available and there would be market environments where funding was more available. And much of it would have -- much of it would be outside our control even if our credit was performing well.

So we thought it was strategically critical to us in our desire to build the sort of largest provider of consumer credit that we needed to have capital arrangements and capital partnerships that could endure through market cycles. And so the -- our putting in risk -- a small portion of risk sharing into some of these deals was fundamentally in exchange for having committed capital over multiyear windows. And so we don't view that necessarily as a requirement and certainly not a bad thing. We actually view it as something pretty innovative that allows us to solve one of the most important sort of fundamental challenges in a fast-scaling business like ours.

Coming back to your question on newer products and prime mix and how that will affect it. I think you are right that the sort of primer products will tend to have a lower sort of level of capital required from a risk-sharing perspective. But I wouldn't necessarily leap to the conclusion that the ultimate mix will be any particular sort of mix of primeness. And that's because we have so much growth in multiple product categories right now. We're putting a lot of focus on growing that core personal loan segment, which is an extremely strong segment for us, a very profitable segment. We have a lot of growth in the auto product, auto, which is a product that every American across the entire economic spectrum needs. And you are right that, of course, the HELOC product is -- it tends to be a more prime product. But depending on the sort of exact rate at which each of these products grow, you could end up either with any particular mix of primeness.

Got it. Understood. That's helpful. And maybe just as a follow-up, sort of applying maybe the same rubric to kind of the medium term, the 3-year guidance. The -- I mean, 35% revenue CAGR is obviously significant and margins going from 21% to 25% is material as well. But the margin increase is possibly maybe not as much of a pickup as we would expect with revenue kind of more than doubling, increasing 2.5x over that time frame. Is that related at all to kind of product mix? Or should we just interpret that as 3 years from now, you still anticipate being in growth mode and probably still in an above-trend period of investment spending?

Yes, more of the latter. So we are expecting this business to grow at a significant amount for a long time. Obviously, we gave the 3-year guidance of 35% CAGR. But maybe the interesting thing about that from the perspective of your question was that 35% was approximately the same amount of growth that we guided to in the first year as the later years. And so maybe you might infer from that, that it's not heavily front-loaded like it might be for a business that you expect to grow a lot and then basically just taper off. We think the market opportunity here is so large across so many new categories of credit that we hope to be doing this for a very long time. And so it is our working expectation that there will be great opportunities to reinvest profits into continuing to grow the business over time. And I couldn't tell you today, given we've got 3 years of guidance, I couldn't tell you what's going to happen in year 4, year 5, year 6. But if I had to guess, I would guess that there's going to be great opportunities to reinvest for some time. And so that's what's in our plan today.

And it appears there are no further telephone questions. I'd like to turn the conference back to Mr. Gu for closing comments.

Great. Well, thank you, everyone, for your questions and for your time today. I just wanted to leave you with a few things I hope you can take away from our conversation. First, Q1 was strong and puts us comfortably on track to deliver our full year outlook on both revenue and profit. Second, core personal loans is our superpower. It has great margins, and we're going to do a lot more of it. Third, home and auto have found their places in the market, and it's time to make them profitable. And finally, the opportunity ahead for us for AI to remake consumer credit is enormous, and we intend to go after it while making every dollar of capital count.

Thank you to our team, our capital partners and our shareholders. We look forward to seeing you next quarter.

And once again, that does conclude today's conference. We thank you all for your participation. You may now disconnect.

Afternoon, everybody. Thanks for joining us this afternoon here at the 2026 Morgan Stanley TMT Conference. Very excited to have Sanjay Datta, President of Upstart. Thanks for being here.

Thank you.

Before we get started with Sanjay, quickly, I've got -- I'm going to read a couple of disclosures. We have ours, which is please see the Morgan Stanley research disclosure website at morganstanley.com/researchdisclosures.

And then Upstart has asked me to read theirs. Today's discussion may contain forward-looking statements that relate to future results and events, which are based on Upstart's information available as of today and are subject to risks and uncertainties. Actual results may differ materially from these forward-looking statements. The discussion may also include non-GAAP financial measures, which are not a substitute for GAAP results. Please refer to the company's filings with the SEC and its IR website for additional information, including GAAP to non-GAAP reconciliations, along with other disclosures. Did I do it all right?

I feel like your disclosure is a lot more cogent than ours...

Well, I think we get away with referring people to -- anyway, thanks for being here. Really appreciate it.

Look, I want to start with just kind of like some of the leadership changes that have been announced. I don't know that they were under adverse circumstances or anything like that, but it does create questions for people. How should we think about strategic continuity versus change with the leadership transitions that have been announced, where should investors expect meaningful change, if any, in priorities or pace over the next 12 to 18 months?

Yes. I think this has got to be one of the leadership transitions with as much continuity as you could hope for. I mean, Paul and Dave have been covering the company since inception for the last 14 years, they've essentially been co-decision makers and aligned leaders throughout. And they started planning this transition -- I mean, I knew about it as of 5 years ago.

So it's been -- they've been working towards it. And it's sort of like -- it's like Dave was running the company, and Paul was his co-leader, and now Paul will be running the company, but Dave is not going far. He's still the Chairman and the biggest shareholder. So he's still [ flocking ] us and giving us his opinions. And -- so I don't -- nothing will change with respect to our priorities with respect to our objectives and strategies.

Got it. So I want to talk about another announcement that you made and one that I know that we in the investment community grapple a lot with, and that is the periodic disclosure of KPIs and how that may impact volatility. I know certainly, I feel like there's probably a lot of opportunity for investors if there was a way to tamp down volatility of stock itself. And one of the things that it seems like it could be a step in that direction was your decision to start to publish monthly transaction volumes.

That being said, is like anything, monthly transactions or any shorter period of time has the potential for more variance from period to period. How should we investors separate what is signal versus noise, particularly in month-to-month prints, whether that be seasonality, marketing cadence, partner onboarding? How should we try to digest those numbers as they come out?

It's a good question. We've sort of done two things with our guidance framework. On the one hand, we went more short-term transparency, right, from just quarterly guidance to like monthly updates on volume. On the other hand, with respect to guidance, we're now emphasizing the longer-term view, the 1-year view and beyond.

And really, what we're doing there is on the short-end side, we're trying to provide real-time transparency to what is going on. There's so much speculation out there and so many third-party data sources trying to figure out what's going on. And they're adding to the noise because sometimes they're right and sometimes they're wrong and...

And sometimes they're right by a little bit and sometimes they're wrong by a lot...

Yes, it's created its own dialogue. And we are just like, look, just -- we'll just put the facts out there. And so it's kind of like maybe just kind of desensitize that a little bit.

Now with respect to what within that is noise and what is signal because I do think like people tend to get maybe overly wrapped up in -- with this month versus last month. So I think the way we'll sort of signal what's important is when we come back and say, well, here's how you should think about how the year is change -- like if there's a trend in those things that are useful. We're going to come back and tell you that the year is kind of changing. And we'll give you a different outlook on the year to sort of reflect that.

But short of that, I think a lot of it is just there's like different calendar days or different business days, depending on how the weekends fall in certain months versus others, there's seasonality, there's all these other things. And so yes, I think what we're going to really try to do, if you're interested in the signal of the business, is really telegraph that through our evolving view of how the year is going.

Right. And it seems like, to your point, is like it's going to put incremental -- create incremental opportunity for us on the investment side of things to try to take into account those things. But if I put it together with the way that you disclosed UMI, et cetera, it seems like there's a lot of message and signal, if you will, that we should be able to deduce as investors.

So let's talk about the additional piece of this communication, and that's the multiyear framework. And give us a little bit of insight, if you can, into the thought process around providing a multiyear framework like you did, especially given the cyclicality inherent in the business, right? Like there's a lot -- and by cyclicality, I mean, the economy is changing, it's you're an economically -- cyclically sensitive business in some ways, et cetera.

Yes, that's an important point. I mean, one of the assumptions underpinning that longer term, I won't call it a guide, but it's the targets that we're operating against, if you will, is that it will -- it's sort of what we believe in a roughly macro neutral environment.

Now the macro won't be neutral probably. Who knows if it will be a headwind or a tailwind, it may be both in that time frame. But really, what it's meant to say or meant to sort of connote is in that kind of a time frame, the underlying growth characteristics of the business will be a secular one. And the secular growth dynamic in our business is related to getting better base models and more automation over time.

So as we do that underlying work, we get models that do a better job of avoiding default, which then reduces APRs for all the other applicants, who are not subsidizing as much default. And you get more automation, meaning you take friction out of the funnel, you take documentation out of the process, and that also acts to improve conversion rates.

And so I think the road maps we have and the backlogs we have of all the R&D projects that are going to chip away at those things, incrementally more automation, incrementally better risk evaluation is long enough that we're signaling that we think secularly, we can -- we're very confident we can maintain that kind of a growth clip over the coming 3 years.

And so it's really meant to telegraph the fact that -- I think -- because there maybe was a mental model out there that's like, okay, I see it's a pretty good core business. It's going to start cash counting in 2, 3 years. The margins will be -- I think what we're saying is, no, we've got a big growth curve ahead of us. Like the out-year outcome of our business should be a very big one. And the sustainability of the growth, I think, in the next 3 years is something we feel very confident in. So that's the main signal.

Now of course, there will be macro puts and takes to that, probably. We can't predict what they will be. And so from that perspective, yes, we're not trying to put a stake in the ground and predict that. But I just -- we're trying to signal that in the underlying growth characteristics of the business, there's a lot of -- there's a lot more persistence that we see.

Got it. So let's talk about some of the components of that multiyear framework. And part of that is the scaling of secured risk -- secured lending. And I guess the thing that a lot of times we're asked is, hey, the unsecured personal loan market is still really large. What is the strategic necessity of scaling into secured lending? Is it risk diversification, capital durability? Are you looking to expand TAM? Or is there something else within that?

Yes, there's no necessity to do it in a sense. I think we could build a nice big business with healthy margins in our core. But I think this is like the difference between big and massive, right?

The secured categories, I think the rationale for why we can and should be successful in them is the same as in the unsecured categories. There's always a combination of too much friction to access to credit and like a significant part of the market that's not getting access because the risk models aren't good enough. And the TAMs on that thing are like multiples bigger on the unsecured side.

So I kind of view it as like, yes, no, we're very excited about the opportunity in the unsecured world and in our core business. But I think there's also this like there's a bit of a race that we're running alone right now, but we don't expect to be running alone forever. And it's like who's going to bring machine learning models into these new segments of credit and then ultimately, for maybe new geographies as well.

And I think there's a huge first-mover advantage because you start generating the -- what's interesting and unique about applying AI to the credit space, unlike a lot of what's being done out there, where the training data is just hoovering up the Internet. The training data is not out there. You have to generate it. And so whoever is first, whoever starts generating that training data, not just through repayment history, but all the variables you collected against it. Where did that borrower study and where do they work? And what's their title and what's their role? And how do they interact digitally with my application? And when I ask them their FICO, how close were they to knowing it before I check -- like all those things provide a richness of data that then train the models. And because it's not out there on the Internet, you have to go and generate it.

And so there is a bit of a first-mover advantage in taking AI into these spaces. And so we're just trying to plant the flag in these areas. We think there'll be massive opportunities. And it's not to sort of diminish the opportunity that we feel exists in the unsecured space. We're a bit greedy, frankly.

Right. So let's talk about mix of secured versus personal unsecured loans, consistent with what you said is like the TAM for secured is much, much larger. And so you've obviously made the assessment or said that you expect that secured products should become a larger share over time. Remind us kind of what that time frame could be or should be in your planning and how you're thinking about it? And what would have to happen in the auto and home funnel performance for them to overtake personal loans sooner than expected?

With respect to time frame, the honest answer is I don't really know. And the reason is because I think we will grow faster in these new categories. But they're trying to catch a course that's moving pretty fast -- and it's a bit hard to tell exactly. I mean I think we've telegraphed a couple of years of strong growth in the core. And the new products will outgrow it, but that's a big and fast-growing business on its own right.

So there's a lot of internal bets as to what we look like in 5 years. Has the core been overtaken? Is it auto? Is it home? I think everyone has a bit of a different take. So I don't think we have got...

Yes, everybody has a different take. I'm sure both externally, but it sounds like even internally.

Even internally, we all have our own favorites and our own -- I mean, we're obviously betting or hoping for all of the above, but everyone has different kind of view. And with respect to -- you said, what would have to happen...

In the funnel or a conversion, like help us think about like the differences between the different products, whether it be auto, home, or unsecured and what you see in terms of conversion rates, et cetera? And even reach.

Reach. Yes. I mean those are the two dimensions, right? One is -- one sort of variable is how quickly are we going to improve the models. And to what extent do we get some unexpected surprises, some unexpected progress. Some of the advancements we make benefit all the models and some of them are sector specific. And so there's different teams working against the different areas, and if they have some big advancement in one particular area, it will improve our conversion mechanically.

And then the other axis really is about distribution. And they all have a slightly different distribution strategy. Unsecured lending is obviously heavily D2C. It's heavily digital. Our main distribution strategy in auto is taking software to the dealerships. In HELOC, we're still early days. We've stood it up as a D2C business. I think there's a lot of opportunity as a white-label strategy, working with other depository institutions and the like.

So again, you can have big breakthroughs like -- just as an example, I think that we, for many -- a couple of years have been trying to figure out the right product market fit with the dealerships in terms of our software. Because once you get the software in there, you can surface the loan offers [ and ] the commercial flow. And I think we spend a lot of time trying to find like something that was very powerful and integrated, but it didn't take too much of a lift for retraining and like changing workflows.

And I think that -- and relatively recently, I think we've hit it. And so I think that, I mean, some of the growth you saw in Q4 in the auto space, that's just like we found the right product market fit with dealers and it's starting to fly off the shelves...

And by product market fit, you mean like the way that the software itself has put together, how it integrates into their own sales flow, et cetera.

How much the dealers are demanding it. Like you can -- like sometimes selling things is hard, and it's like knocking your head against the wall. Other times, they're calling you, "Hey, I heard about this, like can you give me a demo." And I think we're starting to get a great flywheel in that regard. So that's just like one example. You can get these distribution breakthroughs that sometimes create a big sort of unexpected leap forward.

No, for sure. So let's talk about unit economics. As secured becomes a larger share, what's the right way to think about the blended take rate and contribution margins? And I ask this question because inevitably, investors, right or wrong, are also trying to grapple with it and see if the take rates are suggesting some -- one thing or another. But what should we expect headline take rate compression to look like even if contribution dollars improve?

It's a good question. We don't exactly know. But part of the reason we don't exactly know is because the take rate itself within a product has its own life cycle. If you think about the advent of -- or history of our unsecured business, for the longest time, our take rates, I think, were probably in the average of 5% or something like that, okay? And what happens is you can think about the market benchmark in terms of how loans are being priced to consumers as being somewhat credit score centric.

And as our model get smarter and smarter and smarter relative to that benchmark, it's avoiding more and more and more. So if you think about that average price, there's half of the borrowers are riskier and half are less riskier and you get an average. As you avoid more and more of the riskier borrowers, your APRs start to come down. And as your APR start to come down, you create margin opportunity.

And so by 2024 compared to 2018, we were charging 10% to 12% take rates. And we could, and we were still the best offer in the market because of that margin we had created. And that margin did not exist back in 2018. So if we had tried to charge those rates, we would have gotten adversely selected.

So as your models get more mature, as they get better, you create economic opportunity. And that opportunity can go to the borrower in the form of lower rates or it can come to us in the form of higher takes or you could conceivably give it to the investor in the form of overperformance or some split of the three, right, which is typically what we do. And so that same thing is going to play out in the secured products.

So I think we said in the latest earnings, I think the average upfront take rate for secured products out of the gate is going to be something like 4%, but there will be a higher component of servicing economics. When you think about the analog, our servicing rates in unsecured lending are 1% per year. So as the loan amortizes, you get like a ratable revenue stream in addition to the upfront take. In secured products, we think that will be closer to 200 basis points, okay? But there will be a differential between home and auto. And I think both of those segments will evolve. As we get better models, we should create room for more economics.

And so it's a bit of a -- there's a lot of moving parts. In addition, our -- in our core business, our take rates went up to 10%, 12%, we signaled that they're starting to moderate down a little bit because at some point, it's not the best long-term business strategy to run an economic model where you're charging 10% take rates. Like there's some money you can sort of invest forward now in exchange for a much bigger business down the road. And so I think that one will moderate a little bit as we've telegraphed. I think these ones will improve over time and the relative mixes of the three, who knows.

Yes. So let's talk about competition in secured. The competitive set seems like it should change pretty materially in secured. Where do you see your primary advantage versus the incumbents? Is it better credit decisioning, a better conversion UX, better partner economics, time to close? Where do you think you have an advantage initially and which one of those facets or front, do you think will end up being most defensible?

Yes. I think it's different mixes of the same playbook. Almost every credit segment, you've got sort of a hyper-served prime segment and a very underserved, sometimes nonexistent riskier segment. And in the riskier segment, we aim to compete and sometimes create markets just by being good at underwriting.

Now as you get more secured, you probably get more of the sort of prime sort of well-served customer. And then there, you're competing on two things. One is automation. Like how much friction can you take out of the process. And the second is your -- the efficiency of your capital.

And so I think in each of these -- I think we're discovering that each of these areas, we've got opportunities for both. Like let's take HELOC as an example. It takes the industry 30 to 40 days to do a HELOC, we and maybe one or two other fintech players are down to 5. That's like a significant experience improvement. So for people who can access HELOCs already, we're not creating access for them, but it's like, well, you can wait 40 days at your bank branch or you can get this next Tuesday. And so that's a very powerful way of competing.

But by the way, the HELOC space, they're typically coming from the home buyer, like the mortgage space, and they will take loss rates up to 2%. There's not a lot of 4%, 5% loss HELOCs out there. That's viewed as very risky in the home space. But guess what, like we're working with unsecured buyers who are buying 10% loss loans and 5% looks pretty good to them. And so -- but in order to sort of create a new sort of market space in a world where the traditional market isn't touching the risk, you have to have a lot of credibility in your underwriting. And so I think both of those are opportunities, whether it's auto or home or whatever is next.

Why don't I take a breath, see if there are any questions in the audience here. Before we go to funding, et cetera.

So probably the biggest apprehension in the credit market right now, and you can see this even with you or others that they put up good results, be it the stocks or don't react in a way that maybe you would expect, seems to be tied to private credit and the state of capital availability. And there's been a lot of chatter of late on the health of the private credit space. There's an article on the front page of the Wall Street Journal talking about a big provider of private credit and kind of the things they're working through right now.

That being said, recent vintages and their returns have looked really attractive versus benchmarks for you. How should we expect the spreads to express themselves and to evolve as you skew more to the prime and secured? And what are your capital partners talking about that they're looking for from Upstart?

When you say the spreads, you mean like the pricing of the loans...

Yes, the pricing on the loans and the [ fees ] they're willing to pay.

I see. Yes. I mean I think from those guys' perspective, so you got two dynamics. On the one hand, each of these credit classes has some traditional buyer population. And then you get folks who I think are maybe, you could say, coming along with us for the ride, right? And so for that latter category, I think they kind of view all of this as some version of ROE, right? So a more secured credit class compared to unsecured will have more competitive APRs, lower APRs but they'll also have higher financing advance rates because they have lower losses. And it's sort of like levers up to the same mid-teens ROE or something like that.

So I think, like whether the underlying asset is risky or very secure, that will be sort of compensated for in the leverage ratios, and it will ultimately -- you should sort of have a comparable level of risk and return across those products because of how you use the financing.

I think that folks who -- they're sort of like folks who know Upstart very well, but maybe are getting their feet wet with new credit categories. And I think that what they want from us is consistency in performance and improvement. And then there's people we're engaging with who know credit classes very well, but are new to Upstart. And there, it's a lot of diligence, kicking the tires, making sure that we have the right incentives, the right sort of guardrails.

So for those capital partners, both existing and potential or newer, have you seen any reticence in their kind of behavior recently, et cetera, that would indicate like some apprehension that they're demonstrating around their own performance or access to credit and capital?

Yes. None at all, really. I mean the big annual tentpole Structured Finance Conference happens every year, it was actually last week. And I got to say that the tone was constructive and confident. But it's because the underlying credit is benign. The moment that turns, then there'll be discussions. But I think despite all the headlines out there and the talk of equity exposure, cross contamination, like everyone is like, I think these vehicles are very specific to consumer credit. And I think the underlying stuff, as you said, is performing very well, and everyone's happy. So everyone's like looking to deploy. And so that, yes, the tone is good right now, but the tone is always a reflection of credit performance and credit performance is in a good shape.

Yes. So let's talk stay on the funding topic just for a couple of more minutes. Let's talk about secured products and as those scale. Would you expect your ABS forward flow structures to become more central versus whole loan buyers? What kind of constraints, whether it be structure, ratings, data history, do you have to get past for that to matter?

No. The ABS sort of component of our strategy is very secondary in a sense, everything that we do is -- like the predominant objectives are to get the yield in the hands of either lenders, like depository lenders or institutional buyers. And then from there, if institutional buyers themselves want to securitize loans, we'll create -- we create the vehicle for them to do that. But there's very little sort of the prioritization of having us run a securitization shelf off our own balance sheet. It's really for -- it's like a feature of liquidity for the people who are buying the loans.

That said, we do intend to stand up liquidity like ABS shelves for these new products because it's good for the buying community, it creates all kinds of good hygiene in terms of liquidity and price discovery. And I don't know that there's any massive hurdles other than usually in those vehicles you need to prove yourself, right? You need some history. You need to work with the rating agencies.

It will probably be some version of what we've already gone through in unsecured, but hopefully a little bit short circuited because now these rating agencies know us. The first time we said, no, this is not going to perform how it looks like it will perform. They thought we were smoking something. And then like 8 years later, they're like, okay, they're now sort of taking our grades at face value. And so hopefully, that will sort of -- that will play itself out in auto and home lending as well.

So let's touch on balance sheet as I think you guys have gone through the process of educating people pretty well on, hey, as you're experimenting and entering into new markets, there may be some changes in balance sheet, even the amount that you carry initially may grow, et cetera. I think people are well conditioned and informed on that.

That being said, you've been reducing on balance sheet exposure and scaling third-party funding recently. What's the road map to get home lending and that component of maybe that -- what you will carry on balance sheet initially to look -- be auto-like and get to the third-party involvement there so that you can kind of normalize and stay within your real long-term objectives?

The -- I mean it's just doing what we're doing, that's sort of my day job, right? We've got to do the deals. We've got the first couple of HELOC deals in place, and we've got other people looking at it. And yes, it's just a question of cementing those partnerships. Some of them are heavy lifts because they're meant to be multiyear commits. And so you got to do a lot of diligence upfront. But once you lock it in, it's very resilient. So I think we're just maybe a quarter behind where we are [ in a lot of ] the other emerging products.

Okay. Okay. Got it. Okay. So pretty good. All right. So let's talk about back to kind of this medium-term target and the profitability is we've got this roughly 25% EBITDA margin target. On that path, how do you think about how much do you expect to come from operating leverage versus internal AI productivity versus improving contribution margins as products mature? And just help us think through like what the drivers of -- or the points of leverage on your profitability should be.

The majority will -- should and will come from operating leverage. Yes. I mean, I think we view ourselves as a business that can grow the top line pretty quickly with a relatively controlled expense base, a fixed expense base.

Now I think AI productivity will be a component of that, not the headline. But yes, like I think that we would expect to, on the balance to be able to do more growth with the same number of engineering bodies as they become more productive, so that will enhance operating leverage.

I think it will be less so about contribution margin. I think we want those to settle into some nice stable ranges and there should be like, as I said, sort of some incremental improvement over time. But really, I think the headline is the bottom line of this business should largely be a function of operating leverage as we scale.

Got it. So let's talk about your technology moat and operational AI. Through our conversation this afternoon, you've talked about some of the challenges of taking even the emerging AI technologies and scale and applying them to this industry because of data, accessing data, developing the data repositories, et cetera. And something that you guys have been doing for a long time now.

And you've mentioned that model updates incorporate broader outcome data beyond loans just originate on your platform. How are you validating that label and data quality and to avoid selection bias? And what's the impact on approvals for previously lost borrowers? How do you improve not only your data, but your ability to underwrite people maybe that you weren't able to originate for previously?

Yes. So I guess your first question is how do we sort of avoid bias in this expansion of our -- I mean in a sense, it's just -- it's more data available to our models. In a sense, the bias was bigger beforehand. So before, obviously, if you have a bunch of applicants and you give out a bunch of loans and you deny a bunch of people, you can only see the outcomes -- so you have this inherent bias, which is like you may be not approving some people and you'll never learn from that. And what you're referring to specifically is we figured out how to see the outcomes for the people that we decline who get loans elsewhere. And so we can learn from that.

So in a sense, all we've done is expand our sort of visibility set in a way that reduces the bias compared to when we just saw the selection bias of the ones we have approved. So I think it's just like net positive, and it's had a positive -- so the benefit of that, of course, is then you'll learn from those people in ways that you're like, "Oh, I sort of declined a bunch of these people. But look, they got loan somewhere else, and they did pretty well." So then that's additional model learning that your model otherwise would not have had and then that obviously contributes to higher approval rates going forward.

And with that incremental visibility, like just remind us where is that coming from? And what's the opportunity or potential to further expand that capture data capture?

It was a body of work that really involves matching the data we had on applicants that were declined to outcomes of theirs that we found in the credit files.

In the credit files, right.

So it was being able to link those in a large data set. I'm sure there's further opportunity to be better at this, right? Like ultimately, you want the model to see as many outcomes as possible, so you learn as much as possible. So we've got other ideas for how to run sort of R&D programs aimed at learning from declines, but that's a good example of one big outlook.

Yes, really fascinating. I mean, certainly, I think you can see that there's a lot of nervousness in certain elements of what may happen in the market, but it's certainly encouraging and reassuring to hear that things still seem quite solid, both on the consumer front and performance as well as even on the financing side of things. So I really appreciate all your insights here, Sanjay. Thanks for joining us.

Thanks for having us. Yes.

That was awesome.

It was a pleasure.

Good afternoon, and welcome to the Upstart Fourth Quarter and Full Year 2025 Earnings Call. [Operator Instructions] As a reminder, this conference call is being recorded.

I'd now like to turn the call over to Chelsea Williams Investor Relations. Chelsea, please go ahead.

Thank you. Welcome to the Upstart earnings call for the fourth quarter and full year 2025. With me on today's call are Dave Girouard, our Co-Founder and CEO; Paul Gu, our Co-Founder and CTO; and Sanjay Datta, our CFO.

During today's call, we will make forward-looking statements, which include statements about our outlook and business strategy. These statements are based on our expectations and beliefs as of today, which are subject to a variety of risks uncertainties and assumptions and should not be viewed as a guarantee of future performance.

Actual results may differ materially as a result of various risk factors that have been described in our SEC filings. We assume no obligation to update any forward-looking statements as a result of new information or future events, except as required by law. Our discussion will include non-GAAP financial measures, which are not a substitute for our GAAP results. Reconciliations of our historical GAAP to non-GAAP results can be found in our earnings materials, which are available on our IR website.

Before I turn the call over to Dave, please note that concurrent with today's earnings release, we also issued a press release regarding our leadership evolution. Management will address these plans during today's conference call.

With that, Dave, over to you.

Thanks, Chelsea. Good afternoon, everyone. Thank you for joining us. I want to start today with what most of you have heard by now. I'm super excited to announce that on May 1, my cofounder, Paul, will become Upstart's next CEO. While I'll continue on to my role as Upstart's Executive Chairman.

Why now? Well, in a few weeks, as hard as it is for me to believe, I'll celebrate my 60th birthday. And for more than 5 years, we've talked about this milestone as the right time frame for Paul to step into the CEO role at Upstart. I can't imagine a succession plan more thoroughly considered or executed with any finer level of detail than what we've done at Upstart in the last few years to prepare Paul and the company for this day. As I said in the press announcement, Paul and I have been building Upstart together, side by side, for the last 14 years, and I expect we'll continue to build it together for the next 14 or even longer.

