Ambarella, Inc. Aktie

Welcome to Day 2 of Citi TMT Conference. My name is Kelsey, one of the analyst here at Citi covering U.S. semis. We are very excited here today to have Fermi Wang, CEO of Ambarella, here with us to discuss more about the company and to share the outlook.

Perhaps to level set, Fermi, could you give us an overview of the company's heritage portfolio and also what opportunities you are the most excited here today?

Right. Thank you, Kelsey. And first of all, thank you for joining us today. Ambarella is a fabless semiconductor company started in 2004. And throughout the years, we have been starting -- focusing on digital video application now in the last 10 years, really focused on building solutions for edge AI processing. Basically, what we are saying is that we're building a SoC that integrates not only just an inference engine, but also image processing pipeline, CPUs and peripheral so that we can become the SoC for the whole system.

And there are so many different verticals we are targeting at. Anything related to the edge device, which require AI inference, we are basically in there. The biggest markets like enterprise security to the portable video, to the telematics for automotive. And now we are talking about robotic application. In the latest earnings call, we talked about how to penetrate edge infrastructure with new partners like Capgemini and Macnica.

So basically, we are really focusing on edge AI for the last 10 years. And hopefully, the market will start taking off as people expected. But definitely, that's really what we are focusing on.

Right. On that front, what are you seeing today that gives you that confidence that we are sort of reaching inflection point for edge AI? I mean, we have seen a lot of news flow on acquisitions like big analog companies acquiring smaller edge-related companies. So could you talk about that a little as well?

Right. So first of all, I think our confidence is based on our engagement with our customers. In fact, all of the customers we talk to, they all want to put more and more AI performance into the edge device. It's now -- in 10 years ago, we start with CNN type of neural network. Now they moved to transformer, then move to large language model. All of that are available in our silicon hardware software platform.

So -- and one thing that becomes very obvious in the last 12, 24 months is everybody want to continue to increase the AI performance on their device, mainly because of the larger language model. So however, edge AI or physical AI, people have been really talked about a lot about, but the biggest problem for this market today is there is no one vertical like coding for data centers, right? Today, majority data center revenues is generated by one big application, but there's no such application on at edge.

So that's where our strategy always focused on multiple different possible verticals and hopefully that one of them will grow. You talk about the acquisition. This ADI just announced a deal, which we acquired a start-up company for $1.5 billion, which has little revenue and their AI performance -- by the way, the AI performance we offer at Ambarella is in the family between 1 TOP to 1,000 TOP. And based on the public information I can find for that company is really about sub 1 TOP to 1 TOP type of performance. So yes, they are edge AI company by there's a little bit overlap.

But -- and we are generating $400 million revenue versus some. So I really think that's indication that edge AI is happening and a lot of the companies said, I need a road map and they try to find a way to -- by using acquisitions to fill the gap.

Right. On that front, could you characterize Ambarella's key differentiation?

Obviously. So every time we try to bid on a design win, the biggest differentiation for us is always performance per watt. That performance can be AI inference performance per watt or video processing performance per watt or other related functions per watt, right? So at the end, it's power efficiency. We understand that, in fact, most of our customers when they build their first-generation product, they use a GPU to do it because it's easy to program, widely available to do that.

But when some of the applications they are using is really run into the barriers to power consumption because some of them just require battery life or using a very limited power dissipation solution. So from that point of view, power efficiency is the most important differentiation for us. But the second one is really that because of heritage is on digital video application, our digital video processing technology is better than I think most of our competitors -- maybe all the competitors out there that -- that's another huge differentiation.

But the third, I think a lot of people ignore it is really -- if you look at our product portfolio, we have 15 AI SoCs that we offer to our customer. But on top of that, there's a unified SDK called Cooper that cover the whole product line. Basically, if you design one product based on our chip using Cooper, you can easily move the cell application to different silicon without too much modification. And that software architecture, I think, is critically important for our customer who has multiple product line to build and they don't want to reinvent the software from one product to another.

Got it. How should we think about -- as we compare NVIDIA, they have their related Qualcomm, NVIDIA, Synaptics related offerings. How should we think about you relative to them?

So you're talking about 3 types of competitors. NVIDIA, obviously, they can deliver really high performance. That's their strength. But like I said, GPUs power consumption has been well-documented problem, and we think that if you -- suppose our customers who really want to have efficient -- power-efficient solution, GPU is not the best solution. So from that point of view, kind of we carve out that based on power efficiency.

Qualcomm is a little bit different. Qualcomm is definitely more efficient than the GPU. But however, when they compare to our approach, we still have a significant power efficiency. But on top of that, we continue -- but because they are closer, so we definitely have a lot of overlap with Qualcomm. So we always view Qualcomm as the biggest competitor to us.