When we cofounded the company, Paul was barely old enough to drink, and I was older than his parents. So this transition is an incredibly rare opportunity to have a founder-led company for decades to come. To be clear, I'm not going anywhere and expect to be quite involved with Upstart. Beyond my role as Upstart's Executive Chairman, I'll continue to help shape the company's strategic direction and important initiatives.

I want to thank all of you for this opportunity. Building and leading upstart has been the experience of a lifetime, second only to the life I've built with my amazing, my Tiffany and our 2 incredible children, Jackson and Tristan. In particular, I want to thank Upstart employees past and present. It's been an honor to work alongside you through ups and downs, whether a good day or otherwise, it's been an experience I will always cherish. I couldn't be more optimistic or excited about the next phase of Upstart.

Some time ago, Paul said to me that most of the trillions of dollars in interest paid by borrowers in this world is unnecessary in the age of AI. So our mission is a glorious one, and it's one requiring determination and perseverance not just for years, but for decades. I'm confident that in due time, under Paul's leadership, we'll look back with nostalgia at our first 14 years as the early days at Upstart.

Now to move on to our regularly scheduled update. I'm proud of all the Upstart team accomplished in 2025. We grew originations by 86% and revenues by 64% and while growing head count just 18%, a ratio of any business would die for, and we reestablished Upstart as a strongly profitable business. This growth in profitability was driven first and foremost by a significant market share gain in our personal loan product where Upstart grew originations 75% year-on-year.

But just as importantly, our newer products had an incredible year with originations for auto and home, both growing 5x year-on-year and originations for both actually accelerated in the fourth quarter. Even with this Epic growth across all 3 product categories, we reduced loans on our balance sheet by 20% in Q4 and expect this trend to continue.

With 70% of funding for auto and home loans originated in Q4 coming from 11 different partners and an additional 13 signed up for the coming year, we've officially begun a new era as the everything store for credit.

Looking forward to 2026, I feel like Upstart has emerged from a multiyear rebuild and begins the new year from a position of unprecedented strength. In addition to radically improving our AI in the last 3 years, we've also entirely rebuilt our capital supply for resiliency and competitiveness. And as we wrapped up 2025 powered by excellent credit performance and rapidly improving conversion rates in the last few years, we finally took the last steps to launch our home, auto and small dollar credit products into the funding markets.

Upstart is the leader in providing the best rates and the best process across products and to consumers across the credit spectrum. In 2026, we'll be about increasing our lead in establishing our brand in these dimensions. All of this sets us up for building incredible franchises in each of these product areas for years to come.

In the coming quarters, I expect AI-powered lending led by upstart to rapidly gain market share across these enormous unsecured home and auto segments as well as additional lending categories in the near future. We've been building the teams, the skills and the technologies to deploy AI in lending for more than a dozen years, and I believe we are peerless in this regard.

With this strong foundation and expansive TAM in place, I want to share today that we project Upstart to achieve a compound annual growth rate of 35% for the next 3 years. In addition to sharing longer-term guidance, we're also refreshing how we share near-term financial information with the markets. Starting today, we've begun to publish the monthly transaction volume on our platform at the beginning of each subsequent month, providing a fresher look at how originations are trending.

Beyond that, we'll also focus on providing annual guidance, which we expect to update as we report each quarterly earnings. The opportunity to radically transform access to credit with AI is unimaginably large, and we want to offer the world a court-sized seat as this future unfolds. With these upgrades to our disclosures and guidance, investors can better understand both the near-term dynamics and the long-term potential of Upstart.

With all that said, I want to say welcome to the new era of upstart. With that, I'll turn things over to Paul Gu, my Co-Founder and upstarts next Chief Executive Officer. Paul?

Thanks, Dave, and good afternoon, everyone. I've had the chance to get to know many of you over the last year, and I'm excited to spend some time today sharing how I see the business as we head into the next chapter. I'm incredibly grateful to have had the opportunity to learn from Dave these past 14 years. I got to work alongside Dave finding our very first customers, onboarding the first outside capital for loans evolving from one office to multiple to digital first, taking the company public, weathering the boom and bust of the COVID stimulus era and doing virtually everything else needed to build the Upstart of today.

Through that long journey, we built what I consider the very best team and technology and credit. Today, I have the great fortune to take the baton from that enviable starting point. I am also extremely excited to welcome Sanjay, Grant and Andrea to their new roles. Sanjay will serve as President and Chief Capital Officer; Andrea is joining us as CFO; and Grant is returning to upstart as CTO.

Sanjay and Grant have long been key leaders in Upstart's history, and Andrea is an incredibly talented finance leader with a background in complex novel business models. I look forward to partnering with them to show the world the potential of AI-powered credit. As Dave said, Upstart is a far stronger company today than it was 1 year ago or for that matter at any point in our history. We always judge ourselves by the trifecta of growth profits and credit performance and 2025 saw tremendous progress on all 3. Looking specifically at the Q4 results, we grew total loan origination dollars 52% year-over-year.

That is function of finding strong product market fit in our new products, but it's also a testament to the persistent and large opportunity for us to grow with and win share in our personal loan business, where origination dollars grew 41% year-on-year in its 12th year of operation.

Because of our strong unit economics and operating leverage, that growth in revenue flipped our net income from negative $2.8 million in Q4 last year to positive $19 million this past quarter. We expect revenue growth to continue outpacing expense growth and make 2026 an even more profitable year.

I want to pause and touch on our contribution margins. You'll notice that these have declined over the past year. This is expected and intentional due to 2 dynamics, both of which we're actually excited about. First, we have grown rapidly in secured products and prime borrowers, segments which come with lower take rates but dramatically larger market sizes. Second, we increasingly have the financial ability to consider the long-term value of acquiring a customer and the reputation value of saving them more money than anyone else. Together, these are putting us on a path to become the best and most trusted place for every American to fulfill all of their credit needs.

Setting aside a couple of transitional and accounting factors specific to Q4, we expect to be able to accomplish this while still growing absolute contribution profits nicely with the business, albeit with a modest lag to revenue. The growth in 2025 also came alongside another year of exceptional credit performance.

The average return of our last 12 quarterly vintages of loans exceeds that of U.S. treasuries by 608 basis points with every individual vintage exceeding treasuries by at least 270 basis points. This combination of scale, consistency and returns is only made possible by our use of AI lending, and we believe will be a growing draw for capital partners, banks and credit unions.

We achieved this combination of record revenue and consistent credit performance with the Upstart Macro Index in the 1.4 for most of the past year and quarter. By comparison, during our prior peak revenue year of 2021, UMI averaged around 0.8, meaning that statistically identical loans were 43% less likely to default back then as they are today, that our 2025 results surpassed 2021, in spite of that UMI backdrop, is a strong testament to the power of our AI first strategy. Improving separation and automation quarter after quarter is a recipe for delivering results over a multiyear window, regardless of how macro moves the short run.

Now I want to quickly highlight a few of those technology advances from Q4. First, we launched 2 major model updates, Model 24 and Model 25. Model 25 added a significant number of consumers who never took loans from the upstart ecosystem to our training data set by joining past applicant data to similar loans taken outside upstart. This adds a rich diversity to our training data set and better positions our model to understand consumers whom we historically did not approve or win.

These models increased our measured separation accuracy advantage over our benchmark textbook model by about 1 percentage point to 172.2% on previously observed upstart loans, but had an accuracy boost more than 100x as large on those loans we did not previously win. Second, we redesigned our partnership models and integrated new data sources into our partner APIs together driving 24% more channel originations quarter-on-quarter, while lowering latency by 34%.

Finally, we launched a brand-new architecture of our verification models that lowers default rates by 0.8% and launched our first voice LLM for when manual verification calls are required. 2025 was also a year our new products found product market fit. In our auto business, we doubled the number of live lending rooftops for the third quarter in a row and began applying AI for the first time to our auto secured personal loan product.

In our dealerships, we launched the ability for car buyers to sign their loan contracts remotely and begin automating document verification processes with LLM, thereby reducing funding times by 12%. As a result, our combined auto originations in Q4 were up 56% sequentially and 340% year-on-year.

In our home business, we tripled the percentage of HELOCs using automated underwriting and rapidly scaled our campaigns to cross-sell the product to homeowners in our existing member base by integrating property data. As a result, home originations in Q4 were up 70% sequentially, totaling 350% year-on-year.

You may have noticed that I called out technology wins in nearly every area of the business I discussed. This isn't because I'm partial to technology by background, though I am, but rather because technology is the story of how we're building a differentiated business for the long term at Upstart.

In Q4, the scale of our training data crossed 100 million borrower repayment events for the first time, 14 years after we started the company. The next $100 million repayment events will come dramatically faster. And include data on almost every major category of consumer credit.

We believe that proprietary training data, when combined with the ever better learning algorithms our team develops drives a long-run growth trend that persists independent of any ups and downs of the macroeconomy or competitive environment.

We have executed against the strategy and observed its results for more than a decade. And in looking at our technology road map now have every reason to expect that compounding to continue for many years to come.

With that, I'll turn things over to Sanjay. Sanjay?

Thanks, Paul, and good afternoon to all for listening in with us unless you are [ a Sox fan ]. Looking back on what we believed and messaged almost exactly 1 year ago today, I am proud to say that Upstart called its shot in 2025. Last February, we guided to $920 million of fee revenue and $1 billion in total revenue on the year.

We finished the year with $950 million in fee revenue and $1.04 billion in total revenue. We are aiming for positive net income, an 18% adjusted EBITDA margin. We ended at positive $54 million in net income and a 22% adjusted EBITDA margin. Notwithstanding some of the inevitable twists and turns over the course of the year, 2025 landed right on schedule.

Overall for the year, we grew our total revenue by 64%, and we held our fixed expense base to a mere 5% growth, a nice flex of our operating leverage, which allowed us to more than 20x our adjusted EBITDA from $11 million to $230 million, despite concurrently investing in the scale up of multiple new early-stage products.

We took meaningful steps to reduce balance sheet exposure in Q4, while simultaneously graduating 3 new product categories from their R&D phase and almost quadrupling their volume year-on-year. We added new funding sources for our growing product portfolio in Q4 and converted several of our existing funding partnerships into multiproduct relationships.

Notably, for auto lending across both refinance and retail, the most mature of our new bets. 92% of our Q4 originations were funded via third parties, supported by the continued expansion of our institutional and lender funding relationships.

In home lending, we have similarly signed our initial institutional and lender partnerships who have collectively taken a majority of our Q4 HELOC production and expect to attain third-party funding levels similar to that in our auto segment in the near future.

Taken together, these developments have allowed us to accelerate new product growth rates in Q4 while simultaneously reducing our balance sheet holdings by 20% quarter-over-quarter. All in all, we are entering 2026 feeling enthusiastic about the momentum of our financial performance as well as the strength of our capital supply. With this as context, here are some of the financial highlights from Q4 of 2025. Total revenue for Q4 came in at roughly $296 million, up 35% year-on-year and 7% sequentially. This overall number included revenue from fees of approximately $265 million which was up 33% year-on-year and above quarterly guidance, reflecting the impact of recent underwriting model launches.

Within fee revenues, our servicing revenue stream continued its steady growth clip at 28% year-over-year, driven by higher origination volumes and improving servicing fair value marks. Net interest income of approximately $31 million, ahead of guidance by $5 million, resulted from continuing strong return performance on a loan balance that remained elevated during part of the quarter.

As discussed, we made progress reducing balance sheet exposure by quarter end and would expect the contribution of net interest income to moderate as those efforts continue. The volume of loan transactions across our platform is approximately $456,000, up 86% from the prior year and 6% sequentially, representing approximately 307,000 new borrowers. Transaction growth was driven by continued model improvements and continued growth across our new products. Average loan size of approximately $7,000 was 5% higher than the prior quarter, reflecting an increasing mix of nonpersonal loan products, which generally carry larger loan sizes.

Our contribution margin, a non-GAAP metric, which we define as revenue from fees, minus variable costs for borrower acquisition, verification and servicing as a percentage of revenue from fees came in at 53% in Q4, down 4 percentage points from the prior quarter and in line with guidance.

The sequential decline primarily reflects the increased weighting of lifetime value in our pricing calculations, which results in lower take rates but higher future profits. In total, GAAP operating expenses were around $277 million in Q4, up 9% sequentially from Q3.

Expenses that are considered variable relating to borrower acquisition verification and servicing, were up 11% sequentially relative to the 12% increase in the volume of loan transactions. Fixed expenses were up 7% quarter-on-quarter, reflecting continued investment to support the growth of the business.

Q4 net income was approximately positive $19 million, ahead of expectations and reflecting outperformance on the top line, alongside continued discipline across our cost structure. This result builds on our return to GAAP profitability in the second quarter of 2025 and reflects continued progress in the current credit environment. GAAP earnings per share was $0.17 and based on a diluted weighted average share count of 112 million.

Adjusted EBITDA was roughly $64 million, in line with expectations. We completed the full year with total revenue above $1 billion, up 64% from 2024, a contribution margin of 56% and positive adjusted EBITDA of $230 million, representing a 22% adjusted EBITDA margin versus a margin of 2% a year earlier.

We ended Q4 with approximately $985 million of loans held directly on our balance sheet, down from $1.2 billion in Q3. The sequential reduction reflects loan transaction activity during the quarter alongside continued progress transitioning funding for new products over third-party capital.

As we set up for the year ahead, we are evolving our approach to financial guidance by emphasizing the annual outlook and by focusing on the longer-term trends of the business. To supplement this, as David has mentioned, you will also be able to view more frequent updates on origination volumes that will be published each month at upstart.com/volume.

This year, we expect to continue growing our core personal loan business at a healthy clip via consistent model improvements and growth wins. As we increasingly weigh the future benefits of customer lifetime value in our loan pricing, we expect that take rates will moderate reducing current contribution margins in return for higher borrower volumes and lifetime profits.

Notwithstanding this dynamic, we expect absolute contribution dollars from our platform to grow at a robust rate which we are aiming to maintain to within at least 5 percentage points of corresponding fee revenue growth. We also expect to continue the rapid scale-up of our new secured product categories in auto and home which taken together, we expect will contribute over $100 million in fee revenue in 2026.

As they reach mature scale and sell-through, we anticipate this category will attain average upfront take rates of around 4% and in addition to average servicing rates of around 2% of outstanding balance on average loan sizes of approximately $30,000. Within those averages, home lending will tend towards higher upfront take rates, while auto loans will exhibit a higher proportion of ratable servicing revenues over the lifetime of the loans.

Across both unsecured and secured lending, we aim to continue streamlining our balance sheet. And building new committed multiproduct relationships with lenders and investors alike. And finally, we expect to continue maintaining tight fixed cost discipline and improving operating leverage as we further grow our profit base. On the macro environment, as per our tradition, we assume a constant default risk environment with the UMI holding study at its current value of around 1.4 to 1.5 and a static interest rate environment as well.

More specifically, for 2026 in the aforementioned scenario, we are anticipating total revenues of approximately $1.4 billion, revenue from fees of approximately $1.3 billion, and an adjusted EBITDA margin of approximately 21%. Looking even further out, our ambition will be to maintain a relatively steady growth clip over the coming several years, paired with improving margins and operating leverage. For the 3-year period spending 2025 to 2028, in particular, we are targeting to maintain a 35% compound annual growth rate in a macro-neutral environment, with a terminal adjusted EBITDA margin of around 25% in 2028.

We 2025 was a year of tremendous effort and perseverance across the company, and I want to thank the teams at Upstart for their continued dedication to the cause. On a personal note, I'm excited and grateful for the opportunity to be a part of this leadership transition and to contribute to the next phase of Upstart's journey.

I'm also very excited to welcome Andrea to the finance chair starting next month, where she will unquestionably be raising the bar. And on that note, I would like to now pass things back to the operator to kick off the Q&A. Operator?

[Operator Instructions] We'll pause for a moment to allow everyone an opportunity to signal for questions. We'll take our first question from Dan Dolev with Mizuho Securities.

Outstanding quarter like really, really strong congrats. I have a quick question and then a quick follow-up. So growth was really, really strong. in autos and HELOCs in Q4. So how did you manage to reduce the balance sheet loans? And then I have a quick follow-up on that one.

Dan, this is Dave. Yes, thank you. Yes, we were happy. We've been telling the market for a while that the balance sheet loans for the purpose of incubating these newer products and making sure the models were right and beginning to line up partners.

And we lined up quite a few of them. I think we said 70% of these products were funded by third parties and off our balance sheet. So we've kind of achieved lift off on that now. And when you sort of put all the numbers together, that meant not only really fast growth in both of those product areas. But at the same time, the amount of loans on the balance sheet dropped and we expect that trend to continue.

So it's just a matter of doing what we've told the market we were doing. And even that acceleration we saw in the fourth quarter in both HELOC and in auto, despite that, we were still able to actually reduce the loans in the balance sheet.

So I think that's a good signal that we're doing what we said. These products are now mature in the market. We have another dozen or so investors or lenders that are going to participate in those parts of our business in 2026. So we're in a great position with them right now.

That's amazing. And then a quick follow-up on the UMI guys. Like how much conservatism is built into your UMI guidance? Like can you maybe give us like a rule of thumb, your might gets better than your underwriting should improve dramatically. Something like that, I think, would be helpful.

Yes. Great question. This is Paul speaking. Our guidance always assumes a constant UMI. So we wouldn't necessarily say it's either conservative or aggressive. It's just really our very best estimate of what the UMI is And the UMI itself is as a reminder to everyone. It is our measure of the background consumer risk when you hold constant, all of the variation in loans, all of the variation in consumers.

And it gives us what we think is the fastest and most precise read on the background level of consumer risk in the economy.

We'll go next to Kyle Peterson with Needham.

Great. I want to start out on the guide. I noticed you guys guided for -- it looks like EBITDA margins to be down a bit in '26. So I want to see, is this a function of like mix as the balance sheet has been run off a little bit, which is obviously like a positive trade-off? Or are there any like notable investments or something you guys are making in the upcoming year? Just any color there on the EBITDA margins would be very helpful.

Yes. Thanks for the question, Kyle. Yes. So on the topic of the margins that we guided to, we guided to a contribution -- or sorry, to an adjusted EBITDA margin, but of course, baked in there is an implicit set of assumptions about how contribution margin will move. And in my earlier remarks, I described some changes that we've been expecting and planning for on contribution margins.

That's a result of a few different factors, all of those intentional and all of those, we think, actually long term, very healthy for the business. There's basically 2 different things in there. One is a changing mix where we're expanding into the largest markets for consumer credit. These are some of the secured products and these products the loans are much bigger. The profile of the product is different and you tend to have lower take rates in these products, but the TAM that you get access to are just enormous many, many times larger than in just pure unsecured or pure personal loans. So a very good thing for us to have great winning products in those categories.

And the second is that, frankly, I think relative to just the last couple of years, we've gained the ability to be able to take a longer-term view of our customer, the value of acquiring customers, the value of building up our brand and our reputation for being the very best place for consumers to shop for credit.

And we want to lean into that. We want to press our advantages that we've got in the market in terms of having best rates and best process and really just sort of cement our reputation for having that best credit product for customers, and we think that is going to ultimately be winning in the long term.

Now in the short term, that both of those things, the change in mix and taking those longer-term considerations into the mix does, of course, have some impact on the margin. But we think ultimately, that's going to be a very good thing for the business.

Got it. That's really helpful color. So I appreciate that. I guess as a follow-up, I wanted to touch a little bit on private credit and some of the capital partnerships. In the last few months, there's been some spookier headlines about the health of some of these providers, whether it's due to credit exposure on either the consumer side or software exposure or something like that.

But I want to see -- it seems like the pipeline in -- for funding is -- remains really strong there. So I guess is I guess that correct? Is there any change in health or appetite for the consumer credit in the paper that you guys are underwriting? Just any color on how those conversations are going would be great.

Kyle, it's Sanjay here. A couple of thoughts. So first of all, I would say we're very deliberate about who we engage with in the private credit world and I think that we can probably limit our observations to the ones we work directly with. But for those partners that we're working with, I would say that these are counterparties that do deep diligence on both the credit and on us. I think they do their job really well, and I think they take the credit as seriously as we do.

And I think our best reflection of how all of that credit is performing is in how we are assessing our own co-investments in those relationships and as you could see in the earnings materials, they are performing better and better.

And so I think that's probably just a reflection of the kinds of partnerships we've made and the kinds of co-investments we've made together, but everything is, I think, in very good shape with respect to the direct partnerships we have.

Next, we'll go to Simon Clinch with Rothschild & Company Redburn.

I'm kind of curious just your thoughts around the general environment for the consumer, particularly as it potentially of what your models are seeing. In the third quarter last year, you talked about how there are some sort of slightly sort of uncertain signals that sort of materialize in the model and the model reacted. There's been a lot of volatility in the fourth quarter.

There's been some sort of some news that maybe there's been some constraints put by some of your peers perhaps? And I was just wondering if your model saw anything if you react in any way or any sort of signals that are work sort of notify.

Yes, thanks. The health of the consumer is good. We in the UMI values return to what I call post-COVID normalization, which is sort of a long-term multiyear stable to down trend that has been really playing out since the consumer really took a hit back in 2022.

And it is true back in Q3, we had detected a real ultimately temporary increase in the background level of consumer risk. We think back in Q3, our models responded to that appropriately, but that has since faded. And I think we're back to the broader trend that we've been in for the last couple of years, which is which is a bit of normalization to where consumer health was before COVID and kind of stimulus to consumers to an unusual place.

But in short, I think there is always a lot of noise out there about macro and volatility, but if it was showing up in any way in borrower behavior, we picked that up first in UMI. The really good news is that UMI has been showing really good and solid patterns for the consumer lately.

That's great. And just as a follow-up, with the -- particularly with the likes of auto and now moving into home, when we think about the long-term market share opportunity in businesses like this, where it's much larger loan sizes, much larger TAM. I know Dave, you talked about, I think, in the past, aspiring to be like double-digit share in auto.

And I was just wondering, has anything changed in the environment that means that those kinds of targets are perhaps not achievable or are they very much in your size?

I think they're very much in sight. I mean 5x growth in a year, certainly auto and HELOC as categories grew a tiny, tiny fraction of that, meaning we gained rapid market share in 2025. I would say generally that look, our view is that AI-enabled lending as a specialty is going to be the dominant market share in all of these categories, probably a decade from now.

And so maybe the question is, if you're an investor or analyst trying to understand this, is everybody just going to develop this? Or are there going to be real leaders in developing sufficiently different AI models that they become giants. And that's what we intend to do.

We're -- we don't think there's anybody else building the sophistication of models that we are. And we think the economics make it inevitable that an AI-centric platform will be used to originate the majority of loans in these categories over time.

So for us, it's not a question of whether AI will dominate in these categories, it certainly will. The question for us, and for everybody, of course, is what fraction of that will belong to Upstart in our platform, and we're, of course, quite bullish on that.

We'll next go to Peter Christiansen with Citi.

First congrats to Paul. Great to see you transition here. Also congrats to Dave and Sanjay and the team and looking forward to meeting Andrea. And certainly, I want to applaud the added transparency with the monthly disclosure should go a long way towards reducing some volatility in the stock.

My question is your perspective generally on the personal loan market. I -- there's some who opine that maybe there was a bit of saturation, a lot of growth last year. Just curious where -- how you see that perhaps accelerating or normalizing in '26? And what do you think that means for Upstart and their selectivity? And then I have a quick follow-up.

Thanks, Pete. Good to hear from you too. On personal loans and the sort of competitive market there, I think I would say I don't think we necessarily build our business around any assumptions about how the whole market is growing or not growing. But I do think personal loans are extremely useful product. They've generally grown nicely.

But for us, it's all about winning market share. We think we're growing very nicely on our market share. Our product is competing extremely well better than ever on both the basis of rates and process. And we increasingly -- when we do market intelligence, it suggests that, that level of competitiveness is getting better and better over time.

And we're really just in a position, I think, to press and cement that advantage, and that comes from the differentiated technology that we've built. We're going to keep investing in that, widen that lead. And ultimately, that's going to translate into an even bigger lead on best rates and best process for the consumer, and we think that's going to transfer to a bigger and bigger market share in a category that is really useful and important for a lot of people.

That's helpful. And then as you think about the first half of this year, it's not a Big Beautiful Bill, tax refund, no tax on tips, those sorts of things. Likely benefit the consumer shorter term and likely improves credit performance, at least the seasonality aspect of it, do you think that potentially may put a crimp in demand for term demand for personal loans?

Yes. Good question. Generally, borrowers and therefore, we will benefit from looser economic policy that helps consumers. While you're right that when people have more money, there may be marginally lower demand for credit, the much bigger factor in our business, the reason we ultimately end up really aligned with our consumers on their level of financial well-being is that the dominant factor in our business is our ability to approve people for competitive affordable rates.

And that is much easier when people are in a healthy financial spot. So whether it's through interest rate cuts or tax bill changes or other policy changes that make life more affordable for the Media American, and those tend to flow through and be really beneficial to our volumes and business.

And we'll next go to Mihir Bhatia with Bank of America.

Good afternoon. On a second, the congratulations to the team, especially, Paul, also taking the comments on added monthly disclosure. I think the market will appreciate those.

Now in terms of my question, the first question I had was I just wanted to double click in on a statement you made about -- I think you mentioned an increased weighting in lifetime value, which drives lower take rates.

Can you just expand on that? Is really what's happening there are you considering repeat loans there? Or is that more of a product mix comment? Just trying to understand that and what that means as we think about 2026 and into the medium-term guide.

Yes. Thanks, Mihir. Yes. So that comment, I think, both Sanjay and I made reference to that. And it really was referencing the decline in contribution margins over the past year. And the really like -- the important thing to take away is that, that decline is expected, it's intentional, and it is the result of a few different things. One of those is a changing mix into these much larger product categories, but another one is the one you flag, which is the increased consideration of longer-term value coming from both customers and our business.

And really, you can just think of that as it's a little bit like, we think Costco is like a much more valuable and bigger business than Whole Foods. I think we could certainly charge like Whole Foods because we just have such differentiated products to offer the consumer. But we don't want to take all of that surplus even when we can. Now obviously, if we need to, I think we could like in some years past, but we think we can ultimately build a much bigger business, getting more of the value to the customer and just winning a much bigger share of the market being the very best product, saving people and amount of money that's really share worthy. And so we're starting to do that.

Got it. And then as my follow-up, I wanted to ask about the partners you're signing up for some of the newer secured products. firstly, kudos of getting the balance sheet smaller, I think that's a good sign.

But like just maybe just expand on the partners you're signing up for the secured products. Are they different? Are they -- is this more banks? Are they less yield, I guess, focused or is it like pretty similar partners as you have on the personal loan side?

Mihir, this is Sanjay. Yes. I would -- I guess the short answer is it's all of the above. In all of these markets, there's a traditional super prime segment, and there is relative to the existing markets, riskier segments.

We've got depository institutions like credit unions and banks that are now working with us. And in the institutional world, we have a mix of existing partners who expanded their scope with us. And then some new partners as well who are maybe more specialized or focused in these specific secured segments. So as usual, we are building a very broad resilient supply chain that's designed to be essentially the broadest risk aperture in the segment.

And we'll next go to John Hecht with Jefferies.

Congratulations to the team for the executive evolution. First is just a quick modeling question. I mean, I assume the tax rate now, you've been profitable for a few quarters. What -- is there anything you could tell us about your expectations for the tax rate next year?

John, I would expect that our operating loss carryforwards will continue to be in force for 2026.

Okay. And then you guys talked about ABS markets and forward flow. And I think some of the bank and credit unions have been consistent buyers over time, too. But maybe can you give us an update on that channel? And any kind of expectations for changes there?

John, are you referring specifically to the ABS market or the bank channel?

Well, I'm interested in any changes to the bank and saving -- excuse me, and credit union channels and their appetite for purchasing loans.

I see. I think that the existing set of partners we work with in study buyers, and I would say, a sort of steadily increased appetite. As we scale up in these newer secured categories, I think they attract the interest of maybe a larger size institution. Certainly, the bigger size of bank than has been interested in unsecured lending. And so we're starting to develop those relationships in a very interesting way.

All right. And then you had pretty good growth in the core product both in the -- well, you had a good growth in the core product and the super prime product. Any -- and I know you talked about the average loan size, but anything you could talk about in terms of duration or other characteristics that we should be thinking about?

Let's see, duration in our core business has been pretty steady. A mix of 5- and 3-year terms. Obviously, the secured products have a slightly longer duration. So as they become a larger fraction of the mix, they will pull the duration up a little bit. But I think that they're still at a scale for such that their effect is not particularly meaningful.

And we'll go next to Reggie Smith with JPMorgan.

And Paul congrats on the promotion. And Dave, I had no idea you were 60. You're like Benjamin Buttons, aging backwards. But congrats on the move up as well. My question, I guess, you guys have approved like more than 2 million loans over the last 2 years.

And I'm curious like how you guys would rate or think about your engagement with those borrowers during the repayment period, specifically with an eye towards selling additional loans and products. And I ask because I look at your app, and it seems like there are probably some low-hanging fruit from a feature perspective that could probably increase engagement. But I don't want to, I guess, kind of mistake that. So I'm curious how you guys feel about that.

And you also mentioned LTV during your prepared remarks, is there any stats you can share around recent trends there because I'm curious what the repeat usage is or what's giving you confidence that the LTV is kind of bending and the right direction?

Well, Reggie, you've got a good eye and we do have a lot of job openings if you're looking. So I mean you're absolutely right. There is a lot of low-hanging fruit there. We are relatively early in our journey to really building on the engagement of our customers once they've taken out their first loan.

We do have a lot of people that take out second loans in the same product or second loans in another product. But it's really a relatively nascent program that we've built and we don't even have it fully rolled out. We have more limited mobile cross-sell capabilities today. So there is a lot of low-hanging fruit, and we actually think this is going to be a very significant source of upside in the next few years is really getting that whole ecosystem working.

We are starting to invest in it very seriously, both on the tech side, but also as we talked about in the contribution margin discussion on how we just think about the value of acquiring a new first-time customer. We don't have any metrics on this to share yet. But certainly, as this becomes more important, that's something that we'll consider.

Got it. Okay. No, that makes sense. If I could add one more. I guess I'm looking at the mix of loans this last year. And obviously, you've grown your mix of super prime. I see R&D auto HELOC all increasing in ratio, but they haven't increased a whole lot and the platform referral fee take rate has come down a lot this year.

And so I'm trying to think about -- it sounds like these are going to be bigger pieces of the book, trying to get a sense for where this take rate can kind of go because it would seem like it's not really in the run rate just yet, but my numbers could be. Any color you can provide there would be great.

Yes, Reggie, there's a few dynamics going on. First of all, the newer businesses grew really quickly from a small basis. But the core business -- the core personal loan business growing very quickly itself. So the percentages there didn't like move as much as one would imagine when you get that kind of growth. But again, it's somewhat of a rich problem to have that all the products are growing really quickly. I would say, generally, in the personal loan product, also the super prime segment, you can think of that as 720-plus is the fastest-growing part of that.

And when you put that all together, the percentage take rate would expect to be trending down. But at the same time, both super prime personal loans as well as these new categories tend to have larger loan sizes. The dollars that you're -- that we're getting from each aren't necessarily small, in fact, may be larger. So that's kind of the way to think about it.

The percent take rate in some ways, may become less useful as a metric because these other products can be much, much larger, particularly HELOC, for example, and have a smaller take rate, but actually, the dollar margin on them is quite large. So anyway, that's -- it's sort of a bunch of things moving at the same time.

Got it. If I could sneak one more in real quick. Does your guidance, 3-year guidance contemplate any new products? And I'm thinking something like a smaller duration high-velocity loan, like I see some of your peers doing these kind of payday loans, but I guess duration 6-week loans. Are there any other products that aren't out yet that we should be thinking about as feeding this guidance?

Yes, Reggie, I mean, I think we've got a lot of exciting ideas. But with respect to financial guidance, I think you can assume that, that will be largely earned off the back of our existing portfolio, including the new growing products.

Next, we'll go to James Faucette with Morgan Stanley.

I wanted to dig in a little bit on kind of the -- some of the profitability or how we think about contribution, et cetera. And when we -- particularly as you push more into auto and home and when you think about sales and marketing, how should we be anticipating things like target unit economics and borrow acquisition costs for those loans and products versus your traditional personal loans?

James, let's see. In terms of acquisition costs, a bit of a, I think, a unique story to each product. Auto, in particular, as you know, is not a D2C product outside of refinance. It's one where we have enterprise software being embedded at the point of sale in the dealership. And so there's a go-to-market cost there, but once you're working with the dealership then per loan acquisition costs can look very low.

And HELOC as well as a market where there's a big fraction of the market that is direct-to-consumer. And I think we would target a similar acquisition sort of margins per loan as we do in personal loans. Obviously, those loans are a lot bigger. So of course, a bit more dollar cost per loan.

And then there's a large fraction of the market that is done again through maybe B2B type points of sale, either through contractors or in branch at existing institutions. And so there is a bit of a mix there in terms of what is D2C and what is sort of B2B, if you will.

Got it. Got it. And before I forget, and as I suppose my second question, congrats to everybody a lot of exciting moves for everybody personally, it sounds like.

But on my second question, turning back to HELOC. Just curious are bars, are you getting any cross-sell from the existing funnel or from -- or is that from distinct acquisition channels? And then how do we think about the opportunity to cross-sell and how that may be built into your medium-term outlook and profitability targets?

Yes. We are absolutely doing cross-sell for HELOC. It's one of the most important strategies for acquiring HELOC today. And it's because a lot of people that naturally show up in our personal loans marketing have homes, and HELOC is just an incredibly efficient way to refinance other debt they have or get financing for a project they want to do. And historically, the reason more people didn't want to do HELOC as the process was so arduous. And as we increasingly automate that and make that our application to close process faster and faster.

Today, we're at about 6 days versus an industry average of that process gets a lot closer to a personal loan like process and with the much better rates. I mean it's just an incredibly attractive product for many overlapping customers. So absolutely, we do a lot of cross sell.

Our next question comes from David Scharf with Citizens Capital Markets.

Reggie stole my Benjamin Button's observation, but congrats as well. to Dave and Paul and the whole crew. I wanted to take a step back and maybe just ask you for a little more color or the thought process on the guide? And specifically, maybe 2 questions. First is, as you noted that there was a kind of pretty substantial 20% reduction sequentially in the on-balance sheet exposure.

As you look out 3 years, it's $2.5 billion of revenue, roughly, quite a bit of scale. Do you have a particular target in mind or how much balance sheet exposure you'd like to see when the business is at that level?

Yes, it's a good question. I don't think we sort of target a level of exposure. I think we think of it more as appropriately using our balance sheet as a tool to propel the business forward. And that, as we have said in the past, it primarily means incubating new product areas and proving out new parts of the model that we need to do kind of internally.

And a little hard to project it would sort of depend on what new products were in the market with at those times. But just to be clear, we don't aim to be a predominant balance sheet lender of any sort. We don't aim to have NIM, net interest margin, as anything dominant in our financials. So we are using it as a tool. Again, our expectation, and if you look at our guidance for 2026, you can see it's almost entirely dominated by fee-based revenues, and that's how we see our business going forward. But we don't try to make specific predictions on it.

We don't view a $0 balance sheet or 0 loans on the balance sheet as an ideal future state, it might, in fact, indicate we're not doing enough R&D or we're not innovating fast enough. So we always intend to use the balance sheet in that way.

And maybe just to add on to what was saying, David, maybe better said, our strategy, I would articulate on the capital side as being we want to build the largest and most resilient supply chain of third-party capital to power these businesses as we can.

And as you know, our personal loan business, which is very mature, has historically been well into the 90% plus range in terms of our ability to fund it with third-party capital, and we want to get these new businesses to the same place.

And I just want to say, I mean, one of the outcomes of incubating and doing it properly with our own capital is we have 100% retention of private capital partners. So they're depending on us to deliver a quality product that performs as it's supposed to. Part of that is, again, eating our own dog food, if you will, for a period of time.

And that has resulted in, as I said, 100% retention and I think a long list of other private capital or private credit partners that we can potentially work with in the future.

Got it. Understood. No, that's very helpful. And maybe just kind of one more follow-up. As we -- you think about the margin guidance for 2028, it's a notable roughly 400 basis point expansion from the 2026 level. Just trying to get a sense for how to think about -- I hate to use the word mature, you're not going to want anybody to use the word mature.

But nevertheless, somewhat mature advanced margins if there's a ceiling. I know the company had a few quarters above 30% back in the 2021 late 2020 [ date ]. Do you view 25% is an informal kind of steady state level of profitability at scale? Or should we just think of that as nothing more than a 3-year medium-term piece of guidance?

Yes, it's a great question. The short answer to all the questions is just operating leverage. We expect to grow fixed costs by a fair bit less than we're growing the top line and also a fair bit less than we're growing contribution profits.

Now we've answered a few questions about changes in the margin profile on the contribution side because of mix and getting into secured loans and lifetime value and all these things. And so that change within that 400 basis point move to that 25% EBITDA margin is really a whole bunch of different things that are happening.

But ultimately, it's like -- we are building a much bigger business at the top of the funnel. That business is being built in categories that have slightly lower take rates and therefore, a lower contribution margins. But that will still significantly outgrow our increase in fixed expenses. That's partially just because naturally, we don't need we don't need to linearly grow all the fixed functions as we grow the business. There's just natural operating leverage in the business. But it's also frankly because we're pretty bullish about just how much productivity lift.

We're already getting and will continue to get from adoption of AI internally and that we expect to continue to happen and give even more give us even more operating leverage. So I definitely don't expect 25% to be some kind of terminal state. I think if you were to break those numbers apart, there's a lot of room for to keep going higher.

And at the limit, I think it can be approaching what contribution margins look like. And -- but just keep in mind that those won't be quite as high as they were, say, back in 2021, where the only thing we were really optimizing for was really like honing in on those profit levels as we're facing that macro pressure and top of funnel volumes were declining.

We'll take our next question from Kyle Joseph with Stephens.

Good afternoon, and yes, echo congrats to everyone. Most of my questions have been answered, but just kind of want to get your take, obviously, AI, at least in the last few weeks has been really disruptive to the market, whether it was software originally or data services or wealth management today.

But just kind of want to get your updated take on the competitive environment and how you see the proliferation of AI impacting things? Obviously, I consider you guys to have an advantage there. But any updated thoughts.

Yes. Thanks for the question. We've been doing AI and lending long before really AI was even a term or people were talking about this in other contexts and categories.

So I think we'd be really happy to see more players come into the AI lending world. I think it's still pretty thin ranks over here. And I think it's just good for the industry, it's good for the consumer, and we'd certainly welcome it. I think ultimately, if and when that happens, it's going to be a technology velocity question, and that's the kind of competition we like to be in, which is one where we think we've got a 12-year head start across data, across algorithms across infrastructure. But we're really still very, very early in the application of AI to this business and the industry.

And so we expect our technology to get much better at a rapid pace. We talked a little about the amount of data that we have collected and how fast that's growing. We just cleared 100 million training data points over really the first 14 years of the company's life, but the next page is going to come that much faster. That unlocks all of these proprietary algorithms that we've been building for all these years and lets us build more complex models that just can't run on smaller amount data.

And then wrapped around that is this infrastructure layer where as you get these more powerful models with much larger amounts of data, you just need the ability to deal with more memory problems, more latency problems, more cost problems, and you have to build a bunch of engineering infrastructure around that. And so it's definitely the sort of problem with a natural technology flywheel in there. And we think we've got both the best starting position and the fastest velocity. And -- so I think it's -- as more people get into this, we think that's going to be good for the consumer, and we think we're going to still be in a very good position to compete.

We'll go next to Rob Wildhack with Autonomous Research.

And congrats on all the new roles. On 2026 and the outlook there. You're exiting '25 at like a $300 million quarterly revenue run rate. You've got 1Q that's kind of a seasonal low say that's another $300 million.

That would mean you have to average like $370 million in quarterly revenue for the rest of the year. And the high watermark for the company ever was, I think, $310 million back in '22 per quarter. So I'm just curious how you see this all coming together in like the revenue and I suppose volumes to ramping and accelerating throughout the year.

Rob, it's Sanjay here. I think of your back of the envelope math is roughly right. I've checked it specifically, but yes, I think you can assume that Q1 will be one with a bit of seasonal headwind and then from there, we'll continue our growth clip.

And our growth story, hopefully, is familiar by now, but I think we've got a pipeline of exciting model wins and growth wins lined up to continue improving our funnels and allowing us to grow our campaign sizes at constant CAC, which will power our core business, allow us to continue taking share.

And then, of course, we've got these new products that, as Dave said, have been sort of released into the wild now we -- I think we made in hindsight fortunate decision to continue investing in them. And now you're seeing the fruits of that labor.

And I think they'll really begin to contribute in 2026 as well. So I think the combination of all of those things has us and our internal model is feeling pretty good about that.

Okay. And then a similar question on '28. Can you connect the 35% revenue CAGR to volume growth? I know you're not going to give like a '28 volume estimate, but just contextually, like how much of the 35% revenue growth do you see coming from maybe personal loans versus growth in the other products like home and auto?

Well, yes, we're obviously not sharing a more specific mix breakdown nor frankly do I think we could honestly tell you the exact mix in 2028. Now obviously, our new products are growing very fast. They're growing off smaller bases, but they are growing much faster than personal loans. So they're going to be a much larger share than they are today just because they're huge, they're huge markets.

And ultimately, and I can't tell you if ultimately it's going to be in '28 or a few years after that. But I do think, ultimately, those products are going to be bigger for us in personal loans just because the addressable markets are that much bigger. And so it's sort of a question of getting there from here. And I think we're going to keep growing very fast all the years between now and then. And I think it's going to be a race between all of the categories, HELOC auto and unsecured and we're going to be doing cool stuff in all 3. The tech is going to get better in all 3.

And the product is going to be just very, very competitive for the user on both best rate and best process. So I think it's anyone's race to win.

And for our last question, we'll go to [ Arvind Ramani ] with Truist.

I did want to follow up on the question on AI. And we've talked about this before. upstart has been working on AI for a very long time now. But the question I had is, with some of the broader advancements in AI, there is Upstart like sort of like able to sort of leapfrog what -- what are some of the advancements -- broader tech advancements have enabled you to kind of leapfrog some of the kind of AI kind of programs you all had internally And are there any segments where you are feeling a little bit more exposed just because like something that was special that you've developed is not more broadly available?

Thanks for the question. We're really big fans of all the progress in AI. It's a huge benefit for our business in a few different ways. We don't feel threatened remotely by it. We actually feel extremely grateful for it and I'll kind of explain why. So I would say just from the perspective of do we feel threatened by it.

I mean the simple answer is just that a lot of the advances in AI are really good for work that humans are naturally good at. And I think that's what makes it really exciting. So what makes it really useful for so many different kinds of applications. But unfortunately, humans have never really been very good at precisely underwriting loans and figuring out the cash flows they're going to produce for the next 5 years. And that's something that has always been solved as a big math problem, better solve this big math problem. And so it's not something that you can just do because you have a language model that replicates human behavior, well, again, not something that humans are naturally good at.

In order to solve this problem well, you need to have some pretty specific kinds of data that connect lots of information about people at the time they want loans historically with what happens later in the lives of those loans.

And then you need very specific kinds of algorithms to connect those two things in a way that is going to maximize your signal extraction and that's what we've been working on for 12 years. And now turning to like why we're so grateful for all the progress. Well, one reason is actually that a lot of the advances in basic infrastructure or some of the basic types of algorithms that we then innovate and build on actually coming from all of this progress in AI.

When you read about how many billions of dollars are going into funding all the R&D there, some of what comes out of there are just better kind of base algorithms or better kinds of infrastructure that then we get to use. Of course, you think about chips and compute and memory and all of the progress there, like we benefit from all of that. But then at the actual application layer like the types of models you actually need are just totally different. The types of data you need is totally different. So we sort of get the benefits, but we aren't really aren't really threatened.

Now we do also get to use -- directly benefit from and use a lot of the AI technology that's now coming out we get to use that in new ways, ways that are different than what I thought -- think of as these big math problems that traditionally, we were really good at solving.

We now get to do things like if you think about something like HELOC, and how you automate more and more of HELOC, will naturally because so much of the process of securing and perfecting a lean, checking like the property records like a lot of that stuff is mess in a human way and traditionally comes with very high operations cost because you have a lot of people that are checking to make sure things are right.

Those are actually perfect problem to throw sort of LLM style AI against. And so we're doing a lot of work like that now. And that's actually going to help us get the rates of automation much higher in the business than would have been possible, say, 5 years ago. And we absolutely are excited to be at sort of forefront of applying AI in those ways. Of course, also, that's relevant to how we operate and how we make ourselves productive as a team. really relevant to the question of ultimately how much operating leverage the business gets. I think in all those places, we're super, super optimistic about AI.

But ultimately back to the question of like how much does it threat. And what we've built over the last 12 years, I would say it's a huge boost, not at all a threat.

Great. And just -- that's really helpful. Just last quick follow-up on that. In terms of like kind of the AI apps and tools, right, I mean, you'll kind of use open eye, anthropic, lovable cursor, like who are your big I partners?

I mean we are happy to use we use all these models from all of the foundation model providers in the different kinds of uses that I described. But when it comes to the models that we use in our core lending business, those are all proprietary to us. They're ones that we've been building for the last dozen years or so. And those are not things that -- they're not directly built on top of anybody else's technology. And that's really because when I say that we benefit from some of the base algorithm innovations. It's like there are -- there will be a paper about a new kind of neural network, and we'll certainly ingest that we'll try out that new type of network.

But actually to make it work as well as our models do on loans, you have to make a whole bunch of modifications to it to make it care about the right kinds of things naturally like you don't want it to be naturally interested, for example, in getting something right only in a 1 or 0, which a lot of these models are naturally set up to do.

You wanted to really care about the whole series of cash flows, the timing of cash flows, like exactly how many dollars in sense. And -- so you have to change a lot of the internal math of these things to make them really optimal for our problem. And so those are all proprietary algorithms that we've built over many years.

Okay. And so that means the message I can take away is that you all have -- your tech stack is kind of so unique to you all that even if there's some advancements out there, no one can just like kind of just strip together the solution. This needs to be stitched together by hand. Like this is not like easy to replicate.

100%. Again, it's really important to just remember that the LLM models coming from Anthropic or OpenAI or any of the others, Gemini. They are really good at solving problems that humans are good at solving and they can do it at scale. They can work 24/7. You can spin up 100 of them in parallel and have them work.

But no matter how many humans you have, you don't want that army of humans underwriting loans for you.

I'd now like to turn the call back over to Dave Girouard for any final or closing remarks.

All right. As we wrap up, I just want to thank each of you for your time and attention over the past few years. As I said earlier, my time as the CEO of upstart is something that I will always cherish. I couldn't be more bullish about the company's future under Paul's leadership.

We are really just getting started and now more than ever, those who stay will be champions. Thank you for joining us today.

This does conclude today's call. We thank you for your participation. You may now disconnect.

"

"

"

"

"

" Mizuho Securities USA LLC, Research Division

" Needham & Company, LLC, Research Division

" Citigroup Inc. Exchange Research

" Rothschild & Co Redburn, Research Division

" Piper Sandler & Co., Research Division

" BofA Securities, Research Division

" JPMorgan Chase & Co, Research Division

" Morgan Stanley, Research Division

" Autonomous Research US LP

" Jefferies LLC, Research Division

Good afternoon, and welcome to the Upstart Third Quarter 2025 Earnings Call.

[Operator Instructions]

As a reminder, this conference call is being recorded. I would now like to turn the call over to Sonya Banerjee, Head of Investor Relations. Sonya, please go ahead.

Thank you. Welcome to the Upstart Earnings Call for the third quarter of 2025.

With me on today's call are Dave Girouard, our Co-Founder and CEO; Paul Gu, our Co-Founder and CTO; and Sanjay Datta, our CFO. During today's call, we will make forward-looking statements, which include statements about our outlook and business strategy.

These statements are based on our expectations and beliefs as of today, which are subject to a variety of risks, uncertainties, and assumptions and should not be viewed as a guarantee of future performance.

Actual results may differ materially as a result of various risk factors that have been described in our SEC filings. We assume no obligation to update any forward-looking statements as a result of new information or future events, except as required by law.

Our discussion will include non-GAAP financial measures, which are not a substitute for our GAAP results. Reconciliations of our historical GAAP to non-GAAP results can be found in our earnings materials, which are available on our IR website.

And with that, Dave, over to you.

Thanks, Sonya. Good afternoon, everyone, and thank you for joining us. To kick things off, I'll share my perspective on our business.

Upstart today is a dramatically stronger company than it was just a few years ago. Our technology, our business, and our teams have never been better.

The opportunity for AI and credit is unimaginably large, and there's no one better positioned than Upstart to lead this $1 trillion industry to this exciting and inevitable direction.

Now turning to Q3.

Upstart continued to execute on its 2025 game plan of rapid growth, profits, and AI leadership, all under the auspices of exceptional credit performance and precise macro handling.

In addition to 80% year-on-year growth in transaction volume and 71% revenue growth, we were nicely profitable once again. In fact, Q3 GAAP net income grew by a factor of 6 over the prior quarter.

Consumer demand for Upstart continued to grow rapidly with more than 2 million applications submitted in Q3, up over 30% from Q2 and reaching the highest level in more than 3 years. Despite this awesome demand, transaction volume on our platform was less than we anticipated.

Our risk models responded to macroeconomic signals they observed by moderately reducing approvals and increasing interest rates. This drove a reduction in our conversion rate from 23.9% in Q2 to 20.6% in Q3.

If you follow the Upstart Macro Index, you would have seen that this macro indicator ticked up modestly in July and August, which is essentially what our model responded to over the course of the quarter. We believe this to be nothing more than a speed bump with UMI reverting to lower numbers since.

To be clear, we see no material deterioration in consumer credit strength. And in fact, we've seen recent signs of improvement. You can and should expect that our models will always do their best to price prevailing risk appropriately.

Precise and rapid tuning to changing economic conditions is a foundational capability of Upstart AI, and we're confident this precision is winning hearts and minds for Upstart in the credit market right now.

The system is behaving exactly as it was designed. At a minimum, our Q3 results should give you confidence that we don't sacrifice credit performance to achieve transaction volume targets.

Turning to our newer products, which include small-dollar loans, auto, and home. These offerings continue to improve and mature, accounting for almost 12% of originations and 22% of new borrowers in Q3.

Transaction volume for auto, home, and small-dollar each grew in the range of 300% year-on-year.

Our auto retail business, in particular, has really begun to accelerate. We more than doubled the number of live lending rooftops on Upstart in Q3 compared to the prior quarter.

Transaction volume for auto retail also grew more than 70% sequentially. We expanded to 4 new states in Q3 and made some significant improvements to our software. This is really a breakout business for us.

Additionally, we've been quietly working on a hybrid product called an auto secured personal loan that's beginning to gain traction.

As it relates to our home business, beyond continued process innovation, our unique partnerships with banks and credit unions mean we offer the best rates to the primest borrowers compared to other fintechs by as much as 300 basis points.

Best rates and best processes are what we're all about. Our continued process and automation breakthroughs in our secured products, meaning home and auto, give us confidence that they will be real growth drivers for Upstart in 2026.

Finally, with respect to funding on the Upstart platform, we're in an exceptionally strong position in our core business with significant excess capacity.

On the bank and credit union side, we added 7 new partners, our best quarter for new logos this year, and we reached a new all-time high in monthly available funding from these partners in Q3.

On the capital market side, we continue to have exceptionally strong execution with our institutional partners. Having signed our first agreement in 2023, we now have 10 active partners.

In August, we renewed one of our largest partners for the second time. And importantly, Upstart has 100% retention of all private credit partners to date.

We believe that we have the industry's best AI for responding rapidly and precisely to changes in the environment, and this is a central reason why our partners have confidence in us.

In September, we also issued a securitization with strong demand, leading to significant oversubscription of all classes despite upsizing and tightening of spreads.

This ABS deal involved 30 investors, including 7 first-timers, demonstrating the strength of Upstart's reputation in the market. We've also continued to make progress securing third-party funding to support our newer products.

We've signed 17 partner agreements this year, including 9 signed in Q3 alone, and expect to ramp these partners into production this quarter and next.

All in all, we're all systems go to finish the year strong and get ready for what we think will be an amazing 2026 for Upstart.

To wrap things up, we're making rapid progress as the leader in AI-powered credit. The somewhat complicated macro economy we all see today is, in my view, the perfect opportunity to demonstrate the strength of our AI platform, and we're doing just that.

While legacy financial services execs continue to ponder the use of AI and credit, the Upstart platform has now generated more than $50 billion in AI-powered loans since inception.

Unlike other AI platforms, we generate our own training data with more than 98 million borrower repayment events to date, with about 105,000 more repayments due each day, driving improved separation and model accuracy.

This is enabling us to build quickly toward a future of always-on credit, where every American is persistently and precisely underwritten, providing them with the best rate anywhere, 24/7 credit access right from their phone, with little to no process.

That is a proposition and a future that we are betting on 100%. With respect to the investor community, I feel more than ever that those who stay will be champions.

With that, I'll turn things over to Paul Gu, my Co-Founder and Upstart's Chief Technology Officer. Paul?

Thanks, Dave. I'll start by addressing the model conservatism we experienced in parts of Q3. Over the past few years, one of our biggest advances has been our model's ability to respond with speed and precision to changes in macro conditions.

This progress stems from a suite of techniques that are proprietary and critical to resilience through a diversity of economic environments. A few months ago, that led the model to tighten on credit while certain risk signals were elevated before recently normalizing.

That model behavior partially reflects irreducible volatility in the outside world, but is also a function of our model design and sampling variance, both of which continue to improve.

Since the start of Q3, the improvements we've made to our calibration methodology are expected to cut unwanted month-to-month volatility in model calibration-driven conversion changes by about 50%.

As we continue to innovate on our model calibration techniques, we'll increasingly be able to minimize conversion volatility in the business while delivering on target credit performance.

Beyond calibration, Q3 was a productive foundation-building quarter with a number of technology improvements that will power our next phase of growth.

I'll start with personal loans. First, we took another leap forward in our evergreen engineering quest to lower latency in pricing loans. By parallelizing another major portion of loan pricing, we reduced end-to-end latency by as much as 30% and are now rolling it out platform-wide.

Reduced latency unlocks the ability to build larger and more complex models as well as make use of the ever-growing data set of 98 million repayment events that Dave mentioned.

Next, we launched a true machine learning model to optimize take rates. It is our intention to capture value in relation to the value that we create for our borrowers.

We expect that this framework, over time, will unlock a significant improvement in our ability to monetize model wins that benefit borrowers who are already vested for as well as increase our competitiveness in new segments where we're still establishing our edge.

In the domain of customer acquisition, our ability to utilize digital partnership channels that relied on cloud environments or APIs to target offers was historically limited by the complexity of our underwriting models.

This quarter, we built a programming language agnostic framework for data transformation that makes it much faster to translate our models to work with any partner ecosystem.

We also worked with a key partner to enable larger model sizes in their cloud environment, allowing more of Upstart's unique underwriting algorithms to be used in targeting.

On our direct marketing channels, we developed a proprietary technique to target marketing spend based on causal impact to conversion. Compared to our prior, more textbook technique, early results show a 50% uplift in incremental originations from the same level of spend.

Advanced AI and underwriting ultimately need equally advanced AI and acquisition, and the successes this quarter were a big step towards that.

I'll wrap up my remarks with a few technological highlights that are driving growth in newer products. We've made rapid progress automating the process of getting a HELOC.

When we launched instant property valuations back in June, we automatically approved less than 1% of home loans. Since then, automatic home loan approvals have grown to 10% in September and about 20% in October.

While we'd love to simply automate away almost all the documents like we have in personal loans, the world of home loans is just less digital, less standardized, and there are more requirements.

So for the next leg of improving the HELOC funnel, we've begun using multimodal AI to do the work of human document reviewers in real time. Our rapid pace of process improvements makes me optimistic that we're on a path to an industry-leading home equity product.