Synaptics is different. Synaptics is just -- they came in to play this edge AI 2 years ago with -- by acquiring product line. And their coverage is really from 1 TOP to maybe 10 TOPs type of application. So yes, there is some overlap, but it's not a lot. So from that point of view, we don't run into Synaptics that much. But when we go out to compete, we're going to NVIDIA and Qualcomm all the time.

So the IoT segment currently is roughly about 75%, 78% of total revenue today. Consumer -- sorry, security and consumer video, roughly about 45%, wearables, 10%. Could you walk us through the key demand drivers for each of these end markets?

Enterprise security and the...

Yes. So consumer video...

Well, enterprise security is really our largest application that we've been driving for is really for the security camera in the past. But I think that market start changing. The way it changes is that a lot of security camera become serving different purpose. For example, when we talk to edge infrastructure companies, one -- I just give you an example that, for example, in the Starbucks, you have many camera serve as a security camera in the past. Now people say, I want to link, connect all the security camera to a box. And in that box, you run LLM so that now that box can tell you how many people come in, how many people stay there, how many online, how many -- and how much time each one of them spend in the store, what they buy and how they pay for it, whether the cafe shop is clean.

So suddenly, LLM can start providing customer marketing data as well as operation data. So that -- but you're using the same security camera hardware. So from that point of view, I think that the driver in the past was purely security. Today, we are start seeing a lot more marketing-driven solution for that. That's one. On the consumer video side, in the past, after iPhone was introduced, cell phone basically replaced all of the still cameras. But now there's a new trend, right? In fact, people building really high-end video capture device, handheld stabilized handheld device and start selling at a very expensive price, but the volume is at the order of 10 million units a year.

So that's -- those high-end video capture device become a new category that we are addressing. That category including all the 360-degree camera and high-end sports camera, pocket, handheld stabilized camera, drone camera, they all package into that. That market by itself, we estimate anywhere between 10 million to 20 million. So -- and ASP is pretty high because not only you need high-end video processing, but also you need to be able to run AI, a lot of AI on that. So from that point of view, it becomes a sweet spot for us. So that's another driver for the consumer side.

And wearables is about 10%. Is there any growth driver there?

Well, in fact, in the past, wearable is really for police, law enforcement. Now we are seeing a lot more -- for example, a lot of our customers are talking to us that the retail stores, all the people in that store wearing the wearable camera, start documenting the conversation and providing actual angles of security. So wearable camera go way beyond law enforcement and we are starting seeing that.

And right now, that market is really limited by power, how much power you can -- power consumption can be made available in that camera and how long your -- the life, the battery life, you can have those. So that's really played to our strength. However, I rethink we are still at the early stage of this wearable because I think when wearable camera become really expand to all of the service category that everybody should wear one so that they can start collecting -- became not only for security camera, but like collecting the marketing data, the purchasing habit, all of the customer data can be also become quite interesting for the people who use those kind of wearable cameras. So I think wearable camera start from the law enforcement now turning into a marketing tool and will continue to evolve from there.

Is there a way to think about the growth rate for these 3 end markets?

Those markets, obviously, enterprise securities grows at a certain rate because it's very mature. And I think that consumer market is really growing fast in the last few years and wearable definitely grow in a much faster pace. So -- but however, the base is small. And there is also a similar market like robots. Robotic application, people talk about it all the time. But the base is so small, most of the customer there is a small, midsized customer, although the growth rate is high, but the base is low.

But another one that we talked about in this time when we disclose our SAM discussion is really the edge AI infrastructure, which we believe that people will build edge box or edge appliance. In the past, most of them just have a huge powerful CPU in there. In the future, will be CPU plus a lot of AI inference engine. So that will become an interesting market, and we think that's definitely another huge growth area for us.

Could you talk about how that will play out? Is that on our phones? Is that...

No, no, it's -- for example, today, if you go to any retail store, right, if you look at their back end in the store, they have probably a huge box, huge Intel CPU box sitting there for inventory or for the pricing or the cash exchange. And most of people said, okay, I have that CPU box, but I need to add a lot of AI too. That's where we think opportunity. Almost every industrial PC or kind of edge CPU box in the past, when they try to upgrade to the AI -- more AI performance into it, that become opportunity for it.

Okay. Got it. So I know that you guys read had a relationship with Hanwha and signed an LTA deal of about $800 million. Could you share what prompted the move to the LTA deal? And why now?

Right. So in fact, this process is initiated by Hanwha. I think it's -- Hanwha is a huge Korean group that has many, many different subgroups. The subgroup we work closely was called Hanwha Vision, which is their security camera division. In the past, we sell our chip to Hanwha Vision, but Hanwha Vision also developed their own silicon for the edge video product.