Additionally, our small-dollar relief loans continue to make rapid progress. In September, we launched Instant Funding for the first time. Most borrowers who qualify for instant funding see funds in their bank account within around 90 seconds of approval.

While small installment loans at bank-friendly APRs are a wonderful innovation, you should expect to see a lot more from Upstart in this area in the coming months.

Thanks to the team's work this quarter, I'm more excited than ever about our upcoming pipeline of technology wins. With that, I'll turn it over to Sanjay. Sanjay?

Thanks, Paul, and thanks to all of our participants for sharing some of your time with us today. I'll now spend a bit of time reviewing our Q3 numbers.

At a headline level, we were pleased to finish the quarter with healthy annual and sequential revenue growth as well as extend our run back to profitability.

Within that, our transaction revenue this past quarter was marginally short of expectations as our models expressed some temporary conservatism in piloting the current environmental dynamics, but this was largely offset by growth in interest income from the strong return performance of our balance sheet.

Margins and take rates have remained steady, and credit performance continues to land right on target.

We are carrying a larger-than-normal loan balance on our books as we work towards closing a number of deals across all of our new product areas, which will both reduce R&D carrying balances and flow new volume directly to our lenders and investors.

We remain pleased with the progress of those various conversations and expect to have tangible outcomes on this front by the end of the year.

More broadly, third-party capital in our core unsecured lending segment remains readily accessible, handily outstripping our borrower supply, and is currently not in any way an impediment to growth.

Spreads on our third-party capital continue to compress, partially a result of the competitive funding environment and partially as an expression of investor confidence in the steadfast performance of our credit.

With respect to borrower approvability, our model has exhibited some recent caution in response to a UMI run-up of almost 0.2 points that happened over the course of the past quarter before more recently subsiding as well as to a rising trend in repayment speeds, which is generally an encouraging longer-term signal for credit, but in the near term, limits interest income from current loans and requires higher coupons to compensate.

In all of this, we, as always, care, first and foremost, about getting credit performance right, which will always result in the best long-term outcome for our business.

We have an inherent belief that AI models are better suited to navigating a complex and changing environment than human intuition, and we have demonstrated the discipline to heat them even when they express a bias toward moderation as now.

If the currently observed higher repayment speeds and easing consumption growth are indeed indicators of imminent credit improvement, these could represent the long-anticipated tailwinds that could accelerate growth prospects heading into next year.

In the meantime, we continue to be guided by the North Star of prudence in the underwriting of risk on behalf of our lenders and investors. With this as context, here are some of the financial highlights from Q3 of 2025.

Total revenue for Q3 came in at roughly $277 million, up 71% year-on-year and 8% sequentially.

This overall number included revenue from fees of approximately $259 million, which was up 54% year-on-year, but short of our internal expectations by roughly 6%, mainly for the model-related reasons previously mentioned.

Within fee revenues, our servicing revenue stream continued its steady growth clip at a 10% sequential rate.

Much of the shortfall in expected fees was counterbalanced by higher-than-expected net interest income of approximately $19 million, resulting from continuing strong return performance on a loan balance that remains temporarily elevated.

To reiterate, we are aiming to enter into a phase of reducing our R&D-related balance sheet holdings, which we anticipate will gain steam in Q4 and continue into 2026, and we would expect this revenue item to moderate as we are successful.

The volume of loan transactions across our platform was approximately 428,000, up 128% from the prior year and 15% sequentially, and representing approximately 300,000 new borrowers.

The average loan size of approximately $6,670 was 12% lower than the prior quarter from a combination of borrowers requesting lower loan amounts, a model exercising increased caution in improving loan sizes, and a mix shift towards smaller loan products and risk rights.

Our contribution margin, a non-GAAP metric, which we define as revenue from fees minus variable costs for borrower acquisition, verification, and servicing as a percentage of revenue from fees came in at 57% in Q3, down approximately 1 percentage point from the prior quarter and versus guidance as lower conversion rates created some mild upward pressure on both acquisition and onboarding unit costs.

In total, GAAP operating expenses were around $253 million in Q3, roughly flat to Q2.

Expenses that are considered variable relating to borrower acquisition, verification, and servicing were up 11% sequentially relative to the 15% increase in volume of loan transactions.

Fixed expenses were actually down 7% quarter-on-quarter, largely due to a reduction in compensation-related accruals.

Q3 GAAP net income was approximately positive $32 million, well ahead of expectations and reflecting outperformance on net interest income, reduced fixed costs, and a $7.2 million gain on our convertible debt repurchase.

GAAP earnings per share were $0.23 based on a diluted weighted average share count of 110 million.

Adjusted EBITDA was roughly $71 million, also corresponding ahead of expectations.

Adjusted earnings per share were $0.52 based on a diluted weighted average share count of 125 million.

We ended Q3 with approximately $1.2 billion of loans held directly on our balance sheet, up from just over $1 billion in Q2.

As shared last quarter, we have multiple new products simultaneously exiting R&D status and entering the scale-up phase. And our business development efforts this past quarter have been aimed at putting in place the third-party capital arrangements that will enable us to shift away from balance sheet funding on these emerging products and release back our invested capital.

We are very pleased with the progress of these efforts and believe that we are on a path to putting multiple agreements in place across all of these new product lines, which will set them up to further scale in 2026.

Exact deal timing is, of course, not perfectly predictable, and it is important for us to do the right deals with the right partners. So we will take the necessary time to ensure we are well set up on this front for next year.

In the meantime, returns from our balance sheet holdings continue to be strong, delivering healthy spreads above market base rates, as can be seen in the data on Page 23 of our earnings presentation.

As we look to Q4, the broader economic backdrop for credit remains favorable in our estimation. Decelerating personal consumption growth is a signal of improving credit health, if perhaps counterintuitively so.

Against this, we perceive a labor market that has remained at full employment since lockdown, meaning there are as many open jobs as job seekers in the economy, as well as a muted impact of the recent tariff policies on inflation and a gradual easing of the monetary climate.

In this scenario, we once again assume a stable UMI as well as holiday seasonality typical of Q4, which tends to serve as a mild headwind.

We expect the impact of any further rate cuts this year to both improve consumer financial health and lower investor return requirements. But at this stage, any such effects would not be felt until the new year.

In this environment, we will continue to produce model and targeting accuracy gains as well as automation wins to grow our top line. Our net interest income will start to benefit from the returns on our committed capital investments that were made in prior years.

Now that our P&L is once again back to profitability, we will plan to begin dialing up our forward investment into customer lifetime value by slightly moderating take rates in exchange for higher origination volumes and higher repeat transactions in the future.

And as usual, we will expect to continue our fixed expense discipline in how we manage the cost side of our business. With this context, for Q4 of 2025, we are expecting total revenues of approximately $288 million, consisting of revenue from fees of approximately $262 million and total net interest income of approximately $26 million.

Contribution margin of approximately 53%, GAAP net income of approximately $17 million, adjusted net income of approximately $52 million, adjusted EBITDA of approximately $63 million, with a basic weighted average share count of approximately 98 million shares and a diluted weighted average share count of approximately 111 million shares.

For the full year of 2025, we now expect total revenues of approximately $1.035 billion, consisting of revenue from fees of approximately $946 million and net interest income of approximately $89 million.

Adjusted EBITDA margin of approximately 22%, and we expect GAAP net income of approximately $50 million.

Before we move to Q&A, I will take the opportunity to thank all of the various teams across Upstart for their hard work and continuing dedication to our mission. And with that, operator, over to you.

[Operator Instructions]

Our first question comes from Dan Dolev with Mizuho.

Just wanted to ask a quick question on the application demand. It seems very strong quarter-over-quarter.

Maybe, Sanjay, if you can comment on the strong demand in the third quarter, and then maybe just like tie it all into the guidance, which was a little bit below what we were expecting and below the guidance in 2Q. So, how do you square these 2 things together?

Dan, this is Dave. Yes, as we said in the remarks, we grew applications about 30% quarter-on-quarter, which was ahead of the origination, the transaction volume quarter-on-quarter.

And really, a lot of things came together in terms of just marketing programs and cross-selling and all these things. So the application growth is certainly great to see.

I think what it really highlights is that our model took a step towards conservatism during the third quarter, just based on seeing macro factors.

And I think that is just a natural thing we might expect. As we said, it's since reverted, but it was a period of time where it saw signals, and it was moving quickly.

I think maybe overreacting. I think in some sense, having a model that overreacts is better than having ones that underreact because it did revert.

But I think it is useful to point out that the application volume was quite strong, our strongest in 3 years, and grew quite a lot. And I think that's a very healthy statement for the business, even if it didn't in Q3 transfer to as much volume as we expected.

[Operator Instructions]

We'll go to our next question from Kyle Peterson with Needham.

I wanted to ask specifically in auto, obviously, there's been some high-profile bankruptcies and kind of negative credit events in the space.

Have any of the headlines or news impacted your expansion plans or how conversations with customers are going? Or I guess just how has the recent news and events impacted how you guys are viewing things and progressing in auto right now?

Yes. None of that has had a direct impact on us for sure. We have not seen that type of couple of examples that were out there of fraud zone activity. So, I don't think it's anything that we would describe as widespread.

It's not something from our perspective that is widespread. I think when you have examples like that, it does create a little bit of caution in the market. So banks or others providing senior financing probably do a bit more diligence, et cetera.

But I don't think there's any wholesale change in the market, but that is the nature of it. A couple of the larger banks got bitten on that particular auto lender.

But we've been pretty rigorous about building processes to make sure we're effectively underwriting the dealership themselves and mitigating risks against dealer activity that's not what we want. So, this is an area that I think we're handling well. We have not seen any major issues.

But again, I think whenever you read headlines, it does cause a little caution in terms of increasing amounts of diligence or questions that need to be asked, et cetera, but that's part of the course

I guess as a follow-up, I wanted to specifically ask about what you guys are seeing in the super prime segment. I guess, looking at the originations, it was down a little bit sequentially.

So, I guess, was that where the model tightness that you guys called out? Did you see a little more in that 720-plus FICO score versus the core product? Or is there more competition there? I guess, just like what are you guys seeing?

And is any of that concentrated more in the super prime? Just trying to think how we should square that with some of the positive commentary on demand and funding capacity from your bank partners.

Kyle, this is Sanjay. I think it's a combination of things. I mean, the thing you pointed out is definitely a factor, meaning our models reacted generally to some macro signals as Dave described.

I think that was true of the primary segments as well. In fact, if you look at the segmentation of our UMI, the subprime consumer is actually at a relatively low UMI, probably somewhere around 1.2, 1.3. And if you start to look into the segments in the low to mid-700s, it's quite a bit higher.

So, there was definitely a model impact.

I think it's fair to say that it's also a very competitive segment, and we see other growth numbers in that segment, and they're healthy. So there's a price impact or an aspect of competition as well.

[Operator Instructions]

We'll take our next question from Pete Christiansen with Citi.

I want to follow up on some of the earlier questions. As it relates to the improvements that you made with your marketing channels, which sounds pretty exciting and is obviously illustrated by the higher number of applications.

Is there a way to at least get your sense for the quality of these leads? I know the AI system was a bit conservative this quarter. So, taking that into account, do you think that the quality of applications has remained the same or maybe improved or what have you with these new capabilities?

Peter, this is Paul. Yes. So I spoke in my prepared remarks nice wins we had in applying AI to customer acquisition.

And the way you can think about those wins is ultimately, at the point of customer acquisition, we are somewhat indifferent between selecting for people who have a high propensity to apply and people who have a high propensity to convert or be approved.

Ultimately, it's the product of those 2 things that we're solving for, of course. And so the improvements we make can help, one, both, or ultimately just the product of those things.

So obviously, we did have a larger increase in applications relative to where the final originations count ended up.

So I think mechanically, you can infer from that change in the likelihood to convert through the funnel. Of course, the conversion rates are lower.

Now that's in large part, as we already said, because we were knowingly making a choice with our model to be a little bit more conservative on the credit side in earlier parts of the quarter.

So relative to that model, of course, we did end up marketing to people who were a little less likely to be approved or a little less likely to convert, but that's not necessarily a chosen strategy.

Then my second question, non-prime auto has had elevated delinquencies even before some of the more noticeable news events that have been happening in the space for a couple of months now.

Dave, I'm just curious, if we were to see an improvement in that specific category, would that be a needle mover for Upstart's auto originations?

Yes. We've seen very good credit performance in auto. So we do feel good that our models are working that become calibrated.

To the extent others are having issues or what have you, maybe some are withdrawing from the market; those can be good things.

Maybe it suggests a transition or an inflection point in the market. So for us, it is just really important. We get calibration, we get more separation. We bring partners on. We keep refining the processes.

And I think it's going really well on all fronts there. So I think in 2026, we do feel very optimistic that the auto business as a whole is going to be a contributor.

Again, a little disruption or a little noise in the market when you're new like us to it, can be a good thing. It means there's an opportunity when things are shifting.

We'll move to our next question from Simon Clinch with Rothschild & Company, Redburn.

I wanted to just jump back to the first question, really, about the application volume growth that you saw.

And just, Sanjay, if you could just remind us what you said about what's implied in the fourth quarter? Because it sounds like you're assuming that the conservatism in the model is going to continue in the fourth quarter, despite your comments around the UMI actually starting to show some signs of improvement. Is that correct?

Simon, I guess I would note that the improvements in UMIs are materializing. As usual, we are conservative and want to watch them bake, and we're already past the month of October.

So some amount of Q4 was impacted by that UMI rise as well; even though it is now subsiding, we will, of course, follow it with some lag. So, I think that what we described is the model impact in Q3, even though it appears to be abating, will impact Q4 as well.

Just as a follow-up, then, when we look at the broad demand for personal loan growth, I mean, the kind of view I've had through most of this year, and I think is consensus view is that there's a lot of demand for just refinancing credit card debt.

Is that still very much the case that's really driving that personal loan demand? Or are we seeing that demand broaden out into other drivers?

I think refinancing debt really continues to be the dominant use case for personal loans. But it is very much the duct tape of credit. It's useful for so many things. And so there's a very long tail of ways that people use personal loans.

I think in some cases, because the process is so much simpler and the rates can be quite competitive, that it does compete at some places with secured loans, whether that would be to buy a used car off of a website or what have you, places where you might otherwise, or home improvement, where you don't want to get a HELOC.

So I think an unsecured loan, if it's fast, easy, and the rate is competitive, will always be very, very broadly useful to the consumer.

We'll go next to Patrick Moley with Piper Sandler.

I just have one on the balance sheet expansion you saw in the quarter. Just wondering how conversations with some of the potential funding partners of the R&D products have trended recently?

And then you touched earlier on the auto, some of the credit issues we've seen recently in auto, and how that's impacted the consumer.

But has there been any contraction in demand from any of your private credit partners there? And then I understand that they're waiting to see how the portfolios season in some of those R&D products.

Is there anything you can share with us there on how they're feeling about that and how those conversations have gone?

Patrick, this is Sanjay. Yes, as we said in the remarks, we're very pleased with the direction of all of those conversations. We're obviously having them across a number of different new product areas right now. I think appetite is good.

These are large deals, a multiyear time frame, and a large check size. So there's a lot of diligence in these conversations and these processes, and they are not perfectly predictable in terms of timeline. But with respect to progress, I think it's all going well. We're excited about all of it.

On the auto side, in particular, I don't think there are any concerns about credit, to be honest, at least in the loans that we're producing, I think the credit performance is pretty clear.

As Dave mentioned, there's some broader noise about fraud in the space, and I do believe that has probably lengthened timelines in terms of these processes. Everyone's diligence lists have sort of doubled and tripled in size. And so that's sort of a component.

With respect to timing, I don't think it's really changed the motivation or the appetite at all with respect to the specific conversations we're having.

I do believe we now have enough seasoning in our portfolio for people to look at our loan cates and get a really good sense for calibration and for how the credit is performing. So it's really just deal processes, legal processes, getting in place financing, bank relationships, et cetera.

So they're heavy lifts, but I think we're very happy with how they're going. We're very excited about the partners we're talking with. And as we said, we hope to have some tangible outcomes for you guys to digest pretty soon.

And our next question comes from Mihir Bhatia with Bank of America.

I wanted to start with the conversion rate. You talked a little bit about the conversion rate being impacted by higher UMI. Is that the primary factor? Are there other factors that maybe we aren't seeing or you aren't seeing from the outside inside the models that's driving it?

And then just on conversion rate, Paul, I think, mentioned --touched on limiting variability in the metric going forward. Can you talk about that some more? And if there's a particular level that should stabilize that? Like is this low 20s percent the right level?

Yes. Dave, I'll cover the first half of the question, and then Paul can answer the second. No, really, the conservatism in the model is, from our point of view, pretty much the dominant driver of the change in conversion rate.

So it comes in the form of a small fraction, fewer people approved, the rates they're approved at being a little bit higher, which means just marginally less likely to take that load, and then sometimes the approved loan size is a little smaller.

So that is the basics of a slightly more conservative twist in the model. So again, we don't believe this is anything sustainable. And we do think that we'll get to a model that's a little less responsive, honestly, and maybe overresponsive in this particular case.

But no, there's no other factor going on, as you saw the application volume is quite strong.

Then, on the second part of the question, about the reduction in volatility around conversion rates, and specifically around the macro calibration contribution to conversion rates. So, a few things to understand about this.

The first is, as Dave said, one of the single largest contributors, almost every quarter, to the overall conversion rate, which is the state of macro conditions. If borrowers are generally financially healthy, that's going to be helpful.

If borrowers are struggling, that's going to be unhelpful. And that's just because, of course, approvability is such a big, immutable part of conversion.

I want to point out that there is a second component of conversion, which is also not necessarily normative. It's not good or bad, and that's the mix of applicants. We talked a little about this earlier in the question about targeting and what kinds of people come in the door.

And there's always some trade-off between propensity to apply and propensity to be approved or converted. And so there's always an optimization going on there. And I think that's neither good nor bad. It's just that we do what's optimal for the business. And so that can cause it to move a little bit.

But the thing that I was referencing earlier in my prepared remarks about what happened this quarter and the improvements we've made that we expect to be durable and lasting with respect to reducing variance on this metric specifically has to do with managing how the model responds to the latest signals in macro.

So over the last few years, one of the things that we invested the most heavily in was building our models in such a way that we think they are the fastest, most precise at responding to the latest patterns in borrower repayment, including at the macro level.

So if it's like federal employees or if it's like service sector workers or if it's high primness borrowers or low primness borrowers that are being impacted, or it's everybody being impacted by a big macro event, we want our models to be the very fastest at responding and respond as precisely as the data allows.

And so we've made a ton of progress towards that. We're very proud of the sort of system we've designed and built.

But one of the side effects of that system is that it can be a little overly responsive to the latest changes. And that, in addition to being responsive, there's always some kind of sampling and measurement error.

You can think about what we have, of course, a large amount of data, but relative to all people in the U.S. or the whole economy, it's still a relatively small sample.

So there's a natural statistical sampling error that comes about from that. And we were doing a lot of work this quarter on understanding how much natural error there is in the match between the sample and the actual levels of calibration.

Then we devised some techniques to be able to shrink that measurement error by about half, so that we don't have as much what I call unwanted variance in this metric.

We really just want the model to respond to real changes as opposed to changes that are just measurement error, and we were able to reduce that measurement error by a very significant amount this quarter, which means that in future periods, we expect that all else equal, we will see less volatility in our conversion rates as affected by macro. So that's good.

And Sanjay, I think you also called out that repayments have increased in the script. Any theories on what is driving that? Are those folks refi loans away from you at a lower rate? Is that the borrower's financial? He is just improving, so the people are paying off their loans faster.

Can you just talk a little bit about what's going on there and just the credit implications of that? Have you seen delinquency rates already move because of that?

I mean, as we said, it's an empirical observation that repayment speeds have increased. It seems pretty broad. I mean, I think there are a lot of theories as to what's behind it, but we don't know for sure, obviously.

It seems to be broader than just one specific use case, meaning I don't think it's just a refi boom. It seems to be happening across both partial and full prepayments, which would imply that it's something broader than just a spike in refinancing.

As we said, in the broader scheme of things, this is typically a good thing. When repayments are happening faster, you'd expect that it's on some level of reflection of improving underlying consumer health.

You'd expect it to be inversely correlated to defaults over time. So that's what we would like to see.

But in isolation, with all else constant, repayment happens faster. In the immediate term, it means there's a little bit less interest to be earned on the loans to offset the defaults.

And so in the immediate term, your model becomes a little bit more conservative on pricing. It puts a bit more coupon into the loans. So that it compensates for the fact that the duration of the loan has become shorter in a sense.

So, there's a bit of an immediate conservatism by the model. But I think in the broader scheme of things, we're pretty excited to see it because it means that on some level, personal fiscal situations are probably a little bit more stable.

We'll move next to Reggie Smith with JPMorgan.

The origination number again. But I guess my question is thinking about -- obviously, there are 2 components to the conversion rate are what you guys are approving and then the consumer acceptance.

I was curious if either had an outsized impact on the conversion rate.

And then I was also curious, as you look at your application flow, much of it, do you have a sense of what is shown, I guess, comparison pricing with other loan products?

So, like, I don't know if your loans are showing against LendingClub or SoFi or something like that, if you had a sense of that mix. And the reason I ask both of these questions is that obviously, those 2 companies had very strong origination trends this last quarter.

And I'm just trying to figure out if there was a share shift, if you guys were fighting with one hand, time you back because of your model, like just trying to sort through all that stuff. So anything you can provide there would be helpful. And I have one follow-up.

Yes. This is Paul. Yes, the conversion changes were predominantly related to our model's level of conservatism, so reflected in approvals primarily, and that tends to be the single most sensitive metric when you flip someone from approved or denied, you have a 100% decline in their relative conversion rate.

So that's the thing that's most sensitive and tends to dominate changes in the metric, and that's what happened in this particular time period.

And so, I guess you were like declining super prime. I think you talked about there being some sensitivity in that area. Is that the right way to think about it?

No, declines don't happen in the super prime area. The rates just move up a little bit for somebody who would be at the end of the spectrum.

It's at the other end of the spectrum where there are a lot of declines. So the combination of those 2 things is what really amounts to a lower conversion.

And then, if I could ask one more. Just thinking about the HELOC product, and I know it's early days, but how should we think about the, I guess, day 1 economics or take rate for that product relative to, I guess, your base corporate average?

Reggie, let's see. I mean, I think in the past, we've alluded to the fact that take rates will be healthy, but a little bit more modest than in PL, but on much larger loan sizes.

So, without quite knowing exact numbers yet, maybe you could think about a take rate that's maybe roughly half the amount, but a loan size that's certainly far more than double.

And then the last one, I guess, nothing to call out from a credit performance in your book, despite what you guys are seeing in the UMI, just to be clear.

That's correct. We've seen exceptional credit performance, and that's kind of the whole reason for UMI is to make proper adjustments.

Also, Reggie, your question that we didn't get to was what others are doing in the market and if they're growing at higher rates at this particular period of time. We obviously don't know what's behind their rates. We don't know what their models look like, how quickly they respond to signals they see.

But there's no question, if you just look at the nature of lending, that there's always a way to grow.

In our view, the model is always right. The model is going to tell us what's proven at what price, and we don't overrule the model. So I think that's probably our way of looking at it.

It sounds like, if I'm hearing you right, that the model may have given you guys a false negative, and like a blip, and things are better than what may have been showing up a couple of months ago in the model.

I mean, we don't know false negatives. It may be something that's helpful to us down the road, that it saw what it saw and it priced what it priced.

So it doesn't necessarily mean it was, in any sense, a false negative. It's a constantly learning system.

No, nothing to add.

And our next question comes from James Faucette with Morgan Stanley.

Just a couple of quick follow-ups for me. Can you give any specificity to what elements of the model kind of you saw weaken and then subsequently improve, or other indications? Just trying to get a sense of where your systems may have been looking versus the broader market.

Yes. I think, first of all, we definitely wouldn't describe the model as weakening. As a reminder, our primary performance metric for the model is model separation, and our separation accuracy metrics are our highest ever.

The other metric that we track very closely is what we call model calibration, and that's about how the question of credit performance.

And as has been said several times, credit performance really has been exceptionally strong for us in this time period. And so what was weaker in this period was the model's ability to approve as many people or convert as many people.

And that certainly was a direct result of the increased conservatism that resulted from the model observing a couple of months of elevated risk signals in various pockets of borrowers.

And so that you can think of more of what we call a macro change that the model was responding to. I think with the benefit of hindsight, you could call that a bit of a false negative, I suppose.

But of course, I think in the moment, there is a correctness to reacting to the signals that you're seeing. And I think we directionally think that is the right thing to do. That's what we'd like our model to continue doing.

I did say that some of that reaction, we think, was due to a certain natural noise in what I call sampling or measurement error.

And we did come up with some really good ways to reduce that. And so that noise going forward will be a whole lot less, which is a really, really great technical win for us.

But ultimately, there is some level of directional responsiveness that we always want the model to have to the latest changes in what's going on in the world. And if that means that for a month or 2, the model gets more conservative, we think that's just the right thing to do.

And then, as you look forward to the December quarter and as you're forecasting, how are you thinking about exit rates? You made it pretty clear that you think that there will be a little bit of lagging or continuing effect as we go into the fourth quarter.

But are you expecting that by the time we get to the end of the quarter, you'll be back? And how are you feeling about the right way that we should be thinking about the run rates as we go into 2026?

I think we're quite optimistic about the quarter. I mean, I think we have good growth rates. We are taking an appropriate level of conservatism. We have a very, very good pipeline of model improvements that very typically will drive conversion rates up.

So in our view, actually, a lot of things are working really well. And it's really important from our perspective to say that the model taking a bit of a conservative breather is a feature, not a bug. And if others aren't doing the same, maybe we'll figure out why over time.

But it's the strength of the model, not a weakness, that it's making different decisions or taking a different take on the market. But in the grander scheme of things, we think the consumer's health is good. We think our models are getting better.

The new products are breaking out. So we think we're in for a very strong 2026 and feel very good about the fourth quarter as well.

We'll move next to Rob Wildhack with Autonomous Research.

One more question on this subprime, superprime point. I mean, Sanjay, I think you mentioned that the UMI is lower for subprime, higher for some of the higher FICOs.

If we zoom way out, we all see and hear a lot of headlines around this K-shaped economy, where super prime is doing quite well and subprime is struggling. So why do you think there's that difference between what the UMI suggests and what we're seeing and hearing more broadly?

Rob, it's a good question. I mean, we see directly, obviously, the data we have at our disposal. I think maybe it's important to be precise with labels.

So just to be very precise, if you think about the sub-660 population as measured through the traditional credit score lens, that is a population that, in our estimation, is in reasonably, I would say, actually quite good shape with respect to what their same default trends were pre-COVID.

And so consequently, the UMI is relatively modest. If you go into the primary end of unsecured lending, so now let's talk about the 720 to 750 segment.

Those default rates are quite elevated compared to those same default rates pre-COVID, and their UMIs are consequently quite a bit higher. And of course, we would talk about that segment as being a prime segment in the context of unsecured lending.

Now, if you go to an even higher FI segment than that, let's talk about the 800-plus segment. That is a population that I think is actually doing very well. They probably don't do a lot of unsecured borrowing, though. So they're not maybe in our label set or in our data set.

And so I think you have this U-shaped thing in the economy where at the very low end or maybe the low end of the unsecured lending spectrum, let's call it, 600 to mid-600s, things are very good.

And at the very high end, maybe even beyond the unsecured borrowing population, things are quite good, and then there's like a peak in the middle.

And so I think we all use different labels to refer to different parts of that spectrum. whether one part is prime or subprime or super prime or even not even in your data set. But I mean, very specifically, I think that's what we see.

And then just quickly, a couple of the OpEx lines caught our attention. Engineering and G&A were both lower sequentially, better than what we were all expecting again, better than what was implied by the guidance.

Can you give some colors on the drivers there?

Sure. Yes. I mean, some of it is just like our ongoing fixed expense discipline, which we've been focused on for some time. Some of that is, frankly, mechanical.

As we reduce our outlook as a business for this year, we will reduce our expectation for things like bonus payouts and other comp accruals. And so there's a bit of a mechanical adjustment to a lower outlook that sort of reduces the fixed cost base as well, which is working as designed.

And we'll go next to John Hecht with Jefferies.

First question is, you talked about the use case for the broader unsecured loans. It looks like your HELOC loans are $55,000 to $60,000 on average. Can you give us the use case there?

Home equity loans are general-purpose loans. So people tap them for lots of reasons. We don't have a breakout today of what the use case is for ours in particular.

But of course, people know the most obvious thing is oftentimes used for home improvement, but quite often can also be used for other types of debt retirement or anything.

So we think of HELOCs and personal loans as having, in some sense, being trade-offs from each other with respect to a general purpose set of funds, a little bit different rates, different process.

But in some sense, they are substitutes for each other.

And then I know this might sound a little bit like beating a dead horse. But just on this concept of the UMI and the tightening or conservatism and the dichotomy, just from what we've seen, some auto finance companies, some unsecured lenders, subprime, prime.

And virtually everybody we've covered this quarter has experienced good volumes, but not only good volumes, but really positive credit trends. You guys talk about this concept of calibration over and over.

I guess maybe what I'm seeking is, does your engine not disclose to you what it's seeing that's causing the difference between it and the market? Or are you able to see why it's doing things differently?

You mentioned pockets of weakness in certain populations. Again, what population or what demographic was that, or geography or something? I mean, is there something you can point to so we get an understanding of what this black box is doing to some degree?

Yes. I think the principal way that you should think about this is that we've intentionally built our system so that it can respond faster than traditional credit metrics would.

So in our experience, when other players talk about their credit performance, it's a very backwards-looking metric in the sense that you're typically looking at a somewhat mature cohort of loans and you're measuring something like the actual charge-off rates.

If you think about charge-offs in a lot of something like auto, you're often talking about something that could go 180 days since it was first delinquent. And then there are mixed effects, and then there are sort of effects from new populations getting originated and mixed in there.

And the confounding variables that come with all of those things generally create a pretty substantial obscuring effect to being able to tell what's really going on in credit performance in real time.

And so we've built a system that is much better at precisely being able to, in real time, tell you what actually is going on when you control for all of those variables.

So think of it as a system where holding constant all of the changes in your borrower population across, in our case, the thousands of variables that we use to actually underwrite and understand the risk of loans.

When you control for all of those things, you control for the timing, the cohorts, the vintages, then what are you actually seeing? And how does that interact with any of these thousands of variables so that you can actually see the sort of underlying patterns?

And that, I would say, is one possibility that you could see something that's very segment-specific. I don't think that's the story we have in this particular period.

The other thing that it very simply lets you see is if there is an across-the-board move that would have been either detected 3 or 6 months later by traditional credit metrics or wouldn't have been detected at all because it would have gotten obscured by the sort of changing mixes or new originations getting blended in.

And in our case, we're able to see that. Now again, as I said earlier, I think it's possible to be overreactive to sort of that precise, fast-moving signal. And I think we optimized the balance a little bit better through some of our work this particular quarter.

But ultimately, our goal is to be faster and more precise than anybody else in the market can be. And so we don't find it necessarily surprising that there are periods of time where others are saying one thing, and we're saying totally the opposite.

It appears there are no further questions at this time. I'd like to turn the conference back to Dave Gerardo for any additional or closing remarks.

All right. Thanks, everybody, for joining us today. We're excited to finish the year with a flurry of activity and progress when setting ourselves up for an amazing 2026 for Upstart and our shareholders.

Thanks for joining us today.

And this concludes today's call. Thank you for your participation. You may now disconnect.

Up next, we have Dave Girouard, CEO and Co-Founder of Upstart. Dave, thank you for being here. I want to start. I'll do a little housekeeping first. Today's discussion may contain forward-looking statements that relate to future results and events, which are based on Upstart's information available today and are subject to risks and uncertainties. Actual results may differ materially from these forward-looking statements. The discussion may also include non-GAAP financial metrics, which are not a substitute for GAAP metrics. Please refer to the company's filings with the SEC and its IR website for additional information, including GAAP and non-GAAP reconciliations, along with other disclosures.

Thanks, Will.

Thanks for being here. Okay. So maybe we could start with big picture. Upstart is a lender that uses advanced machine learning techniques to make credit decisions. So can you talk about the importance of AI in your models and how you think about the extent to which you can improve upon the credit outcomes of other lenders using more traditional methods?

Sure. First of all, I wouldn't describe us as a lender. I would say we're a lending marketplace. There are hundreds of lenders and private credit institutions, et cetera. We're more a market maker in the middle. But the whole premise of our business is that lending and credit can be completely modernized in ways that take advantage of technology like machine learning and AI to improve access to credit from the borrower perspective and improve the product from a lender perspective. And we started the company 13-plus years ago, really with the idea that these technologies were coming to a place of maturity. Credit and lending in the consumer segment really hadn't changed much since the FICO score came about in the late '80s. And I was at Google, and the idea generally was the types of technologies we were building at Google for a very different reason could be applied to this very old and important industry.

And essentially, what that means is we've built technology and models that are much better at assessing risk and understanding who is likely to default and who's not likely of default, et cetera. And also just kind of automating the process so that the experience of the borrower is much, much better. Today, in excess of 90% of the loans on our platform are approved in a moment. There's no human involved, completely automated by machine learning techniques. And also what amounts to the risk separation and being able to price the right person correctly.

And so this is really something that just accrues value and gets more accurate as we have more data, more sophisticated AI models, more training, et cetera. And in our view, at least, if you looked ahead a decade to us, it's inconceivable that lending and credit around the world wouldn't be AI-powered in almost every location you see it because the economics are so much better, the opportunity to create a better product is so obvious. And I think the world is now, in the last few years, really begun to open up to the possibilities of AI.

Yes. So I want to talk about maybe the trajectory of the business longer term. I think you said at AI that you want to have the best offer for all Americans, which is a bit of an extension from the TAM where you currently operate. How do you think about competing in some of the more competitive areas like prime lending, for instance, in the U.S.? And then how close are you to having the most competitive rate outside of some of the near prime personal loan categories?

Yes. So I think it's to say the way lending has worked for many decades is every bank or credit union or other lender would kind of spin their own model, they would get a score like a FICO score or something else. And it was -- some were a little better or a little worse than others, but it was a place where most of the technology was just homegrown. And with the advent of machine learning and AI models, that sort of changes things pretty dramatically. It's not realistic for a lender to just spin up a model and hope that it's competitive. So what we basically came to the conclusion a few years ago is lending won't just be this little thing that appears at your bank or at your local or in a neobank, et cetera.

It's a very, very sophisticated technology opportunity that requires intense focus and investment. And so we are building toward a destination for credit that our hope and expectation is 100% of Americans will be persistently underwritten, meaning once we've seen you, once we know you, you are permanently underwritten. There's no process after that. And we can provide access to the very best forms of credit that you might need, whether that be a home loan, a personal loan, a car loan, a revolving credit, et cetera, all different flavors of credit. And at some point, we would like to be able to guarantee the best rate as well.

And this is kind of what AI does. It separates and allows you to get to the vast majority of Americans that are extremely creditworthy and make a better product for them. So we've just started a year ago, as you referenced. Internally, we say we want the best rate and the best process for everybody because that's how to build a lasting brand in a huge enterprise if you can have that sort of brand value. And we're working toward that. I would say, in our personal loan, we feel very good. We're on the cusp of really having the best rates for everybody. And there's a few subtleties to having all the right capital behind that, but we're making great progress.

And in our home product and in our auto products, I think we're also on a very good trajectory. And that would be the goal is to have a trusted brand where you know you will get a phenomenal rate. The process will be effectively no process whatsoever. And as we kind of expand on the product set, and this is our chance to be the largest player in AI. I think we are the only company building a foundation model for credit in AI and the proceeds of that will be coming in the next decade.

And I thought AI did a really good job showcasing how Upstart is ahead of the pack in terms of leveraging some true machine learning techniques in real time, constantly innovating on that and deploying it into the market. It's clear that a lot of the banks are not doing this. When you look at some of the more modern competitors in the space, like do you see signs that others are beginning to apply some of these more sophisticated techniques? And how do you think about the risk longer term that competitors' underwriting models begin to converge and produce kind of consistently different underwriting results than the traditional credit models.

For better or worse, I don't think there's like a wide array of competitors trying to do what we're doing. I think there's a few, if any. If you look at people that are in the kind of buy now, pay later space, people like Affirm, Klarna, et cetera, they tend to have a lot of focus on fraud, given what they are really kind of payments first. And the credit they do is very short term. So it's a little different. And so they have, I would say, some genetic comparability to us, but they're not really pursuing the products that we have. So in that sense, there's others a little bit like us.

But if you think about longer-term credit, 2 years, 5 years, 10 years, 30 years, the players in that market, in our view, are not AI native, and they're not really building AI native products. They aren't building products with instant approvals. They're still based on someone going into a branch, sitting across from a loan officer many times. You can't realistically get a home equity loan from your local bank in a matter of a day or 2, and you can't do it from your home. You're going into an office. So I just think that the industry at large is not building AI-centric models.

There are a few -- I think there are a few maybe copycats, if you will, but there's not a lot of. So I don't see a convergence. I mean, certainly, some others will step up. I think the opportunity is so vast that it will become more apparent to people. But you can't spin this up in a matter of weeks or even years. I mean we've been building models that the data set that it's trained on has been growing for more than a decade. Every loan, every repayment, every late payment on our platform becomes part of the training set. And you can't replicate that overnight.

Yes. Makes sense. Maybe just switching gears to the customer value prop. It clearly beneficial for a customer that can't get access from a traditional provider to gain access to credit because of the underwriting. How do you think about the ongoing relationship with that customer after that transaction in that first loan? And then what do you track on repeat usage that maybe shows the outlook of Upstart kind of being more of a customer acquisition vehicle than kind of like an originator?

Well, first of all, I'd say the idea that you come to Upstart because you can't get the product from your sort of core banking provider would be a very old view of the industry. Banks don't hold their customers tight that way. Credit is -- essentially has been shopped for a decade or more now. If you're doing any form of accessing credit, whether it's for a card, home, whatever, you are generally shopping. And so if you deliver the best rate and you have a brand behind that, you will be a first look, not a second look. And I think that's where we're quickly evolving to. So I think -- another transition we're making is from someone that just gets a transaction with us, gets a loan once. Generally, they're paying it back in an automated fashion.

They don't have to talk to us again. I think you will see more and more from us. About 1/3 of our loans today are from repeat borrowers and people that had great experience came back for another product. We are very rapidly moving into a place where someone might come for a small unsecured loan. We discover we can offer them a better auto loan than the one they have today, and we are able to cross-sell that or somebody comes looking for a personal loan, again, our core product, and they discover they can get a larger and maybe less expensive home equity loan.

So cross-selling products and having that long-term relationship, I think you will see us more and more have products that people interact with more frequently. But today, we've had phenomenal unit economics and lowering acquisition costs through our entire history. If you just sort of went all the way back to the pre-public days, we've been extremely good at improving conversion, having very -- decreasing cost of acquisition. So that's all been on a great trajectory for a long time.

Yes. Makes sense. Maybe we just hit on the macro environment, the kind of obligatory macro question, the upstart macro index. Wondering how you've used some of the more recent revisions to the job data, particularly weaker unemployment environment. How do these data points inform your particular machine learning approach to underwriting standards to the extent that you're feeding that in?

We aren't per se feeding in Fed data. I mean, honestly, most of the data you would get publicly would be so dated, it would not be of interest to us. I think that what we are trained on is just literally the tens of thousands of repayments that come in or could be delinquent any particular day. That is far more real-time accurate view of what is going on in the real consumer economy. But I should say, overall, the consumer has been in a stressed place for a long time now, probably a year or 2 at least as they came out of COVID and the stimulus went away and consumers or families have been like overspending relative to what they earned for some time.

If you look at every category of credit today, student loans, cards, auto loans, et cetera, default rates are much higher than the long-run average and much higher than 2019 before COVID happened. So that's not anything to do with us. That's the way the world is right now. That has been priced into our loans for a long time. Of course, interest rates are higher, too. So we have grown really quickly. We had triple-digit growth rates in Q2. But frankly, that is against a headwind, which is both higher interest rates, which is part of the cost of funding as well as a higher risk premium in the market.

And at some point, that will unwind. But our view generally is the consumer has already been stretched very fully. And I would generally say we're not macro forecasters, but we are not terribly worried about unemployment. I just generally -- our view is there's a sort of secular lack of workers in this country. And if anything, reduced immigration, et cetera, is only kind of stressing that even further. So I don't -- of the many risks, and we tend to worry about everything, at least somewhat. But I would generally say the risk of sort of super high levels of unemployment feels like pretty remote to us.

Got it. Yes. So I guess, summary then would be you have more frequent data that's already sort of priced into the model. There's not like a management overlay or anything that you would do around underwriting to react to some of the recent data that's been coming out.

No. The model is generally targeted to be somewhat buffered against expectations so that if there was some worsening in the future, we'd be fine. But generally speaking, we're using real-time repayment data as the basis for what we would call calibration.

Got it. Okay. So sticking with sort of a macro theme, the overall funding environment, there's been a lot of capital flowing into the consumer credit space in general, the rise of private credit, alternative asset managers looking to increase their allocations to the space. You and a lot of the consumer lenders out there in the space that I cover, a firm, SoFi, et cetera, have had significant access to, call it, alternative capital supply, off-balance sheet supply. How do you think about the sustainability of this capital through a credit cycle? And what are you kind of watching for signs that some of the fixed income investors out there are getting more nervous about consumer credit allocations?

I view it as, first of all, it's just the early stages. We really created the first partnership with private credit in consumer lending in 2023. So it's only been a couple of years. But it's important to say these partnerships were created and structured purposely to survive through credit cycles. If you looked at the way funding and sort of online lending, et cetera, in the prior era, it was very much dependent on hedge funds. It was at low funding. It was very ABS dependent. If the securitization markets were fluid and pricing well, there was more money than you know what to do with. But of course, that could change, and it did several times change. These are long-term partnership structures that there is alignment, there's co-investment. And they're designed essentially to work through credit cycles. Private credit companies have long-term locked up capital. They're looking for a certain mid or high teens ROE on that capital.

And we are very well aligned to do that.

So as long as we're co-invested and the structures are set up so that through good times and bad times, we can continue to deliver on that yield, then I think they're great partnerships. They've also begun to get longer. The first partnerships we signed were single year, moved to 18 months. They're moving to 2 years. So I do think there's great alignment. This is not like the subprime mortgage crisis with originate to distribute and no verification of anything. These are highly aligned business models. So I don't necessarily -- it doesn't mean everybody will perform because our skin is in the game. And we think that's the right way it should be.

How can anybody commit long-term capital to you if you don't have any skin in the game. But at the same time, we need that long-term capital. If you look at what we filed when we share in our sort of earnings deck quarterly, we've returned something in the range of 9%-ish ROA on average, if you invested since in our platforms, I think, since like 2018. So that ROA very easily translates into the ROE that these firms need. And that's the way to think about it. Over a long haul through whatever cycle, you can deliver that. And it's a little insurance like in the sense that there could be worse quarters and better quarters. And collectively, it's -- there's actually great alignment.

Got it. I wanted to talk maybe around pricing then. I think during the quarter, you mentioned you did some testing on pricing elasticity. And so I was wondering if you could just expand a bit on what that means and how you think about the pricing opportunity in the business. But I'll lump it in with maybe a bigger picture question just on spreads in general. So several years back, you saw a big increase in industry loss rates. Upstart had to put some loans on the balance sheet. Since then, funding markets have come back, obviously, in a big way, there's not an issue anymore.

I guess fixed income markets tend to be more backward looking than forward looking sometimes when it relates to kind of lender performance and track records. Do you feel like you're getting credit for the improvement in underwriting models over the last several years? Or do you think lumping it back with the pricing question. Is there more room to go to see improved spreads on some of the loans that you distribute?

I think we're getting credit where we need it. I mean the market is not what it was. I mean, today, the influx of private credit dollars isn't remotely like hedge funds buying, no risk retention by the originators, et cetera. That world does not exist today. Even ABS markets have improved a lot. The bonds have -- the yields have tightened and the spreads have tightened, but there's really not a fluid market for the equity or the residuals. So like I don't view this as like a crazy market of just money flying all over the place. These are very carefully structured co-investment type markets.

But I think, first of all, our credit has been performing really well for some time. I think the partners we have would probably gladly take more from us, which is really what led them, I think, to go to other platforms. We'll see how everybody performs on that. But right now, we're very confident in the performance of our models. They calibrate really quickly now. They -- our separation continues to get better. And I think this is just a natural -- this is an AI shaped problem. And we feel very good about that. I think the private credit coming into this market is a realization that depository credit is just one form of low-cost dollars. There's plenty of other dollars that need yield over time.

The relationships between private credit firms and insurance companies means you can have a blend of capital to solve these types of problems. So I think it's actually a natural evolution. And I love that our platform has a combination of depository capital funding for the primmest of borrowers and then private credit to have more structured style funding for less prime parts of the market. Our product set is now going from home equity, which has super low loss rates, 1% or less, all the way to higher risk smaller dollar products. But that's as it should be. That's what a true credit market looks like.

Yes. That makes sense. Okay. On one of the near-term questions, specifically on approval rates, they stepped up meaningfully in the most recent quarter. I was wondering if you could talk a little bit about what drove that? And then bigger picture, talk about how you view the normal cadence of model improvements on an ongoing basis.

Sure. So our models were sort of continually doing research and making improvements to our models. They can come in the form of -- or come from new sources of data or which have some ability to improve the models or improved algorithms. So this is just like OpenAI is trying to work on the next version of their ChatGPT or some other product. We are constantly doing research and then periodically, we actually release new models quite regularly, probably in the order of every 2 or 3 weeks. But those can just be some of point releases and then maybe more of a once-a-quarter cadence of kind of a primary release. Those releases essentially drive better separation. Like that's the whole goal is to better separate good risk from bad risk.

When you do that, effectively, you, in the end, are getting some likely credit losses out of the pool, and that allows you to lower the rates to everybody else, which improves conversion. So calibration neutral, meaning if the economy is not changing a bit, then each new model is generally going to improve conversion, which is how we grow. That is the sweet spot of how we've grown over years is just improving conversion rates. I would just say that is in contrast to the entire world of credit and lending, which generally has very static models. They might upgrade them every year, every 1.5 years.

They aren't generally looking for a model win to improve conversion. They're more in an exercise of loosening or tightening based on perceived performance of their credit. This is more structural, a structural change, which says a newer model is trained by more data, has more sophisticated algorithms behind it, can separate good risk from bad risk better and results in better conversion. We're at the early stages of this, but I think it's a new world that is completely different from the history of credit and lending.

So I want to pivot a bit and talk about some of the newer verticals. You've been testing this for some time now. This quarter, your commentary seemed to signal kind of moving to later stages of operationalizing them. I guess where are you in terms of testing and tweaking the underwriting models? Is there anything left to do before pursuing some of the funding arrangements that you brought up on the call were kind of slated to go live later this year.

Yes. So we are in the process of becoming a 4 or 5 product company and expect we will be within a few months. And I'd say that we've had products that we offer to consumers for some time, as Will mentioned. But really to get them ready for third-party funding, you have to have, first of all, a significant enough volume that credit is performing, and you can sort of prove that on a statistical basis. Unit economics are good. You're originating at a return level that is acceptable to the market, which we've been doing for some time.

So all those sort of check boxes are done. We're not needing to like tweak those. These products are ready for market. We're essentially in the final stage, which is you have to have the agreements in place, you have to have the operational abilities in place to transfer loans, et cetera. So I would just say our goals and our expectation is the majority of these products will be third-party funded by the end of the year. So we're in this last phase. It was, I would say, unplanned and a little unusual that we have 3 products, all of which are going through this final step at the same time.

As I've said, it's like Monday is the first day of school and the triplets are going to all start on Monday. We wouldn't have exactly planned it this way, but it turned out this way. But I think the opportunity really is that our expectation is by next year, these products will be vast majority third-party funded. And our balance sheet, I expect, will begin to turn into more cash than anything else. And so this is just that final step. There's nothing holding us back, and I think we're on target to have this done by end of the year.

Got it. Makes sense. Let's -- maybe let's dive into a little bit on each of these. First, on T-Prime, could you talk about how you view the opportunity in T-Prime loans? Prime loans, obviously, a much larger part of the market, more competitive, more bank involvement. So do you see the same inefficiencies and opportunities in the prime segment as you see in kind of the traditional near prime personal loan space?

Yes. I guess the context -- T-Prime is a term we use to market to banks and credit unions. The opportunity to come and get very, very prime borrowers who are, of course, attractive to that type of institution to cross-sell into other products. So a bit over a year ago, we basically came to the conclusion we were known for and very experienced at serving non-less prime, near-prime borrowers and really did not have good offers for someone who would be an obvious prime person. Think of somebody with a 750-plus credit score and solid income, et cetera, a traditionally prime person. And we kind of felt this was not long term in the best interest of the company.

It's hard to have a brand just to say we might have an amazing rate for you, but we're not sure compared to what you could get elsewhere. So we basically said we want to get to a place where we can have the best rate for everybody. And that really meant bringing more depository capital in that can serve the lowest rates to the best borrowers. And that effort has really gone incredibly well. Our market share in the super prime end of the market, which has moved up very rapidly. You can see in our earnings deck, it is in our core product, maybe 1/3-ish of our borrowers already in just a year.

And again, it's that best rates, best process for all, we really needed to go after the super prime part of the market. I would also say that it's interesting that we don't think that has to be served entirely by depository capital. Private credit, along with their kind of insurance partners really can create very, very compelling sources of funding for the whole market. So that's who we're becoming very quickly. And of course, home equity is itself a very prime product. So you will see from Upstart over time, our goal to be best rate for everybody and have the best experience for everybody. And T-Prime was really the kickoff of that effort way back in mid-2024.

And it seems like a good time to be getting into home equity loans. The industry has been in long-term decline into the last couple of years, and I'm sure some of the elements on interest rates and home prices have something to do with that. So what do you kind of see in the market opportunity right now, consumer interest for home equity loans? And then just how are you approaching that opportunity?

Well, yes, I mean, home equity loans, in some sense they're popular when you want to tap the equity in your home, but often in a high interest rate environment, refinancing your mortgage doesn't make a lot of sense. There's no opportunity there. So they're just a way to tap into home equity for whatever expense you might have. That could be an improvement to your home, but it could be a wedding, it could be paying off some other debt. So there's a lot of reasons why home equity loans are popular. Traditionally, they've been the type of loan that you go to your bank or you go to your credit union, you will take 5 to 6 weeks on average to close a loan. So to us, that's like as that exists, not a very interesting category.

What we're building toward is a product that is really very close to our personal loan, if not instantly approved within a very short period of time. The hard part, of course, is you have liens and titles to deal with and state regulations and such. But we're moving very quickly toward a place where a consumer can come in. If they want an unsecured loan, they can have it in a moment. If they want a little bit more effort, they can have a home equity loan. And we're not talking about 5 or 6 weeks, we're talking about a few days. And hopefully, at some point, less than a day, we are getting into the first automatically improved home equity loans.

So it's a category that there's a couple of online digital types and then there's -- it's a very fragmented market. There's a lot of banks and credit unions who have cheap capital, but they have a process that's archaic and slow. So in our view, this is a category where we will be the most AI-native participant. We can grow market share and hope to grow market share for a very long period of time. And we hope -- we would expect over time to have a whole suite of home lending-related products.

That makes sense. Can you talk a little bit about just some of the operational differences in that product relative to the unsecured space or in auto, things like collections, foreclosures, just mortgage servicing in general, tends to be a little bit more labor-intensive. Like how are you approaching that? How are you kind of dealing with the cost associated with it?

Well, both home and auto, yes, you have an asset out there that at some point, you can repossess or not repossess. So first of all, it's important to say one of the great things about home in particular, is the default rates are much, much lower. I mean, less than -- typically less than 1%, right? Nobody wants to lose their home. That's why they're right at the top of the repayment list. I think we're just growing into and getting better at servicing in the auto segment. We're not servicing yet in the home segment at all. But generally speaking, our servicing and collections capabilities have gone from, I think, pretty mundane a couple of years ago to quite state-of-the-art as we're beginning to introduce AI and machine learning models into how we do loan servicing.

And the opportunity there is much, much larger, I think, than anybody appreciates because the subtleties of when to intervene, how to intervene, what message to deliver, when you might settle a loan versus let it go further or sell it off to a debt buyer. I mean there's so many avenues and paths in terms of how you deal with a delinquent loan and servicing and collections that AI is a very natural fit there. And I would just say we didn't focus on that for a long time. Only in the last couple of years, we're beginning to gear up. And the results have been pretty extraordinary. We hope and expect over time, we will have an independent credit servicing business that we would begin not just to service credit originated on our platform, but everywhere. And I think that's another very large market opportunity for us.

Yes. I mean, since you brought up, I think you mentioned on the call, I think there was some servicing related improvements to the model this past quarter, isn't that right?

Yes. We have very ambitious goals in the range of like 20% reduction in defaults over the course of a year, not by anything to do with the credit decisioning at the front or by the macro economy, but just better servicing and collections of loans. So we've been able to achieve those kinds of improvements. And I think, honestly, we're still at the like earlier stage because we're barely getting the first machine learning models into production. And the opportunity there is very large. I mean it's just hard to imagine -- if you just think about it, like, for example, it's very common that once someone has charged off, meaning you've charged off a loan, then you might pursue a settlement on that loan.

But there's no real reason. Maybe you should pursue a settlement earlier if your model tells you that's how you're going to get the maximum dollars back. So you can just see how over time, it's very obvious, there's a lot of different choices you can make, interventions you can make that the dollars back can get better. And when that happens, by the way, that goes directly back to a better price to the next borrower, right? And that's kind of the essence of the model. The more you can collect better, the more -- reduce the cost of origination. All these things accrue to a better product to the consumer. So there's a great flywheel there.

Got it. Okay. In the last minute or so here, I was hoping we could talk through just economics on some of the new products, how you view sort of the stack ranking of sort of like unit economics on dollars of GMV from kind of a top line and a contribution margin perspective.

Well, the products are very different. I mean we literally have a small dollar product that could be a loan for a few hundred dollars all the way to today, a home equity product that could be a $70,000 home equity loan, and we'll have larger loans and we'll have smaller loans. So it's just going to really cover the vast. I think generally, what we are optimizing for as a business is contribution dollars, right? We are not in the business of making loans where there's nothing in it for us or building for some magic future. But at the same time, I mean, the percent can vary a lot by the nature of the product. But when you think about how we originate, what we originate, how we spend on acquisition, we're looking for net additional contribution dollars. And I think that's for us always the thing, along with keeping our fixed expenses extremely low as a business.

And you can already see, if you've just looked at our economics in recent quarters, growth goes immediately to the bottom line. We've moved back into GAAP net income profitable. It's our expectation and our plan to stay that way and grow our profits. But that's the nature of our business. It's low headcount. It's super efficient, super automated credit origination. In the last quarter, 92% of the loans on our platform had no human involved in them whatsoever. They are approved in a moment. We're getting to some what we call loans where there might be some other process going on after the fact, but there's not a human involved in it. So as you start to think about all the possibilities of AI, I don't think there's anybody in our industry who's as AI forward as we are, and I think the advantages will continue to accrue.

Great. Well, I think with that, we're out of time, but thanks so much for the conversation. Thanks for joining us.

Thank you, Will.

Good afternoon, and welcome to the Upstart Second Quarter 2025 Earnings Call. [Operator Instructions] As a reminder, this conference call is being recorded.

I would now like to turn the call over to Sonya Banerjee, Head of Investor Relations. Sonya, please go ahead.

Thank you. Welcome to the Upstart earnings call for the second quarter of 2025. With me on today's call are Dave Girouard, our Co-Founder and CEO; Paul Gu, our Co-Founder and CTO; and Sanjay Datta, our CFO.

During today's call, we will make forward-looking statements, which include statements about our outlook and business strategy. These statements are based on our expectations and beliefs as of today, which are subject to a variety of risks, uncertainties and assumptions and should not be viewed as a guarantee of future performance.