And they approach us because as a whole group, Hanwha decide they have to have a unified edge AI strategy. They want to have one hardware, one silicon, one platform for the whole group. And that makes a lot of sense. But however, they try to evaluate who can build, help become a pioneer on that. The requirement is very simple. They have to be very power efficient. They need vision processing. They need a very powerful AI inference engine and they need to access 2-nano, 4-nanometer process node, right?

If you look at the combination of that, Ambarella become one of the very few candidates to talk to. So we talk to them, we quickly come to the conclusion that it's beneficial for both sides to have collaboration. Obviously, there's other business discussions come at the end. But at the end is really that we help Hanwha to build a family of chips that they can use internally and also give them a certain benefit. In return, they're basically helping -- basically give a back and making sure that we can become the dominant supplier -- semiconductor supplier to the edge AI strategy.

Are there any other opportunities where you see these kind of long-term agreements will happen?

Absolutely. In fact, when you look at the data center market today, almost all the large companies are looking for their custom chip. Everyone wants to build their own chip. And because they can afford it, Google, Amazon, they all can build their own chip. But when it come to edge AI market, there are a lot of large corporations. But if they want to do a custom chip, they will find out that they -- the economy doesn't work out for them. Because the size is not as big data center, the price is not as big.

So they -- the best case is if they want to build a custom chip themselves, they found out that to build 2-nanometer chip is too expensive, they cannot justify. So this is why the semi-custom chip business model comes in. We believe that we can help those large corporation build a chip they can use. But in exchange, we can sell the same chip to other market that they don't care.

So just I believe that just like in the data center, many large corporations want to control their silicon road map, that apply to edge AI market also. When edge AI market become bigger, I expect more and more large corporations will say, how can I control my own silicon road map. But then I think that will add to our opportunity to grow in this business.

Got it. I'll pause there for -- and open it up to the floor if you have any question.

So your SoC platform clearly in the competition, especially on the energy perspective. Have you received any validations from automotive OEMs specifically like your software layer on top of that?

Well, we don't have any design win we can announce. But however, we've gone through many cycles of evaluation. So the -- from the auto OEMs point of view, we have announced Level 4, Level 5 truck deal like Aurora. And also, we announced automotive OEM -- sorry, Tier 1 like Conti and Bosch, they are partnering with us. We obviously continue to talk to potential partners in that. So that's on the autonomous driving side.

But our total revenue -- our 30% of total revenue comes from automotive. That's come from other areas like telematic -- for commercial vehicle telematics, several large companies in the telematic is using our solution today. And also that electronic mirrors, look at ADAS, DVRs, we have significant design win. So in fact, Q2, we have a record automotive revenue. But all of that has come from the one side of this autonomous driving, we still need to get more design wins from big OEMs.

Any other questions? Yes.

[indiscernible]

Between the OEMs -- sorry -- right, so first of all, we start serving this edge AI with our, we call CV2 family chips 10 years ago. We started developing that. That family of chips really focused on relatively simple compared today is CNN-type network, base detection, object detection, those simple tasks. That's where we start. Then starting 5 years ago, transformer type of network come in and large language model happened. That's where you start triggering fast pace of AI performance requirement.

And every time there's a new model coming out, that trigger more AI demand. Just one example, robotics right now talking about using VRM, VLA, now VRX, just look at the performance requirement among those models is tremendously different. And every time the model changes, we constantly engage and say, okay, how are we going to handle this? So from that point of view, different applications definitely have different paths of the AI performance requirement, but all of them is on the path of getting higher and higher. In fact, we talked about X7, which is our accelerator product just in this earnings call.

The reason for that is when we talk to our existing customers say, okay, the LLM is changing so fast. And in the past, every time we build a product, the product exists for the market for 4 to 5 years. Today, nobody believes that for one product defined for one -- this year, 4 years down the road, the same LLM still works. So I think then they become, how are we going to address that problem because the performance requirement is so big. So the easiest way for today is we design a baseline product with one of our SoC chip, and we leave option to add our accelerated chip like X7 into that board when it's required.

So you design in a fundamental hardware and software and just -- and also add option that can increase the performance. And those kind of option was not required in the past. And the only reason for that is really that nobody can predict how fast LLM evolve in the next 5 years. So that's just kind of example how the conversation we have with our customers.

Could you talk a little bit about robotics and humanoids? I think this is a theme that people are just betting on or thinking that's going to be coming. How close are we? Or what are some of the customer conversations?

Right. So first of all, robotic and humanoid is quite different form, right? In fact, here's my personal opinion. I think humanoids from a technology point of view is more difficult than Level 5 autonomous driving car, okay? And Elon Musk promised us in 2015, we will have Level 5 cars in 2016. And in 2026, we still don't have that. I just show you that the best engineering team, the best engineer talent still haven't really resolved the Level 5 autonomous driving car.