Actual results may differ materially as a result of various risk factors that have been described in our SEC filings. We assume no obligation to update any forward-looking statements as a result of new information or future events, except as required by law.

Our discussion will include non-GAAP financial measures, which are not a substitute for our GAAP results. Reconciliations of our historical GAAP to non-GAAP results can be found in our earnings materials, which are available on our IR website.

With that, Dave, over to you.

Thanks, Sonya. Good afternoon, everyone. Thank you for joining us today. Before I begin, I want to welcome my co-Founder, Paul, to the call today. As most of you know, in May, we hosted our first Investor Day, what we call AI Day. For many investors and analysts who cover Upstart, it was their first exposure to Paul. Unsurprisingly, AI Day generated a lot of interest in how he and his teams are creating the world's leading AI lending platform. After the event, many told us they'd like to see and hear more from Paul, so we asked him to join our quarterly earnings calls. You'll hear from Paul in just a bit.

On to the update. On our call a year ago, we provided the first signs that Upstart was returning to growth mode. And today, you can see it in full bloom. The second quarter was exceptional for Upstart. In addition to achieving triple-digit revenue growth, we reached GAAP profitability a quarter sooner than expected. Additionally, our newer businesses, Home and Auto, actually accelerated off the amazing growth you all saw from them in the first quarter.

Originations on the Upstart platform in Q2 were $2.8 billion, our highest volume in 3 years. Revenue in Q2 grew 102% year-on-year, helping us deliver positive GAAP net income for the first time since Q2 2022. Our Auto business grew 87% sequentially, while our Home business grew 67% sequentially. While this friendly sibling rivalry tends to go back and forth in terms of growth rate, I can happily say both businesses accelerated meaningfully from their Q1 growth. For the first time ever, more than 10% of our originations came from our newer businesses, including our small dollar loans, which grew 40% sequentially. Our teams couldn't be happier. After a long period of super focused execution, it all just seems to be working right now.

Once again, our growth last quarter was not a result of dramatic macro improvements or Fed rate decreases. In fact, the Upstart macro index has been largely stable for several months now. Our growth was primarily on the back of model improvements, which helped to drive conversion rates from 19% in Q1 to 24% in Q2. These wins came first and foremost from Model 22, which we launched in early May. Paul will share more about our model advancements shortly.

In addition to our ML team, our growth and operations teams continue to do amazing work to drive down the cost of acquisition and origination. These are technology-driven economic wins that result in a superior product for the consumer and a sustainable advantage for Upstart.

As I mentioned earlier, our emerging businesses are growing really quickly. Small dollar loans and Auto each crossed $100 million in quarterly originations in Q2, and we expect Home, the new kid on the block, to follow soon. Our newer products collectively drove almost 20% of new borrowers on the Upstart platform in Q2.

For each of these emerging products, we're now reaching the point where credit history is sufficient and volumes are substantial enough for third-party funding. In fact, we have a goal to transition most of the funding for these products off our balance sheet by the end of 2025, though deal timing is always hard to predict.

It's worth noting that our Auto Retail product, that is our software installed at car dealerships, has really gained traction and momentum in the last couple of months. This product has always presented unique challenges relative to our others, and it's clearly taken Upstart some time to get it right. Several months ago, we took the decision to narrow the focus of our software on an exceptional financing process, and this focus has paid off in spades. The dealership adoption right now is like nothing we've seen in the past, and the volume of loan requests and closed agreements from our dealer partners is on a steep climb. This is a recent phenomenon, and I expect we'll share more about it as it plays out.

In our Home business, we're increasingly confident we're on a path to building the best-in-class HELOC experience. Home is a massive and fragmented category with few players versed in AI and its amazing potential to power superior home lending products. In Q2, we launched instant property verification, with the first applicant completing the entire verification process in under 1 minute. Our system automatically verified their identity and income, assess the property's value and any existing liens and confirmed ownership and vesting information, all the key steps needed to close the loan. I believe this speed and efficiency in what is normally a slow handcrafted process is without precedent.

We continue to strengthen the funding supply on Upstart's platform. Our funding partnerships have been both durable and scalable, allowing us to grow rapidly while delivering the target returns our partners expect. With respect to banks and credit unions, we expect to reach a new all-time high for monthly available funding in Q3, surpassing our prior peak from early 2022.

The funding markets continue to improve as the year progresses, particularly since the Liberation Day fears in early April subsided. In June, we priced and closed our second ABS deal of 2025, delivering significantly improved execution compared to our first, which closed in April. It's worth noting that the more recent transaction had nearly twice the number of investors as the first, including some new names.

We feel increasingly confident that these committed funding partnerships can scale with our business as needed and will play an important role as we begin to commercialize our newer products.

Before I turn the call over to Paul, I'll share a few final thoughts. Looking over the last couple of years, we've done a lot of work to run our business more efficiently and streamline our cost structure, but we had conviction that investing in much larger Home and Auto opportunities made sense. These categories are ripe for AI disruption and they've expanded Upstart's TAM by more than 10x.

Our considerable investments in Home and Auto are really paying off with fast growth, strong credit performance, rapidly improving separation and commercial readiness with 9 lending partnership deals recently signed across one or more of our secured products already. To be clear, our goal is market share leadership in each one of these product categories in the future.

As our CMO, Chantal mentioned at AI Day, we're building the always-on everything store for credit, aiming to persistently underwrite 100% of Americans with the best credit products in the world just to click away, and we're off to a great start.

Thanks. And now I'd like to turn it over to Paul, my Co-Founder and Upstart's Chief Technical Officer. Paul?

Thanks, Dave. Our aim at Upstart is to win by having objectively the best rates and process for borrowers and technology, specifically AI is how we do that. To that end, I want to highlight several areas of recent progress.

First, we've continued investing in our core AI advantage. Model 22 made use of neural networks at every level of the model architecture, whereas prior models only made use of neural networks in the base layer. That may sound like a subtlety, but it increased our separation accuracy advantage over our benchmark textbook credit model by 17 percentage points to 171.2%. Equivalently, it decreased the inaccuracy remaining to be solved to 87.5%. This is a metric where the starting point is the benchmark textbook credit model I described back at AI Day and 0% would be a model that gets every credit decision perfectly right. As you can see, there is a long way to go, but fortunately, we have a commensurately long road map of model improvement ideas to get there.

As of the end of Q2, core underwriting had 91 million borrower repayment events to train on, up from 86 million at the end of the prior quarter. To support the larger and more complex models, we invested in further parallelization and cache-ing solutions that cut up to 17 seconds of latency off borrower pricing and saved on model costs. Those time and resource savings can now be reinvested in yet more powerful models.

Second, servicing is the newest frontier for us and realizing loss reductions via best-in-class servicing has been a major focus. Over the past year, including the most recent quarter, we launched numerous improvements and optimizations to how customers can pay, how much they pay and when they pay. As a result, year-over-year population-adjusted delinquency rates are down 20% and raw delinquency rates are down 32%.

Machine learning is already informing many of these optimizations and will soon allow us to determine the causal impact of servicing actions we take. This will include assignment of specific agents, hardship programs or settlement offers to specific borrowers. We also plan to apply machine learning to the problem of individualized recovery prediction for the first time ever, replacing a fixed assumption about an economically significant portion of loans’ cash flows with machine learning. Servicing wins directly improve loan loss rates on loans, which in turn improves the pricing and approvability of new loans.

Third, we made strong progress in Q2 generalizing our AI technology across product verticals. I want to start by noting that even with accelerating growth in new products, our share of fully automated loans actually kept up this quarter. That will be challenging to keep pace with, but we're encouraged by wins we had across new products.

As Dave mentioned, HELOC had its first instant property verification, which involves solving for over a dozen facts or documents that previously required waiting for a manual verification. In Auto refi, we launched full automation of the remote online notarization process. Both of these wins remove major procedural barriers to model-driven automation, which we've seen relentlessly drive the percentage of loans fully automated up in core personal loans over the past few years.

Our growth in auto has been supported by and coincides with strong advances in generalization of our core underwriting technology. Auto is the first area where instead of directly training an auto model, we start by training a foundational credit model on data from multiple credit categories and then apply fine-tuning to arrive at an auto-specific model. We are now working to add embeddings to the auto retail model, along with generalizing what we call “APR as a feature” and our macro framework from personal loans. This type of model generalization is powerful because it means all of our loan products can learn from repayment patterns observed across our platform, not just within their individual category.

Lastly, I want to touch on generative AI and its applications to our business. I'll start with the table stakes. Like any good tech company, we've realized solid productivity wins from application of large language models to our internal operations. 60% of our developers are weekly active users of LLM-powered developer tools and teams all across the company have built over 700 custom GPTs to automate various internal workflows.

More interesting are the applications to the end borrower. We've already launched early versions of borrower impacting generative AI tools around model explainability and customer service. We will continue to build on these with an eye towards eventual agentic management of our consumers' credit needs.

As Dave has discussed, one of our key priorities in 2025 is to 10x our leadership in AI. We continue to have a robust pipeline of modeling wins, and I'm incredibly proud of the team and what we've been able to accomplish so far.

With that, I'll turn it over to Sanjay. Sanjay?

Thanks, Paul, and thanks to all of our participants for sharing some of your time with us today. I'll now spend some time giving context on our numbers.

With respect to its impact on financial performance, the credit environment we operate in was largely a non-story in Q2. The emergence from last quarter's tax seasonality played out roughly as expected. The broader macro has been idling in regards to its impact on credit trends, registering as neither a significant headwind nor tailwind over the past 6 months.

As Dave alluded to, the strong sequential momentum we achieved in Q2 is largely due to the strength of our model launches during the quarter. In addition, take rates and contribution margins increased in the core personal loan business, although in our aggregate numbers, these dynamics were partially offset by the continued rapid scaling of the newer Home and Auto products, which still have immature unit economics.

The combination of these effects allowed us to beat our guidance across both top and bottom lines in Q2 and break through to GAAP profitability a quarter earlier than anticipated.

We have been able to comfortably fund the ongoing growth in the core personal loan business through our existing lending relationships and capital structures. The main source of pressure on the balance sheet as it currently stands is from the continued scaling of the new products and an increasing priority for us this year will be to finalize and implement our third-party capital plan for these new products.

With this as context, here are some of the financial highlights from Q2 of 2025. Total revenue for Q2 came in at approximately $257 million, up 102% year-on-year. This overall number included revenue from fees of approximately $241 million, which was up 84% year-on-year and 15% better than guidance.

Within this, transactional revenue more than doubled year-on-year, largely reflecting the influence of the aforementioned Model 22. Separately, servicing fee revenue grew by nearly 20% year-on-year as the outstanding book of serviced loans continued to expand.

Net interest income represented roughly $17 million of overall revenue, ahead of guidance by $2 million, reflecting the growing volume of new products being incubated on our balance sheet and in particular, the Auto book of loans where our return on investment has meaningfully strengthened.

The volume of loan transactions across our platform was approximately 373,000, up 159% from the prior year and 55% sequentially and representing just over 250,000 new borrowers. Average loan size of approximately $7,570 was 15% lower than the prior quarter as model advancements drove higher approval rates in smaller loan amounts.

Our contribution margin, a non-GAAP metric, which we define as revenue from fees minus variable costs for borrower acquisition, verification and servicing as a percentage of revenue from fees came in at 58% in Q2, up 3 percentage points from the prior quarter and exceeding guidance. This improvement reflects a strengthening take rate in our core borrower segment in addition to the acquisition and operational unit cost efficiencies driven in part by Model 22.

GAAP operating expenses were roughly $252 million in Q2, up 16% sequentially from Q1. Expenses that are considered variable relating to borrower acquisition, verification and servicing were up 21% sequentially relative to the 55% increase in volume of loan transactions, supporting the higher contribution margins previously referenced.

Fixed expenses were up 13% quarter-over-quarter, largely reflecting a onetime catch-up in the compensation-related accruals, which on the current business trajectory, we expect will normalize in the back half of the year.

Q2 GAAP net income was approximately positive $6 million, well ahead of expectation and reflecting outperformance on fee revenue against our tightly managed fixed cost base.

Returning to GAAP profitability has been an important objective of ours over the past year, and I am proud that our team has reached this milestone ahead of schedule and while subsisting in the persistently high default environment that still surrounds us today. Now that we are over the line, we will look forward to continuing the positive momentum of our bottom line and to improving our profitability profile as we scale.

Adjusted EBITDA was $53 million, also scaling nicely in accordance with our operating leverage. Adjusted earnings per share was $0.36 based on a diluted weighted average share count of 118 million.

We ended Q2 with approximately $1.02 billion of loans held directly on our balance sheet, up from $815 million in Q1. This sequential increase is mainly due to the continuing growth of our new products, which have all simultaneously entered the transitional period between R&D and commercialization, a period in which we must ramp deliberately in order to demonstrate credit performance and our ability to deliver meaningful volume before obtaining third-party funding commitments.

In this regard, we are in a bridging period with these new products, which is precipitating what we expect to be a temporary expansion of the balance sheet usage that we intend to reverse as these products exit incubation. As Dave mentioned, we have already begun the process of securing external capital to support these initiatives, and we believe these efforts will allow us to transition away from direct balance sheet funding of these in the near term.

As we plan for the back half of the year, our macro assumptions remain consistent with our prior view, which is to say a steady environment with the UMI continuing in the 1.4 to 1.5 range, steady interest rate levels and a labor market that remains resilient in the face of unpredictable policy shifts. Inflation will remain a near-term risk. In this environment, we expect to continue to launch model enhancements that will improve conversion rates, our take rates and contribution margins will remain robust, and we will continue to scale and fund the newer products.

In this scenario, for Q3 of 2025, we would expect total revenues of approximately $280 million, consisting of revenue from fees of approximately $275 million and total net interest income of approximately positive $5 million. Contribution margin of approximately 58%, GAAP net income of approximately positive $9 million, adjusted net income of approximately $44 million, adjusted EBITDA of approximately $56 million with a basic weighted average share count of approximately 97 million shares and a diluted weighted average share count of approximately 105 million shares.

For the full year of 2025, we now expect total revenues of approximately $1.055 billion, consisting of revenue from fees of approximately $990 million and net interest income of approximately positive $65 million, an adjusted EBITDA margin of approximately 20%, and we expect GAAP net income of approximately positive $35 million.

These numbers are, of course, the outcome of a lot of hard work and great execution by the various teams across Upstart. So I'll take this opportunity to both thank and congratulate all of those teams.

And with that, operator, over to you to kick off the Q&A.

[Operator Instructions] Our first question comes from Peter Christiansen with Citi.

Really great results. Dave, Sanjay, I just wondering if you could chat about ABS -- the health of the ABS market. Are you seeing any appetite for equity tranche investments at all? And then I'm curious if you had any thoughts on more competitors coming into the more near prime space, also the prime from a loan platform perspective. Are you seeing any competitive pressure there?

Pete, Sanjay here. Great to hear your voice. The ABS markets, let's see, we -- I think have returned to being a somewhat regular issuer now on the cadence that we would like. I think the bond market is very constructive. The residual or equity market that you asked about I would call it an opportunistic market right now. There are buyers. I don't think they -- it is an efficient market, and I think buyers sort of pick and choose deals. So I wouldn't call that market “back in the way that I think the bond market is”. But overall, it's a constructive market, and we're certainly happy to be a regular issuer again.

Pete, this is Dave. I'd just say on the competitors, to the extent the funding markets have improved and capital markets have improved somewhat through the year, I think that does tend to bring more competitors into the space. So unsurprisingly, it's a fairly competitive game these days. But again, we're very focused on having best offers, both at super prime level and at our core business as well and also very confident in our ability to grow our market share and keep our strength in those markets as well.

[Operator Instructions] We'll take our next question from Ramsey El-Assal with Barclays.

I wanted to ask about the increase in the loans on the balance sheet. And Sanjay, you mentioned that those you would start transitioning to external funding in the near term. I just want to kind of zero in on what that means exactly. Should we expect next quarter that amount to begin rolling down, decreasing? Or is it going to take a little more time to line up the external funding category? And I guess what will be the pace of decline there that we should expect over the next few quarters?

Yes, Ramsey. Yes, as I mentioned in the prepared remarks, a lot of the volume on the balance sheet today are from our new products. So our core business, I think, is well funded. Categories like Home and Auto are growing quickly. So it's a bit of a good problem to have, sort of the original use case for the balance sheet, which is R&D and incubation.

In terms of time frame to get the flows moving to third-party capital, I think we're looking at a time line that's sort of roughly between now and the end of the year. And a lot of that is just about new originations and getting that flow to capital sources. I think as we make those deals, obviously, we'll opportunistically use our balance sheet to seed those relationships. So I think as we sort of transition the new flows, you should see our balance sheet start to release as well. But yes, I would give sort of a couple of quarter time line on that dynamic.

Okay. One quick follow-up. The super prime percentage of loans as a percentage of total just declined a little bit quarter-over-quarter. I'm just curious what is the driver there? Is it mix? Is it underwriting decisions? How do you -- how is that trending right now?

I don't think it's anything in particular because I think we have enormous room to grow in our core as well as in super prime and both are growing very quickly. So it wouldn't surprise me that it kind of goes back and forth. We have enormous market share opportunity across those product segments. So to me, that's not surprising.

We're also really just beginning to build in the depth of funding in the super prime segment. So we do have very competitive funding in super prime, but we need to build the depth there, and that will allow us to scale while keeping prices in a very competitive place. So that's kind of the process going on there. I wouldn't read anything more into it than that.

Our next question comes from Simon Clinch with Rothschild & Company Redburn.

Maybe I could just start with the pretty impressive step-up in contribution margin versus your guidance. And Sanjay, could you perhaps break that down? I mean I presume some of that is due to sort of lesser mix of prime within the mix of loans that you've originated. But also the comment around take rates being higher. Could you just talk to that and elaborate a little bit more, please?

Yes. Sure, Simon. Yes, I guess the overall contribution margin improved. That is in the face of yet sort of growing new products, which have immature unit economics. So you may infer that the contribution margin of our core business grew by even more. And within that, there's both a mix benefit from having slightly more core borrower segment loans versus super prime. Those loans obviously have higher margin than very super prime loans.

And then within that core borrower segment, there in of itself, our contribution margins and our take rates improved. And some of that is a result of the model launch we had, an improved model improves conversion rates that decreases acquisition costs like-for-like. And so there's sort of benefits to the unit cost side.

And then even our take rates within that core borrower segment in the personal loan business saw some improvement, I think largely a result of the ongoing cost or sort of the take rate optimization that we're always doing, trying to understand elasticities in the different bands and optimizing against them. So that gave us some opportunity to improve our take rates and our contribution margins overall.

Okay. That's great. And just a follow-up, just on the outlook you've provided and your comments around the macro that we're seeing. Perhaps you could just give us a sense of what your -- what assumptions you're making around that are sort of fed into your guidance? Because I see that the third quarter is a little bit above consensus, but the fourth quarter looks like it's kind of unchanged relative to consensus. And given the substantial beat this quarter, it feels like you're holding something in reserves. So I just want to get a sense of that.

Well, let's see, one part of your question was about macro assumption. We're typically conservative with respect to the macro. We sort of roughly expect the status quo. So the main way that we measure that, of course, is in UMI, the macro index we have, which is hovering in the 1.5 range, and we sort of plan to a consistent UMI for the rest of the year. So remaining in a relatively high default rate. We plan for no real cuts in interest rates in the market. Obviously, there's a lot of speculation around what that might look like for the rest of the year, but we certainly don't bank on anything in that regard.

And we're sort of continuing to rely on a relatively resilient labor market, notwithstanding the noise of the last week or so. I think the labor market continues to be in relatively good shape in terms of how many open jobs there are out there versus how many people are seeking jobs. So that's sort of the totality of the macro assumptions that go into our planning, and I think it's a relatively conservative sort of kind of a status quo, if you will.

You also asked a little bit about the shape of the guidance. I would just say that we have relatively direct line of sight into what things are going to drive our Q3 numbers. Those projects are very sort of near term and rounding the corner. And so we feel very confident in being able to guide against them. I think there's a lot of things we're excited about with respect to Q4 and how that's going to go, but I don't think we quite have the line of sight required to guide specifically against them. And so I think that a lot of the near-term uplift you see is just excitement over some of the sort of the projects and the dynamics that we can see much more in front of us.

And our next question comes from Dan Dolev with Mizuho.

Amazing results, as always, very proud of you. My biggest question is the UMI, what could make it go up or down as we move throughout the year? This is the key question.

Great to hear from you. Let's see. UMI, I mean, it is ultimately a reflection of the impact of the macro on credit trends. I think the things we think most about in terms of what could move it, what could make it go down? I think the main thing is improvement in savings rates, improvement in consumption patterns relative to income. I think we've been consistent in saying that the American consumer in aggregate is probably overspending relative to the income levels that we're earning, and that's been true for a while now. And if that balance improves, we would expect that credit trends would improve as well. In the opposite direction, you might imagine things like a reacceleration in inflation or significant unemployment.

And we'll take our next question from Kyle Peterson with Needham.

I want to start off on like the average loan size and the take rate. It seems like that's been drifting down, at least in the core personal loan product. Should we continue to see that drift down? And I guess like is that a strategic shift? Or is that like a broader response to the macro where that's where you guys are seeing like the most favorable risk reward here, I guess, just trying to level set on kind of whether -- what's strategic versus like what's the market conditions response here?

Kyle, this is Dave. I think you can, I guess, categorize it as strategic, meaning it's intentional because it really is reflecting the very rapid growth of the small dollar product, which for us is really, again, pushing the boundaries of the credit model, getting much more people onto the platform. It accounts for a lot of our new users. They can be upsold to other products like Auto refinance, et cetera. So it is definitely the fact that, that product, which has much, much smaller loan sizes is growing very rapidly.

And again, our goal is to have like every American persistently underwritten on the platform. So having more and more ways to get them in is, from our perspective, good and having more products to cross-sell to them. So that's all part of the larger game plan. And I don't know if you'll see it continue to go back down. Right now, products at the other end, like mortgage and home loans are growing, but -- so at some point, these will outweigh each other. There's no real change other than the product mix itself is getting more diverse.

Okay. Okay. That's really helpful. And then I guess on some of these new products, obviously, it seems like a really good opportunities are in like HELOC and Auto. How would, I guess, like you guys compare the competition there versus the core personal loan product? Is it equally as much of a knife fight? Or like is the sledding any easier or tougher? I guess, just kind of how should we think about the economics of these products, especially as they shift to external funding and the ability to scale quickly? Any color there would be really helpful.

Yes. I think I mean these products are different and our ability to create a very differentiated product happens in a different way. I would say in the unsecured products, the underwriting itself is a big part of the advantage and the magic that we bring to the market, and that is what's built the company. And the newer products, particularly like Home, Auto refi, they're actually quite low loss rates, prime-ish products. But the real opportunity for both Home and Auto is to create a very differentiated experience and process that costs a lot less to originate and also just creates a far better consumer experience.

So the relative ability to price differentiate isn't as great as it is in unsecured, but the ability to create a very much differentiated experience for the consumer and also a lower cost origination is much larger for those products. So overall, that's what we kind of keep pushing on and sharing. We mentioned automating a lot of the process of getting a home equity line of credit. That's a product that normally would take more than a month on average to get from your local bank or credit union, and we can do it just in a few days or even faster. So that, to me, is important. AI can bring not only pricing things properly, but also just eliminating the friction and reducing the risk in highly automated, very efficient ways.

The next question comes from Mihir Bhatia with Bank of America.

Maybe just starting with the newer products, particularly the Home and Auto. You mentioned you're working on additional funding partners or funding partners to get some of that off the balance sheet. Can you provide a little bit more color on that? Are those going to be more bank partners? Are you thinking securitization? How are you thinking about that?

Mihir, this is Dave. I mean it will be a combination of banks and credit unions. For both Home and Auto, they have a lot of history and familiarity with those products. Particularly HELOCs are something that are extremely popular in the bank and credit union world. So I would say on a relative basis, they would probably have a larger play there relative to the institutional capital, private credit capital. Though as we go, we will always find the right most competitive combination of capital to have the best product in the market. And I think that's actually what's unique about our position as we have both depository capital as well as private credit and other sources of institutional funding in effect, competing with each other to make the best product for the consumer. And I think these things relative to the unsecured product, this will swing -- I think the other products will swing a bit more toward depository capital.

Got it. That's helpful. And then just turning maybe I think on the conversion rate improvement, you mentioned the biggest driver was the new product, the new model, which launched in early May, if I heard that correctly. So does that mean you only got the benefit of 2 months. So 3Q conversion rate should be even higher from there? And just if I could also just throw in there, if you could just talk a little bit about the Walmart partnership, any call-outs there?

We aren't really forecasting anything about conversion rates for current quarter. There's always puts and takes on conversion rates. When it goes up, we often end up spending more and pulling it back down intentionally. In other words, turning conversion rate into extra growth. So there's just not a straightforward kind of up and to the right on conversion rates.

If you sort of see the chart that we provide in the investor deck, I think it's a great illustration of how conversion rate trades off with volume. And so I'd say that. On Walmart, we continue with that partnership. It's been a great success for us thus far, but we don't have anything new we want to share about it today.

And our next question comes from Reggie Smith with JPMorgan.

Congrats on the quarter. Really strong quarter. I had a follow-up on the conversion rate. I'm not sure if you guys have shared this in the past or if you're comfortable sharing it. But I'd love to hear about, I guess, the 2 elements of conversion rate, the approval rate and then kind of the acceptance rate is how I've been thinking about it. I guess, one, am I thinking about that right?

And then two, can you talk about how those ratios have maybe changed versus the prior year? And then as we think about the new model, maybe anecdotally, talk a little bit about the types of people or the profile of the people that may have been rejected before that they're being approved today. Obviously, I don't want to give your secret sauce, but just any color you can give there? And I have one follow-up.

Reggie, your sort of decomposition of conversion rates is correct. It's sort of a product of our approval rates and the subsequent exception rates of the loan. We've never really decomposed it in how we analyze externally, and I don't think we have any off-the-cuff narratives around the relative subtrends there. It may be something that if it's interesting, we can sort of look at exposing over time or in the future, but I don't think that's anything that we have any great soundbites for you as of right now.

This is Paul speaking. Just on the second part of the question about specific types of borrower characteristics that we may be waiting more or getting more of. I think the short answer is that like we've said a number of times at AI Day and other instances, the real power of our model comes from its ability to find many, many small subtle relationships in the data, and that's happening at multiple levels of the model architecture as we described with Model 22.

And the unfortunate result of that is that it's not like there is one -- well, unfortunate for answering the question is that there's not really one simple answer of like we have suddenly got more high credit score borrowers or more low-income borrowers or anything like that. It really is just picking a couple of borrowers from many different sort of parts of the credit fabric, if you will, and then finding borrowers who are more likely to repay than their sort of conventional credit characteristics would suggest.

That makes sense. Okay. And then if I could, Slide 23 in the deck, you guys give this every quarter. We see, obviously, the numbers are increasing. It looks like the assessed value is greater than the co-invested value up until this point. Maybe help us understand like how should we interpret this slide? And what should we take from it? And then as far as outperformance or underperformance relative to expectations for this piece of your portfolio, where does it show up? And how do we see that flow through? Because it looks like maybe things are better than you even thought when you put these loans on the books. I'm just curious where that shows up.

Reggie, well, as far as what takeaway from this slide, I mean, it does pull together, I think all the various ways in which we are co-investing in risk capital deals. So it hopefully gives you a holistic perspective of what that investment level is at any point. I think in terms of thinking about how to model it, we're sort of in the ramping phase. To the extent that these deals start paying back a couple of years into the deal, you should expect this to sort of ramp probably for a few more quarters, and then it will start to level off as the amounts we're investing in new quarters are roughly offset by the amounts coming back in from prior deals.

And so there's sort of a ramp-up and then a sort of a platforming of this amount. And then, of course, this tells you how we're doing on those investments. I think early on, in the early sort of instances of these deals, our goal is to make sure we were preserving capital. And so you want to make sure that the way that we're valuing what these positions are is at least on par with what we invested. I think more recently, we have an intention to start earning returns on these investments. And so you'd expect or hope that the assessed value of these positions starts to grow in relation to the invested capital. But I think the idea of this slide is to give you a picture of how this investment is trending and how the returns are looking.

In terms of how it shows up in the P&L, it's probably a much more complicated answer to your question because the reality is it hits on various line items depending on the structure of the deal, and there's a lot of different structures at play here. But I think in a -- at a very high level, you can expect that the amounts that we're assessing at current value, if they were to hold, they will make their way back into the P&L largely in the form of fair value improvements or net interest income really. And so while there's a couple of different paths back into our financials, I think that's probably like if you were just to really crudely simplify it, that's how it will show up.

And we'll take our next question from James Faucette with Morgan Stanley.

I wanted to just talk about really quickly, you mentioned your CAC and some benefits you're getting there. But you've also always been really CAC efficient during even lean periods and whether it be your organic or your own CRM mind leads. As you expand though, how should we be thinking about how much is purely organic traffic to upstart.com, how much is originated via direct mail and how much is sourced via third-party marketplaces? And how should we think about the evolution of those types of channels?

James, this is Dave. I mean, I think the long-term trend has been more repeat borrowers, both in the core products we started with the unsecured product and increasingly cross-sold into other products. So those can be, as you might describe it, mined from our database or just people that have a relationship with us, and that's largely close to 0 CAC and an increasing fraction of our loans. The offset will always be how fast are we growing and acquiring new borrowers. So it's not necessarily bad, of course, if new borrowers are suddenly coming on board much more quickly and most of those are paid for one way or another.

So I just think we're moving toward a place where reliance on aggregators or other forms of sort of people that represent our brand is sort of slowly declining over time. And we have much more sort of direct relationship with the consumer and more things to offer them. I would think the thing that is really of note in the last few quarters is we are beginning to really get much better at how we -- how to properly cross-sell and through lots of testing and things, somebody that might have gotten a personal loan or a small loan, and we are able to like refinance their auto loan at a lower price. That's something that works much better as a cross-sell than it does at a first-time acquisition.

So I think as you see the Auto and Home categories grow, you're seeing a lot of cross-selling to them. And that's just the road that we're on, which is, I think, more repeat longer-term relationships with consumers. But also, by the way, it also sort of -- I think we're proving the long-term value of a customer that we serve, and that allows us to invest a bit more upfront with the confidence that it's going to generate more margin downstream to cross-sell to other products or second loans, third loans, et cetera. So I think that whole model, of course, is something we've worked on for a long time, and it's how we've been able to grow and have kind of acquisition cost per loan actually for years has kind of generally gone down.

Got it. And then I wanted to ask, you guys often give updates on what portion of the loans are being handled completely automatically. And I don't know, I may have missed that this time, but just it was interesting that you called out within some of the new products like HELOC and Auto that you're looking at taking advantage of your improved automation, basically that becoming part of your brand. And I've got to imagine that's really helpful in this cross-sell. Where are you at in terms of the automation levels now for -- and where do you think you can get to ultimately?

Yes. Thanks. Great question. So we obviously haven't broken out the exact fully automated percentages for the new products at this point. But I think it's safe to say that they're a fair bit less mature than they are in the core personal loan product. And that's not surprising. I think they're both newer products, but also they start out with more challenges. I think ultimately, we believe that those challenges can all be overcome.

And I think we shared in the preprepared remarks that we made quite a lot of progress in both Auto and the Home categories in this quarter. And in particular, we had our first instances of fully automating several new parts of the home loan process for our HELOC product. And there's still quite a bit more work to be done before those numbers will reach the personal loans level, but I think they are on a very good trajectory towards that. And I think in the long run, that would be our expectation.

And our next question comes from Rob Wildhack with Autonomous Research.

On the fair value adjustment in the quarter, I noticed that was higher or more than a negative than it's been in the past. Can you speak to the drivers there? And then I think along the same line, the aggregate NII guide is down for the rest of the year despite the increase in balance sheet loans. So how you're thinking about fair value marks for the rest of the year?

Rob, yes, I mean fair value has some volatility to it. I would say the rep contour is UMI, our macro index sort of fell by a lot in the second half of 2024. So we took some of those gains as fair value marks as it persisted into the year. So in Q1 and partially in Q2, you saw some of the benefits of that declining UI. Now UMI has drifted up a little bit in the first quarter or 2 of this year. So that's starting to reflect itself in Q3. In Q4, I think you're starting to see some of the benefits of the risk capital deals, some of the earlier vintages of risk capital deals that are starting to repay. So there's sort of some benefit happening there. And you sort of have a bit of a dead period in Q3 where the UMI has drifted up. So there's a bit of pressure there. The risk capital deals are not quite yet starting to materialize as benefit.

And then you have this sort of phenomena that depending on the seasoning of your loan book, if I could describe the fair value of one loan in isolation, it would be really high at the beginning of the loan because there's really -- there's a lot of interest and no charge-offs. And then at some point around month 12, you hit the peak charge-offs. So the fair value of that loan at that time tends to drop. And then at the end of the loan, the charge-offs, the curve has sort of run off and you get pure interest again and it's high again.

So you have this sort of natural phasing of the value of a loan where it sort of goes from high to low to high. And when you project that over an entire portfolio, you can sometimes get these effects where the value, even though it sort of averages out to 0 over time, has a bit of seasoning volatility, if you will. And some of that is playing in Q3 as well. So I know that's like -- that's a large recipe of a lot of ingredients. But the reality is fair value is quite a complex sort of topic. But I think those are the main things that are impacting the trends over the quarters.

Okay. And then bigger picture, competitively, it seems like all the personal loan origination -- or excuse me, originators are growing very quickly lately, even someone like SoFi is loud and clear that they're drifting into near prime with their loan platform, which is kind of your space. You grew quite a bit in core loans, too. I'm wondering how you guys think about adverse selection in a competitive environment like this one.

We certainly think about it. I would say, first of all, it's one of the reasons we really focus on making sure our cost of capital is competitive across the spectrum or anywhere we want to participate or else you are at risk of adverse selection. So at the conceptual level, we just have to make sure that the fuel, the dollars funding the loans are as good or close to as good as anywhere, and then you have much less concern about that.

Also, I mean, the nature of our models that have gotten sophisticated enough to handle issues of like how does the price of the loan affect the performance of the loan. If you recall back last fall, we introduced what we call APR as a feature, which was quite an innovation on our side, which helps us make sure the price of the loan is considered when you're measuring the risk in the loan. And that was a giant leap forward back in model 18. So we kind of feel from a technical perspective and also from a business perspective, we are both aware of and responding to potential for adverse selection. So a very competitive market is not new to us and something we feel pretty good about.

Sorry, if I could just sneak one more in on that, Dave. How do you compete with deposit funding from somebody like SoFi or LendingClub in terms of cost of capital?

Well, about, I would say, 25-ish percent of our loans are funded by deposits. So it's just that they're not our deposits, they're from credit unions or bank partners. So that's an important way. And that's, of course, for the primest of our loans. So that's deposits compared to deposit. We're just doing it in a distributed manner as opposed to it being Upstart Bank or something.

The other thing I think is important, which is happening quickly is I think non-depository capital is getting more competitive at the primary end of the spectrum just because there are sources of funding that have all sorts of different blends of risk and reward appetite. And through the worlds of private credit, insurance, et cetera, there is non-depository capital that may not be exactly as inexpensive as depository capital, but getting closer and closer. So I think that difference, which historically was quite large, is actually beginning to blend much more these days.

And our next question comes from Kyle Joseph with Stephens.

A lot of them have been addressed, but just thinking about the product mix, obviously, you guys talked about the benefits of AI to the personal loans. And then obviously, you're seeing good growth in Home and Auto, but factoring in the law of large numbers, just kind of how you see the mix going forward and any implications in terms of margins and/or customer acquisition costs?

Well, let me just speak to the mix to the extent we can, and maybe Sanjay can comment on what it might mean for margins. Look, we, of course, feel like there is enormous growth potential in all our product areas, including in our core personal loan product where we have very, very well-established margins and automation, et cetera. And then these newer categories are generally much larger categories, but we're much earlier into penetrating them.

So we think we have a very attractive, like very large addressable market. We're very early into it. And almost inevitably over time, I think the secured products, meaning Home and Auto are going to grow as a fraction. And could they become much larger than unsecured over time, it's quite possible. I don't know that we know that. I think unsecured as a category even though it's smaller, is growing, it's quite popular with consumers.

So I guess that's a long-winded way of saying, look, we're in like single-digit market share in unsecured, which we have enormous advantages and strengths and just at the cusp of really beginning to move in Home and Auto. So the potential for very, very rapid growth over a long period of time in all those categories is pretty significant.

Yes. I would just add with respect to margins. I mean, the margin profile of these new businesses is still materializing to some extent. But all these markets have the same rough shape as the core market we exist in today, which is the unsecured market. And that's that -- there are some segment of very well-served borrowers where prices are competitive and margins are thin and you largely compete on process and distribution. And then you always have a part of the market which does not have access or is underserved, and there's a lot of opportunity to create volume and margin. And I think that's no different in the auto space or even in the HELOC space, frankly.

I think these loans that are secured will have larger loan sizes, maybe smaller percentage take rates and just a similar dollar revenue and/or profit per loan that we have in our unsecured business. But we don't know the exact precision of that yet. We're still -- that's still materializing as these categories are growing for us.

And we'll take our next question from John Hecht with Jefferies.

Like Kyle, most of my questions have been asked. I guess I'm curious as to kind of your sense of what interest rate reductions would do. I assume the Auto refi business would pick up, and the HELOC business, I imagine as well. But in the core products, do you pass on rate changes to the customers? Do your -- some of your partnership agreements with the private credit, do those contemplate changes in interest rates? Just to kind of get a sense if we go into your rate cycle, what to think.

John, yes, a reduction in rates would telegraph itself into our core business virtually one for one, not immediately with some lag. But if rates in the economy go down, that means financing rates go down, and that means your -- the sort of ROA of your unlevered loan can go down commensurately. And so it would result in lower rates to our borrowers and conversion improvement. And yes, the sort of the types of structures we have in place with our committed partners contemplate this and would sort of act in this fashion. So it would be unambiguously good if rates were being reduced, at least in a direct sense because it would mean that rates for borrowers over time would sort of reduce commensurately.

It appears as there are no further questions at this time. I'd like to turn the conference back to Dave Girouard for any additional or closing remarks.

All right. Thanks, everybody, for joining. Q2 was a great quarter for Upstart, no doubt. For those of us in the inside who saw kind of the radical makeover we've been going through in the last couple of years, it wasn't a surprise, but it was very rewarding. So thanks again. We're very excited for what we'll do the rest of this year, and we'll see you all in November. Thanks.

And ladies and gentlemen, this concludes today's call. Thank you for your participation. You may now disconnect.

All right. We'll go ahead and get started. Thanks for joining us, everybody. I'm James Faucette, Senior Fintech Research analyst here at Morgan Stanley. Really appreciative of Sanjay Datta of Upstart being here with us today. But before I get started with him and really kind of put your feet to the fire, I do have a quick disclosure to read.

Today's discussion may contain forward-looking statements that relate to future results and events, which are based on Upstart's information available as of today and are subject to risks and uncertainties. Actual results may differ materially from those forward-looking statements. Please refer to the company's filings with the SEC and the IR website for additional information and disclosures. The team at Upstart, wanted to make sure I read that.

And then for my own compliance, for important disclosures, please see the Morgan Stanley research disclosure website at morganstanley.com/research [ disclose ]. Also, the taking of photographs and use of recording devices is not allowed. If you have any questions, please reach out to your Morgan Stanley representative.

So yes, with the legal leaves out of the way, maybe, Sanjay, great to have you here at the Morgan Stanley 2025 U.S. Financials Conference. Maybe for those that aren't familiar, and it's hard for me to imagine that's the case, but just in case, could you give us a brief overview of your business? What are your problems that you're trying to solve? And how do you fit into the broader ecosystem?

Sure. Yes. It's great to be here. Thanks for having us. So let's see. We're a company that's called Upstart. We've been around roughly since 2014 in our current incarnation. We are a platform for consumer credit. We bring together borrowers requiring credit and funding sources, seeking yield. And the core of what we do is we try to create the technology in the middle that largely has to do with risk modeling. So the underwriting of the credit, and the modeling of the fraud.

I guess from a first principles perspective, we've always believed that the existing space of consumer credit on the one hand, is the most advanced market in the world. On the other hand, the risk models empirically are not that good. And as a result, a lot of people are left out from having access to credit who are otherwise perfectly good credit. And many of those who do a large percentage of the APRs is paying default subsidies for other people who will default, they have never met. And if you have risk models, the math is pretty compelling for how you can include more people, approve more people and reduce the default subsidies everyone is paying and thereby reducing APRs.

So we have this vision of trying to improve approval rates and reduce APRs through the simple sort of application of modern technology to the prediction models and credit.

Got it. No, that's really helpful. And it's interesting because I think the way you fully described it, a lot of times it's shortcut as just AI-based underwriting and that kind of thing. And we've heard that's a topic that comes up obviously now with every senior management, et cetera. And we'll get more into how you're able to differentiate yourself there, both from an underwriting as well as a process standpoint.

But I want to ask just quickly on recent conditions. What's the trajectory been like through May? Maybe you can help paint a picture for us in terms of what you've been seeing in the recent couple of months of April and May as it relates to origination volume growth and delinquency trends. And in particular, are those factors trending in line or better than your expectations? What's happening? Can you talk a little bit about the macro index that you use and how that's been evolving?

Sure. Yes. I guess maybe for both understanding volumes and delinquencies, it's useful to sort of understand the context of the macro environment or at least specifically what we care about, which is consumer repayment patterns and default patterns. And if you sort of take a step back and you look at those more broadly, and you sort of take maybe a reference period of pre-COVID. For many years, pre-COVID, the world was kind of stable, both in terms of interest rates and default trends. And then the world got really good in about 2021 and default rates apples-to-apples got really low. And of course, that was largely an outcome of all the stimulus that was flooding the economy and the money supply got really big and liquidity was flush.

And starting in about 2022 and basically throughout 2022 and 2023, those default rates went from very good up into the right to very bad. And in rough terms, if pre-COVID was like, call it, 1.0, 2021 was maybe a 0.6. So 40% lower default rates than pre-COVID. And that number sort of went up into the right and peaked probably somewhere around 1.6. So 60% worse than pre-COVID. So it was a really tough 2 years in which defaultiness, if that's a thing, was just kind of consistently getting worse month-over-month for the better part of 2 years. That sort of thing stabilized coming into 2024, was stable for much of 2024. And then coming into '25, it moderated a little bit, like it's probably now down to 1.4, 1.5. So that's sort of like a writ large, maybe a view of how consumer default trends have sort of evolved.

Maybe separately, if we want to go into why all of that has happened because it's not obvious at a surface level why the world should have done that in a world where there's been very little unemployment. But nevertheless, it has. And that sort of is reflected in origination levels and delinquencies. If you look at origination levels, as default rates, as default risk started going up into the right, originations in the industry plummeted, and we were on the front end of that for some reasons that had to do with phasing and how that worked its way through the various segments within the borrower base. And I think that contraction ran for the better part of 2 years, sort of stabilized when default trends peaked. And now that it's not going up into the right anymore, and in fact, it's normalized a little bit, our originations growth has picked up.

And that's maybe another piece of the puzzle that's important to understand about our particular business model because we have a unique growth model in credit. In credit, if your underwriting is relatively static, your growth levers are limited, right? You can either spend more on marketing to acquire more or you can loosen your box, if you will, and lose your credit standards, neither of which are particularly palatable. If you can improve your underwriting over time, what that means is you're sort of on the margin doing a better job of avoiding some of the defaulters in the system. And so everyone else is paying a lower default subsidy. APRs go down, acceptance rates goes up, everything gets better.

And so that is something that happens consistently for our business. It's been like a consistent pace of model improvement over time. That got overwhelmed by the negative environment that existed. Now that the environment stabilized, we can sort of grow through model improvements again. And that's what you've seen from us in the last couple of quarters. Now that we're no longer facing a stiff headwind, every month, every quarter, we have a model that's a little bit smarter. It's a little bit better on the margin, avoiding an incremental defaulter. Everyone's APR has come down. We can underwrite better for the good borrowers, and that creates growth.

So to get back to your original question, originations were down and sort of troughed for a while. I would say in the last 2 or 3 quarters, we're back on a growth trajectory. We're obviously guiding growth for the rest of this year. The assumption behind that growth is that the macro environment will not get worse. It doesn't need to get better, but it just needs to be sort of stably high. And delinquency trends will reflect that as well. Delinquencies like our loan cohorts underperformed a lot when default trends went up and into the right, obviously, we sort of recalibrated to that new world.

And now that, that world is stable, I think delinquencies are relatively consistent. They're flat, and they're in line with how we've calibrated the models. And of course, everyone is asking, well, what's happened since Liberation Day? And the answer is in our repayment data, nothing yet really. Like none of that has trickled through into real-world behaviors on credit repayment or default.

Got it. Got it. And just as part of that, one of the things that you typically see is -- and we talk a lot about this internally with the Econ team, et cetera, is that a lot of times, when the market is moving around, maybe the lower income, less wealth-exposed customer, maybe they don't respond, but the higher income or at least higher wealth customers will. But you haven't seen any of that either.

Sorry, responding to what?

Just responding to changes in market values, stock market indices levels, et cetera?

Yes. Yes, nothing really. I mean in terms of asset values, I mean, the market has gone down, but it's kind of come back up. There's been a lot of uncertainty and volatility, but it's not clear what direction it's all going in, so.

So that volatility doesn't seem to have created any impact across the borrower base that you have.

Not on the borrower side. You can imagine it creating some funding challenges because those funding sources don't have volatility and uncertainty. But on the borrower side, I think the impact to sort of someone's day-to-day life has not necessarily yet materialized from all of this stuff in the news.

Right, right. So let's talk about on the funding side, to your point, that's usually where I think about there being stress emerging first, especially in volatile market environments. How does that stress manifest itself for you on the funding side, especially in a situation where you guys have been working very hard for a while to have more secure sources of funding and more predictable sources of funding?

Yes. Traditionally, as markets become more uncertain, spreads get wider, ABS markets become harder to navigate, then the funding sources tend to become more scarce. I think in our particular case, we spent a lot of the last couple of years trying to do deals with counterparties in structures that are sort of designed to survive as [indiscernible]. So they're meant to harvest some overperformance in benign periods. Certainly, there will be periods or vintages where there's macro surprises and that leads to underperformance. A lot of these structures are sort of predicated on how performance works over the duration of that cycle.

And these are with counterparties that are themselves more resilient than what we've worked with in the past. So they've got LP bases and funding sources that are a bit more durable. And so they're as yet untested. But in theory, I think we've got more resiliency on the LP base of these capital providers. We've got contractual commitments that are bidirectional. And we've got mechanisms to figure out or to try and make these partnerships whole over the duration of the cycle and not a specific vintage that requires on ABS trading and not with counterparties that are historically maybe a bit more fickle when it comes to redemptions and LPs sort of resiliency and things like that. So we put a lot of things in place. I guess, at some point, it will be tested.

Got it. Got it. Appreciate that. So let's go back to the underlying technology, though, model differentiation. This was a particular focus at the Investor Day, especially I think the team did a really good job on illustrating some of the improvements you've made with respect to calibration and handling the dynamics of the macro environment.

Perhaps explain what those dynamics are first for the audience, like what are the things that you track in the macro and the impact on loss variance? And then let's -- I'll just set it up for you so that you can then explain why you feel so confident now that the model is appropriately primed to react to changes such that you can, by extension, be more confident in your ability to deliver stronger credit outcomes.

Yes. So I guess, maybe, again, part of the context is a very good machine learning problem is speaking out a lot of information about the borrower and asking your model to take a point of view on the borrower relative standpoint. That's always been our bread and butter. A less good machine learning problem is trying to get it to anticipate what's going to happen in the macro because every macro event is a bit of a sort of a unique event. So we've poured a lot of effort into making our models not predictive of the next macro event, but much more reactive to it.

And because we collect a lot of alternative data upfront when we originate loans, our models can actually do this in a very nuanced way now. So for example, if tomorrow -- I don't know, maybe you might have a hypothesis that government employees are suddenly a higher risk now because of those. I mean that's probably a level of assumption that -- well, maybe that one is obvious, but there's many that are subtle. But because we collect a lot of employment information, we know in our borrower base, employment profiles and such. And if something were to happen such that, that population will become more risky, imagine they inflect with their default rates, these models will pick it up, and they'll pick it up instantly and react to it.

So the precision with which we can detect things happening in the macro and the speed at which we can sort of react to them is much greater than it was maybe even in 2022. But those have also given us the tools to sort of not anticipate these things, but like maybe manage with conservatism through them. So if the world is at -- in a place where it's defaulting at a 50% higher rate than pre-COVID, these are now tools which we can use to say to the models, look, I want you to assume that it's going to be 60% going forward. And so we can create a bit of conservatism in how we think about the environment, give it some buffer to get worse. And those are the tools by which we manage these committed capital deals because we do need to create some upside in the deals in benign periods to pay for downside.

So there's a lot of like subtle things in the models themselves now that they're not going to allow us to avoid underperformance in whatever happens next in the macro. But I think our models compared to 2021 will react much more quickly and much more precisely to it. It would have dramatically limited the underperformance that we did see in those vintages if we had the tools that we do now.

Got it. So that's underwriting and underwriting performance. But another part of the work that you've done that I consistently am impressed by is level of automation for the loans that you're underwriting. You've talked about automation rates reaching an all-time high of 92% for unsecured loans. How do you see this impacting your ability to serve different borrower segments, especially as you expand into the prime and super prime market? How important is this?

It's very important. In fact, there's sort of 2 different sides of the same coin if you think about like risk and friction are related. If you want to lower your risk, you can put a lot of friction into the process. You can ask for a lot of documentation to make sure that you can verify every fact they've given you. On the other hand, if you want to reduce friction by not asking for that documentation, you can introduce more risk. And there's always like a trade-off frontier. And what our modeling allows us to do is push that frontier out. How do you reduce friction and documentation but not take on increased risk? And the answer is by being better at fraud modeling.

And so there's kind of 2 different manifestations of the same thing. If you're working with high-risk borrowers, the value is probably much more in reducing price because default subsidies are very high. If you're already working with very prime borrowers, there's not as much loss or price to take out of the system. But you can manifest that goodness as sort of instant process, and that's much more valuable to the prime borrowers. So you talked about the fact that we are becoming a little bit more aggressive in competing for prime borrowers. Historically, that's not our sort of our sweet spot. But it's largely a game of cost of capital, but the technology contribution we are making is by making it a very frictionless process because those borrowers that are very prime have a lot of alternatives. And so the less friction you put in front of them, the more positive the selection bias.

So when you think about like your relationships with your bank and capital partners and those people that you're originating loans on behalf of, et cetera, how do they -- how do you think they weight the advantages and benefits of underwriting versus automation? Like you said, it's part of the same coin, but it seems like depending on what they're trying to do, maybe it's just a simple, hey, if I want to increase my portfolio exposure to prime and super prime, I like automation more. And if I want more subprime perhaps with a higher return, maybe I weight the model more. But I'd love to hear from you.

Well, with banks and credit unions specifically, there's a very specific thing going on, which is we're not selling them assets. What we're doing is we're giving them technology. They are the brand. They are underwriting and originating the loan -- sorry, they're using our technology to underwrite the risk, but they are originating the loan as an entity. So that looks like a consumer transaction to them. They care a lot about the experience.

I think the financial services famously has relatively poor NPS scores. So when you can give them a process that's very seamless that the consumer loves, that's a big deal to them. And they do tend to originate borrowers that are primer that -- there's not a lot of pricing to take out of the system. So they want a borrower that's relatively prime with little adverse selection and a very seamless process is really what matters most to them.

Got it. Got it. Got it. So is that something that you can continue to push is like that seamless experience, the automation? And where do you think you can get to from that 92%?

Well, in our core business, I guess, I mean if you think about the theoretical limit, it's 1 minus the fraud rate, right? Fraud rate, you measure in basis points. I imagine it's 100 basis points. Your theoretical limit of automation is 99%. You should give 99% of the people the money and put all the friction on that 1% that you are worried about fraud. That would be our mission. Obviously, that's a theoretical limit.

But maybe the other way to think about this is 90-some percent of our loans are fully automated. But if you look at it on the basis of the applications, it's only about 70%. And why are those numbers different? It's because the ones that are eligible for automation convert like 3x better. 70% of the applications represent 90% of the loans. But that means there's 30% of the applications that are not converting well because there's so much friction. There's still a fair amount of process we can take out of the system. Considering -- I mean, I'd used 100 basis points, but the reality is fraud is, I think, 30 basis points.

So like 30% of the applications we're not smart enough to automate yet, but only 30 basis points of them are actually lying to us. So that suggests there's still a long way to go.

Got it. Got it. And then on that point of 3x the conversion, is that substantially different between prime and non-prime borrowers in terms of the conversion like that...

Conversion?

Yes.

It's universally similar. Universally. Yes. People like the moment you ask for a pay stub or something that there's a conversion drop-off that's very clear.

Got it. And sticking with the theme around Prime, you've been meaningfully mixing your -- or meaningfully shifting your mix, excuse me, towards the prime segment, and this is generally what we think about 720 FICOs and above. And you mentioned that your March originations were up roughly 125% year-over-year and now are about 32% of your overall origination mix. What's going to get us to 50%? Are we going to go over 50%, especially since those borrowers tend to be able to borrow more? How should we think about that?

I mean I think it's important to say it's not necessarily an intention of that, right? We're not targeting some mix. We didn't have an ambition to increase the prime mix. We want to be successful everywhere. We just happen to have more success in the prime segment in the last quarter, partially because we're sort of still kind of more getting started in that segment. And we had a lot of success. I mean I think a lot of the competition or a lot of the competitive dynamic in prime lending is cost of capital. And we had a lot of success working with banks and credit unions to reduce their target returns, and it gave us a boost.

So that was like an outcome of goodness that we benefited from, but it wasn't as a result of us trying to change the mix. We would love our mix to be representative of America. The rough -- the sort of back of the envelope is half the country is prime and half is not prime or something like that. So that would be a good outcome. We would be like equally competitive across the board.

But if you're 50-50, but then by exposure or origination volume, doesn't that mean you're going to be substantially more prime or not necessarily?

Yes. By construction, it would because loan sizes are larger in the group segment for sure, yes. That's probably right.

Got it. We've been going for about 25 minutes here. I want to see if there are any questions from the audience. Let me...

There's a question there.

Well here in the back. Let's give you a microphone.

Are there any constraints -- I mean, one of the things the new administration, Mickey Bowman gave a speech on Friday talking about community banks. She comes from banking family. Her family actually owns the oldest bank in the state of Kansas and a lot of her speech was about community banks. Are there any things that this administration can do to help -- I don't know, to help some of the pain process of like growing in this area of consumer credit. Do you feel like people are -- do you feel like bankers are constrained at the community segment?

It's a great question. Yes, banks are undergoing a lot of evolution, and they have been since '23 or maybe even slightly earlier when a lot of the turmoil happened and some of the failures happened. I think that -- so banks have definitely pulled back as direct lenders in this economy for certain. I think there's both regulatory scrutiny and very punitive capital charges that are at work. And I don't actually have a pretty much fast view as to whether those are good or bad things. If you wanted banks to step in more as direct lenders, you would have to relax those regulations and those capital sort of charge constraints. And I don't -- again, there's pros and cons to doing that. We do want deposit-taking institutions in this country to be very safe and you don't want them to take on too much.

I could make an argument that I have in the past that it's not clear to me why banks have a history as direct lenders and insurance companies that have a similar problem, right? They've got float and they need capital preservation, and they sort of show up as senior lenders. They buy sort of over collateralized securities and banks do direct loans. And I don't know. I think there's a good argument to be made that maybe banks should increasingly become financiers instead of direct lenders in this economy or something.

So I don't know if that's me now just riffing about what are the different sort of scenarios are. But if you wanted banks to be reengaged as direct lenders, I think you probably have to allow them to take a bit more risk in ways that I think are very hard for them to do today.

Got it. Got it. Wanted to talk about other parts of the offering and one of the things that Upstart has tried to do is be innovative in its offering to the customer, et cetera, and maybe speaking a little bit to what you're saying is that allowing maybe the traditional banks should be financiers and allow you to kind of figure out like what customers are going to be most responsive to.

You've talked about experimenting with subscription offerings and revolving products. What would that look like in your mind? And do you view them as an opportunity to compensate for the low level of frequency that we tend to see in personal loans? Like how do you think about like those as relationship builders?

Yes. We've talked a little bit about this in the past. I think in a very -- maybe still ethereial way, I don't have any concrete announcements on that. But I think in a larger scale perspective, if you think about the credit landscape for consumers, the 2 big areas we've not really sort of gone after yet are revolving credit and purchase mortgage. And I think those will be interesting areas for us to explore at some point. What we like about revolving credit is what you said, which is there is a much more engaged relationship.

And today, most of our products are quite transactional. And I think there's a lot of advantage in having a more recurring dialogue, both from an acquisition and sort of engagement standpoint, but also an underwriting standpoint, frankly. And so I think those features are such that we'll have thoughts or plans on those segments at some point in the future.

Got it. I appreciate that. So I want to talk on a couple of things where we get a lot of questions, especially as it relates to thinking about the macro environment and the sensitivity of Upstart generally. In the past, you've talked about 35% or just over 1/3 of your borrower base has a student loan debt. And you think that around 2% to 3% of this group is in some form, noncurrent. What's -- how much confidence do you have on the mix that the mix of noncurrent won't step higher? And if it does, how should we be thinking about the impact to your business?

Well, the mix itself, I guess, just a function of what we're originating, what's coming into our funnel. So we are aware of it. I think the more interesting question is what is the risk of that segment. Today, as we watch it, we've not seen any real difference between their performance and the control. There's a thesis that it might change with the change of the moratorium. But I think there's also scenarios in which it doesn't change.

And this is an example of something that's like if something changes in that segment, our models will react to it quite quickly. It's aware of these facts. Every time we underwrite a loan, the model is aware of what their status is on their trade lines and on their student loans in particular. One question might be, well, why don't you anticipate it and start pricing them as being riskier starting today? And with things that have to do with the macro, and this may be a little bit counterintuitive, but like I think we probably have a 50-50 hit rate in guessing.

Like here's the fact about the student loan moratorium. When it was first enacted, it didn't help credit performance at all, okay? So people didn't get a break on their student loans and they start paying their credit cards, they probably bought an iPhone or something, okay? So it didn't help credit performance. Now the question is, will it hurt credit performance to take off the moratorium? Is that money going to come from excess consumption? Or is it going to come from the installment loans that they're now not going to pay? I think there's a 50-50 there.

And by the way, when the moratorium was first ended, which was during the Biden administration, we got the same question as now, like why don't you start pricing these guys as being riskier? Well, it turns out the Biden administration said, yes, well, we don't care if the moratorium is over. We're not going to enforce it. So there's like trying to guess what's going to happen in the macro is a bit fraught. I think you're probably as right off and as you are wrong.

And to us, the better answer is unless it's a very large exposure, something on the order of 1% or 2% or 3% of our book is in the rough bucket of macro things that could go one way or the other. And we want to react to them as quickly as possible. But if I were to definitively say today, although that's going to be a riskier population, it's not clear. [indiscernible]. And if you're wrong in the other direction, you start getting adversely selected.

So there is sort of the risk is fraught. But if it's something on this level, there's a whole basket of macro things that I think are at any given time being juggled and they sort of fall into this bucket of like, okay, are we going to react to them appropriately as they happen.

Got it. Last couple of minutes here. I want to go back and compare where we're headed to history again. Let's talk about the specific factors that you think we would need to see align for you to return to the types of origination volumes that we saw in 2022? And how much of that depends on ongoing model improvements that you can manage and control versus macro conditions?

It's a good question. Both of those are paths. They have different timings. If the macro sort of stays roughly as it is, and in particular, if that default index is like 1.5%, we will get there over time through model improvements, maybe 2% to 3% at a time. We're probably at half of what we were in early 2022. So I don't know. It will take us a year or 2 to slowly work our way back there. If that macro were to drop, we would get there much more quickly, right? Tomorrow, default risk in the environment subsided for all borrowers, it would be an immediate tailwind.

And to be clear, like your perception of macro and default risk or as you're measuring it better said, is that defaults right now are about 50% above where they were pre-COVID.

We're still in a very high default.

Very high default environment, et cetera. Well, that's great. Really appreciate you spending time with us today. It's a really interesting conversation. I think that it's something that everybody aspires to. But what I find most compelling a lot of times with Upstart is that you're not only taking on the hard challenge of underwriting, but we've heard a lot of banks talk about, oh, we're going to use AI to improve things like customer interaction and automation, et cetera, and you clearly are already well along that path. So I look forward to seeing how it develops on a go-forward basis. Appreciate it.

All right. Thanks for having me.

Thank you very much.

Thank you, everyone, for joining. It is the Upstart session. So welcome to the Upstart session. I'm Mihir Bhatia. I cover consumer finance and payment companies here at Bank of America Research. For your tech investors, you can vote -- also vote in the consumer finance or the payments categories in II. So please vote for us.

Before we get started, I do have to read some of these disclosures. So today's discussion may contain forward-looking statements that relate to future results and events which are based on Upstart's information available as of today and are subject to risks and uncertainties. Actual results may differ materially from these forward-looking statements. The discussion may also include non-GAAP financial measures which are not a substitute for the GAAP results. Please refer to the company's filings with the SEC and its IR website for additional information, including GAAP to non-GAAP reconciliations, along with other disclosures.

So with that out of the way, it's my privilege to be hosting Sanjay and Paul. Sanjay is the Chief Financial Officer at Upstart, Paul is the Chief Technology Officer. So thank you both for joining us. Really appreciate you guys taking the time to come in today.

Thanks for having us.

So now usually, when we have these discussions, we start the macro backdrop and then start getting down to company level questions. But you guys just hosted at AI Investor Day. We are at technology conference. So going to try to do things a little different, and we'll start with a couple of questions around AI and Upstart. And like the first question -- the first thing I think we should acknowledge is that Upstart has been talking about AI much longer than the last 24 months when it has become much more of a buzzword, right? You've been using it, machine learning techniques, which -- some -- and AI to a certain degree for many years?

The question really is, how do you distinguish yourself or your models from other lenders, right? When I talk to other lenders, they tell me, "Hey, we use AI too. We have machine learning techniques, we've been using. And like we're talking large sophisticated companies, the capital ones of the world, the discovers, large companies that have been using these for a while. So what is like different about the way Upstart uses AI and what's the sustainable advantage?

Yes. Yes. I mean we get this question all the time, we've been really getting it set started the company all the way, all the way back into the 2012, when we first got started in the end out. Maybe I'll start sort of from the zoomed out answer and then I'll give some more specific details. But whenever I talk to my team, I think one thing I'd like to say is that ultimately, in the long run, the only actual competitive differentiator is speed and I believe that very deeply in the sense that there's something fundamentally that would stop any of the player existing word view from attempting to build what Upstart has built. Anybody could pick up and decide to start. Now in order to actually start, they would need to do a few things.

The first is that they would have to invest in a fairly expensive technology team composed of machine learning researchers. We have a team today about 70 machine learning researchers that you would need to build a team like that, that can build this kind of technology. We need to empower that team to be able to actually do work that gets to production. This is something that turned out to be quite challenging at legacy financial institutions that have very particular kinds of risk controls, risk committees.

And then you would need to start gathering alternative data, which is something that has been pretty important to us. You would need to gather alternative data about actual borrowers that you could connect with prepayment outcomes on consumer loans, which typically have a life cycle of multiple years. So you have sort of this multiyear data gathering process. And then you need to actually build the actual machine learning algorithms, now that you've got the team, you got the data, you need to sort of build models that are capable of making use of these things. And it turns out that a lot of the work that is done, it's not like you just I mean we just take sort of off-the-shelf algorithms that are open source and apply that to the problem you have to do a whole bunch of modification to make them work sort of specific case of loans, which are lot a bunch of technical properties that are quite different than sort of other machine learning domains like image classification or text prediction or what have you.

And then basically, you need to sustain this effort through: one, regulatory pressure; and two, the sort of very significant likelihood that your early models will very significantly underperform when you don't yet have enough training data, and so you sort of incur a fairly significant amount of financial and reputational and regulatory risk in order to have possibly the outcome that 5 to 10 years from now you will have a model that will be sort of in the neighborhood of Upstart's.

And I do think people could decide to go and do that. We haven't, in our time, seen a lot of activity in that direction. But in theory, I think anybody could decide to start. And of course then, the question is like, would we still sort of differentiate the enterprise value in that role or let's say, 5 people decide to start doing this today is like, well, what will we accomplish in those 5 to 10 years that they're pursuing the work that we did over the last 10. And the answer to that, of course, is just like, well, what does our team get done during that time? What's our current rate of progress.

And the thing that has been really consistent throughout our history is that there's been essentially no diminishing of the sort of returns to research investment. So we've basically been able to continuously compound sort of model accuracy wins, sort of quarter-by-quarter over our history. And so we've been in to make the models better by continuing to sort of be innovative on the algorithm design getting more columns of variables, getting more rows of data just as repayments come in, which unlock for complexity in the models and then investing more recently in the sort of the compute speed and memory layer that unlocks the ability to support larger bots.

One of the things you mentioned was about -- is Paul's mic on?

Yes. It's on.

Is that coming through the PA? Yes. There we go.

One of the things you mentioned, Paul, was your continuous improvement in the model. I think you've talked on calls about 2% to 3% a month. I understand it's not linear, but talk a little bit about -- more about that. How does the model improve? What are you doing every month to make this model improve?

Yes. So the areas of improvement are divided into the 4 categories I referenced earlier. The most common one is investment in model architecture. So this is essentially a work in algorithm design where we might introduce different types algorithms into the ensemble so like a different kind of neural network or sort of a different way that the models get ensembled together or a different way that we modified the loss function. So all of the sort of inner workings of machine learning algorithm. And this is where probably the plurality of our research time goes, and we've gotten more wins from this area than any other?

The second is just getting new types of data about consumers. There are a lot of third-party data vendors. There are a lot of ways to collect data directly from the consumer, and we're always experimenting with new ones and seeing which ones add incremental value on top of the existing ones. And historically, every couple of quarters, we found something new that sticks.

The third bucket is investments in compute and memory. This basically unlocks the ability for us to support larger, more complex models in production without running into the latency constraints. And if you know, your model runs too slowly and the consumer leaves because they're tired of waiting or the training cost is too high to be able to regularly retrain this model.

And then the last thing that kind of just happens in the background is getting more rows of repayment data. Rows of repayment data basically are sort of the hard constraint on how many variables you can have in the model and how sort of fancy your algorithm design can get -- and as you get more repayment data, you basically can just like unlock more variables or more complexity in algorithm design.

Another area we've talked previously before about model improvements has been calibration and the macro calibration. I think one of the statistics mentioned that the AI was a 55% of excess defaults that you experienced in 2022, you would have experienced if you had a similar event. Now talk a little bit about why that is the case? What gives you confidence? Obviously, the next macro crisis might be different. It's not going to look like 2021, hopefully. We don't have a pandemic and similar stimulus. So what gives you confidence that the model is much more macro resilient today?

Yes. Yes, so this one is actually pretty straightforward, unlike a lot of other things that we do. So historically, we basically didn't care about macro at all in our models. We viewed our models as having one job, and that was just to separate risk, rank risk, figure out which borrowers are relatively more or less likely to pay back their loans than others. And what that meant was that the models are essentially implicitly calibrated to the entire period of observed training data, which you can think of as the past 10 years or so.

And after the events of 2022, we realized that was actually quite valuable and important for our models to care about the sort of current macro conditions and not just the average macro condition for the last 10 years. So we introduced a number of things into the model that had the effect of making the model time aware. By the time of where I mean it became aware of what calendar months any given payment due date was in and it was able to isolate out from that, first, the sort of aggregate macro effect, which we've subsequently published as the UMI or Upstart Macro Index, which basically is just a statistical measure of the sort of likelihood of defaulting when your payment is due in a particular calendar month when you hold all the other characteristics of your borrower population constant.

And then the second thing we're able to do is let the model dynamically interact that time variable with any of the other borrower level characteristics. That means that maybe there is an aggregate relationship, but maybe there's a sort of different aggregate relationship in just borrowers with higher or lower credit scores or white collar versus blue collar workers or people with more or less types of education or maybe people in a particular industry like nursing or people who study computer science. And if there's a particular sort of sector-specific shock, the model is able to dynamically pick that up.

And so to your question of like why do we think this is sort of something that will work in the next macro shock, which is certainly going to be different than the last one. The answer is that there's nothing we did here that was specific to the 2022 macro shock at all. There's actually nothing sort of -- there's no kind of like economic macro forecasting that's going into this model. It's sort of just a fully general machine learning solution that just as its core innovation is just making time and operable feature in the model that can interact with any borrower level characteristic.

And so this will work as long as it's the case that the next shock happens in a way that is related to any of the thousands of borrower level characteristics that we observed, which is really very broad set of things. So it could be happening for certain occupations or certain skill sets or certain sort of socioeconomic strata, certain education levels, like all of that stuff would work.

Now obviously, of course, in theory, it could happen to in some way that isn't reflected. But I think for the vast majority of the things that people talk about when they think about macro shocks, those are things that are reflected in the borrower level variables. Now what that would mean, assuming that plays out, as I described, is that it doesn't mean that there's no effect to the business. The business will still be impacted because we won't be able to approve as many people for loans, but the credit performance will be in a much better place because it means that new loans that get made will very rapidly be issued with a model that is properly calibrated to the new time environment that's observed.

And I think it used to be 8 quarters it took to adjust the model. How much would you say it is now? Is it immediate? Is it?

Yes. So in the actual history of the sort of 2022 shock or the end of stimulus shock, it took 8 quarters for new loans getting issued to be properly calibrated to the new environment. And that was just because, again, the model was basically training on the whole sort of history of what we'd ever observed for the past 10 years, it was calibrating to that. Whereas with the sort of new setup the model is able to start picking up changes within the sort of very first quarter, that's why there's -- that we have that fact going about you get half the loss reduction or half the excess loss reduction almost immediately. And then within just a couple of quarters, you basically get to fully calibrated. So instead of like 8 quarters, you can think of it as being 2. And even within those 2, you're already cutting the excess losses very materially right away.

Got it. Now Sanjay, maybe turning to you. We talked about the underwriting model improving, becoming much more resilient. Where do you see that impacting the business today on the financials? I understand in a future recession, future macro turbulence, model will adjust fast. But do you feel the effects of that in the business today, whether it's from borrowers being more willing or loan funders or loan buyers being more open to buying an Upstart loan, whether it's moving up, just in general, are you seeing any positive impacts from that in the business today?

You're asking specifically what Paul just said?

Yes, just including the macro calibration model improvements?

I mean everything in our business over time flows from credit performance. If you cared about one thing, if there's 1 KPI, I had to track, it would be the credit performance. That gives confidence to the capital markets and the investor side of the equation over time. It allows us to more confidently sort of approve and underwrite the borrower side. So it will -- I mean, immediately show up as capital resilience, if you will. But the whole business model we have is predicated on us being smart at credit, and this is a huge upgrade in our ability to do that. And so to me, it's hard to -- like there's not a line item on the P&L where it shows up.

Right. No. And that's why, like I guess, maybe lift the hood a little bit, how do these forward flow agreements work? Like big private firm comes to you and they like how much testing do they do? How much do they actually dig in on your historical underwriting or what the underwriting model? Like how does that actually work? What is that process like? Where are you seeing like the benefit coming through?

Extensively, is the answer. You can imagine that these folks who are there -- I mean they're in the business of credit and consumer credit, and they go very deep in their diligence. They kick a lot of tires. Sometimes, they'll even before getting to what you described as a forward flow agreement, which is essentially a commitment to purchase sort of flows of loans in the future. First, they'll buy an existing portfolio and just watch it, get sort of acquainted with it. There's a lot of tire kicking on us as a team.

And it's one thing to sort of take a point of view on how the credit looks but when you're making a forward commitment that's on the time horizon that we're talking about, you have to worry about how your counterparty is going to show up when you get into economic stress. And so a lot of that is just evaluating the counterparty and just making sure that's we want to be in a partnership with. So that whole process is expensive.

I think things along the lines of what Paul described are helpful. Ultimately, the proof needs to be in the pudding in credit. Like it's not like some of the equity style investing, particularly at the venture stage, is a bit more of an art, in credit, it's not. In credit, there's like deep science. And we could talk about what Paul talked about thematically, and it would be interesting to them, but they would want to see how that looks in the next cycle. And they'll put up the appropriate guardrails between now and then to ensure that they are getting into something that is controlled risk for them.

Maybe switching gears a little bit. One of the changes at Upstart maybe compared to 2022 other than just the credit model of changing is the diversity of products. Talk a little bit about that, the expansion more fully into auto, HELOC, Prime. What's driving that? What are you seeing? Like is that probably your partners, the funding partners? Is that like more driven by them and the demand from them? What made you choose to do those products? And why do those?

Yes, I can start. I think ultimately, our product strategy is just designed around like what do borrowers need. And the first product we came to market with the sort of unsecured personal loan was really like the product where it was sort of blindingly obvious that there was a hole in the market, especially in the sort of like nontraditional prime sector where there was many very significantly under approved and mispriced borrowers.

And I think that is, to some extent, true in some of these adjacent markets and motivated are entered into some of them. We certainly think that there are a lot more people that you can approve for good bank quality credit in the auto world on both the refi and the purchase side. We think to some extent, that's true in HELOC in some other categories. So that's sort of like the first kind of motivating reason we get into a new product.

The second one is that in some of these areas, maybe there is sort of only a limited amount of mispricing of the credit, but there's a significant opportunity for a reduction of the sort of process required to get the credit. I think there always being the sort of frontier of trade-offs between the risk you can underwrite to and the process that you put the borrower through sort of at the limit. If you put someone through enough diligence even with no AI of any kind, you could get your risk probably to 0, but no one would sign up for a process that looks like that.

And on the other extreme, even with the smartest AI, if you didn't collect a single bit of data about the person while you would have very high losses. And so it's always about sort of getting to a higher frontier on that trade-off curve. And so in some products like in sort of maybe the prime HELOC category, for example, there's more of an opportunity to reduce the sort of process of getting credit, holding risk constant.

And then finally, the last thing I would say is we have also been increasingly motivated by the recognition that our borrowers are going to be borrowers not just sort of for one product but over their life cycles for multiple products, and that's increasingly played out in a way where we see that. Once someone's in the ecosystem and they've had a really good experience getting an Upstart loan while when we have more data about them than anybody else has, we have sort of a competitive advantage in underwriting them correctly for all sorts of different credit products. And so we want to be able to serve them over their whole sort of consumer credit life cycle, and that means offering the full coverage of products.

The one thing I'll add, maybe at a higher level here is that, so as Paul said, we've always believed these would be obvious opportunities of expansion from our core modeling technology. Back in '22-ish, when the world not really stressed for us, at that time, we had sort of 4 little bets in incubation. We had -- we had auto lending at the sort of early instance of home lending, which is a HELOC for us now. We had small business lending at the very sort of small proprietor end and this small dollar short-term duration loan. And when stuff got really stressed and like many tech companies in the valley are going through rifts and layoffs, we had a decision to make. It's like how much do we protect versus sacrifice these things that were -- they were taking resources at that time. And we made the call to essentially sacrifice one of them but protect three of them. The one we sort of put on ice was the small -- it was the small business lending.

But we've protected these other 3, and we went through the pain -- we reduced our fixed cost base. We increased our margins and our take rates. We hunkered down. But for the last couple of years, we've been quietly incubating these bets. With the idea that like one day, we would be happy, we would be thankful that we kept them. And I think we are getting to the point now, maybe between now and the end of this year or coming into early next where we're going to really start to see the fruits of that decision because they all look like they're emerging now, and they are, they all have great men's and we're very happy about the trajectories they're on. And the default environment is such that they're now, I think, ready to sort of scale. But it didn't come out of nowhere. It's because we've been quietly working on this and sort of tooling it under the hood for the last couple of years. And so I think that's a very exciting time from the perspective of these new products.

So at the risk of asking, which is your favorite child, which of these three is your favorite. Well the most excited about from just an opportunity standpoint, maybe.

I mean we have debates on this at our leadership. I don't know, I'll let you go first, Paul.

I mean they're really different, and they serve different parts of the market. And I think for each of us on the leadership team, there's like maybe like different versions of the sort of credit problem that are the most exciting to us. We've talked a lot about our expansion into the sort of prime audience. I think if you look at from that and sort of the fact that we can become relevant to sort of essentially the entire spectrum of the U.S. population certainly, I think something like HELOC is sort of the most like complementary to our existing product from that perspective where we used to be a solution for kind of people who didn't really have too many net assets. And the HELOC, of course, is the very opposite, sort of end of the spectrum. And so having a great product there is something that we're really excited about.

We really do care about getting past the point where marketing needs to be so highly targeted that it's like if you're in this particular sort of cross-section of the population, Upstart is the best for you to get into a place where it's just generally true that if you need credit, then Upstart is the best place for you. And so I think HELOC is a big step for us in that direction. I think at the opposite end of the extreme, the small dollar product that we have is just the one that's sort of like goes most directly to sort of, in some sense, like the sort of original reason we started this company, which was like the observation that there are a lot of people who are very underserved from a credit standpoint.

And having like 5-year, $10,000 personal loans is good, but it's not really the sort of true marginal consumer of credit, and that consumer needs something like the small dollar product and being able to get places to get to a place where approval rates can be super, super high and people can just be very confident that if I need credit, I go to Upstart, I'm going to get credit, like the small dollar moves us in that direction, which is, in some ways, like opposite to what the HELOC thing does, but it's sort of very exciting there.

And then like auto is sort of -- has its own case for it, which is that I'd say of all the different credit products, I think auto is, in some sense, the one that is most sort of universally needed and most universally used words like there's only certain types of people that would ever need a HELOC or a certain type of person that would ever need like a small dollar or a personal loan, but essentially no matter who you are and auto loan is useful for you, whether you're a super prime person or you're like not at all a prime person.

I'm pretty much auto is like it's actually a thing that like is useful for your life, it's not sort of like at risk for kind of just frivolous and it's relevant to everyone across the whole population.

That was a very fulsome answer. I'll just say in a very summarized way, I think the small dollar product is our most strategic bet. I think the auto product is our most audacious bet in what we're trying to do. And if we pull it off, what it could be.

Okay. And we want to dig into that, but we only have 5 minutes left. So I'm going to switch a little bit to the year and now maybe 1 of the -- your guidance for the year is profitability in the back half of the year.

Yes.

Walk us through what needs to go right, the pieces, how do you get there? What are the pieces that need to change? Is it -- are you just on a glide part there now? Or is there some risk to that?

It's very simple. There is some risk to it, but the core assumptions are basically 2: one, the macro as we experience it through the default risk levels, will roughly stay static. And two, we will continue to execute on the model -- the types of model improvements Paul described at roughly our historical clip of improvement. And if those 2 things happen, our fixed cost base is very steady. Our margin profile is resilient. So it's really just those 2 things. Now in that, there is execution risk. Maybe we won't drive the technology improvements that we historically have or think we will. I like that risk because it's in our control, and we're good at that. The other risk is that the macro environment is not static, it could get worse, and that would be a headwind. But it's really down to those 2 things?

And then maybe another question we get a lot from investors on the co-investment model on these forward flow agreements. What kind of guarantees or what kind of risk are you taking at Upstart, like are you guaranteeing a certain level of performance for the -- in the forward flow agreements, like if the loans underperform, does it fall back to you? How much of this risk is there on versus off balance sheet? What are you comfortable with on that?

Yes. I mean it's not really a guarantee per se. We do recognize in the context of these agreements with our counterparties that are meant to spend cycles. There will be good vintages and maybe some underperforming vintages because the types of shocks we've talked about, which our models will react very quickly to but not instantaneously too. And the equation that we need to get right over that duration is that in benign periods, these vintages are meant to overperform, that overperformance, we will harvest, Upstart will as the counterparty.

We will put that in the vault, and where there's underperformance there's a macro shock and there's a few bad vintages. The over performance needs to be sufficient to pay for the underperformance. In that sense, it's a bit of a -- we're creating a bit of a macro insurance layer. Now if the underperformance is extreme or it's higher than the amount of over performance we were able to harvest some of that we are at risk of. It's on balance sheet. I think we disclosed it in our investor materials very clearly. There's a certain aggregate amount that's theoretically at risk, and we have a certain valuation of it. Right now, we value at it slightly above par, if you will. And so that's the variable we manage from a risk management perspective.

Right. From your perspective, like is there a target? Is there like some kind of ratio, something you're looking at in terms of what that dollar amount is?

Yes, it's sort of a low to mid-single digit percent of the overall origination volume that we do through those deals. So just for sake of argument, take like a 5% number. That would be our sort of basis in the risk.

Got it. So we have a minute left in case anyone has a question. Go ahead.

Your [indiscernible] perspective, I think as a [indiscernible] investor, we disconnected lions about economy versus the power data and the harbor is stop beta, right? , et cetera, you feel elaborating some we basically convert back up? Or do we have a kind of cash down? Or how do you feel about that kind of a expected or is the progress in this traditional?

So you're asking about the sort of the sentiments about the economy versus the hard data on the economy and how they reconcile? Yes. Some of this is down to averages versus distribution, and some of this is down to what the sort of the mainstream narrative cares about with respect to macro. And so what I mean by that is, I mean, the 2 most salient macro things that are talked about our GDP, which is essentially consumption and production in the economy and unemployment, obviously.

GDP has been remarkably resilient. It's been like since through the -- since the period of the stimulus to today, we've always celebrated the strength of the American consumer. And that I think is evidence that the -- or the economy in aggregate is strong. And of course, the labor market itself is extremely resilient as well, and that's sort of data that suggests that the economy is strong.

But I think that if you look at it in a more nuanced way, that the GDP that is sort of materializing and the consumption that it's underpinning it, you can ask the question how affordable it is, okay? So normally, we had a certain level of consumption in GDP, and we were putting away 7% to 9% of our national income. That was a savings rate.

And today, we are spending higher amounts than before. GDP has grown, but we are barely putting any money away. On average, it's about, I don't know, 2%, 3%, 4%. And -- but again, that's an average. I think if you would look at the distribution of that number, there's a significant fraction of this country that's barely getting through the month, right? And they have -- there's an increasing reliance on these cash flow products to get through the month -- so you have this we disconnect we're like, we're all spending a lot of money. Arguably, we can afford it. Clearly, there's no money for a significant fraction of Americans going into the bank at the end of the month. And are they discounted about that? Sure.

And then you could say, well, why don't we spend less? And I don't know the answer to that question. We seem to -- our spending habits got turbocharged with the stimulus, and they never got unwound. And maybe some of that is at the heart of the disconnect. We've gotten used to a certain way of living. We can't really afford it. All the base numbers are good, but I don't think people feel like they're on -- there's much of a safety net around them. And the labor market does remain extremely resilient.

And frankly, we don't see that really changing anytime soon because from a structural perspective, like most Western countries and particularly the Asian ones that are leading the charge on this we're sort of running out of workers every generation. And we reached full employment economy back in 2019. And it was interesting that nobody ever really talked about it because we then went into the pandemic. But I think we have, on the 1 hand, an extremely resilient labor market. But it's not producing for the -- for some large percentile of Americans, the types of incomes that will both support the consumption habits that have materialized and the savings sort of practices that we've typically been used to and that sort of creates security for us. And so I think the result of that is you've got a lot of consumption and a little unemployment, but very little financial security and a lot of people who feel like they're going backwards?

All right. So with that, we're at time. I only got through about half of the question. So we'll have to have you back here next year to continue. But thank you so much for joining us?

Great.

Thank you.