I'm not saying humanoid is going to take 10 years to do it. But if that -- if I'm right, if I -- if that problem is harder than an autonomous driving car, it will not happen next year. It will take more time to evolve. So from today, to when the humanoid become available, we're going to see multiple generations of different type of robots performing different functions at different AI performance level.

And just that's an opportunity we should focus on instead of trying to really put our eyes on the home run opportunity. So our approach is we engage with humanoid company. We have a couple of design wins in that category. But the volume is not really -- I won't bet that's going to change materially about our financial performance, but that's really to try to prepare everything on the road map we have both on the hardware and software side. So we won't miss on the opportunity when it happens.

Could you talk about your foundry strategy? You've indicated recently that Samsung capacity is getting tighter. Could you remind us, like how do you work with fabs? Is it all across the board? And how should -- is there any risk to your securing more if demand exceeds?

We have a strong relationship with Samsung fab. And there is one fact I don't know whether people are familiar with. There are so many design houses there, private and public. We are the only one exclusive with Samsung. And that has been one of the things that we think is the right thing for our size of the company because think about this, we are doing 2-nanometer chip, while our first 4-nanometer will be in production. If we stay with TSMC, we won't be able to get much allocation because there are a lot of big guys in front of us.

So staying with Samsung as long as they can give us the technology that we need for our market, we are happy that they can give us more attention and it will get better allocation. From that point of view, I think we are happy with the Samsung's not only the allocation but also support, so I'm not saying we'll get everything we got, but every time the conversation is that if we have a demand, we talk to them early enough, Samsung will find a way to work with that.

So that relationship definitely matters. I think that's just like any other business. When you have a good relationship with your suppliers or your partners, the conversation is easier. I really think this is the time that I hope that we'll take advantage of that because the overall semiconductor supply can become extremely tight, not only foundry, packaging, testing, everywhere is becoming tight. And we hope that the relationship we have with our suppliers that we have been exclusive not only with Samsung but ASEs that with this kind of strategic partnership, we can continue to get a little better treatment than others.

Got it. You've guided to 10% to 15% revenue growth for fiscal '27. How do you think about the growth drivers we've discussed today? What's going to drive upside, downside or the range of outcomes?

Well, upside, we talk about this whole thing, right? It's really about edge AI. The downside is only one, it's memory, right? This is well documented. And the problem is when I talk to our customer, for Q3, we kind of -- we are comfortable with the current forecast with a little bit of uncertainty. But in Q4, when I talk to our customers, they told me that none of the memory supplier can even give them commitment about allocation for November, right? They don't even know how many memory or what's the price they're going to get in November, how can they commit to us. So that's the biggest variable in my opinion, the biggest uncertainty that I think everybody is dealing with and not just us. Everybody is dealing with.

How do you see that as we look out to next year?

I don't think it will end this year. It will definitely go to next year. Who knows how long, but hopefully it won't be too long because otherwise, it's going to eventually impact us.

I believe you have noted that the SoC ASP for a company is about $15 in fiscal '26. How should we expect ASP to fit?

It will continue to go up. In fact, the reason is just we talk about, the AI performance continue to increase. And that means we have to build bigger die and so that we can deliver better AI performance. Therefore, people understand they have to pay more to get AI. So the ASP is going up. It's not that we're going to increase our gross margin. Our gross margin will probably stay the same. It's just that with the -- ASP growth is proportional to the AI performance growth.

Got it. And just to make sure, so if the foundry were to increase pricing, you're able to pass on that to customers?

I have no choice. I know that our customers are going to hate me for that, but we have to. But just like my supply chain just increased the price to me, and I was extremely unhappy with it, but that's just a fact that we have to deal with.

Got it. We haven't touched on automotive yet. That's roughly about 30% of total revenue. So you did mention a little about the drivers, telematics safety. How should we think about the growth prospects? And what are you -- which area are you targeting for content expansion?

We definitely believe the few market that we target at will continue to grow. We're going to grow with the market. So from that point of view, I think our automotive revenue should grow. That also reflect on our SAM discussion. I think the biggest wildcard for us is I still want to get some design win on the automotive OEMs or consumer vehicle on the Level 2+, Level 3 cars. I think that's the biggest wildcard. We have talked about this for years, but we haven't really delivered that. But that is where I think we haven't -- we still continue to focus our current engineering resource on. So hopefully, we can deliver on that.

What is the kind of content step-up that we can expect?

Any consumer car design win, we are talking about a few hundred million dollars per design win. So that is meaningful for our current revenue stream.

So as you move from L2 to L3, is there like a rule of thumb to think about what is the potential content increase?

The content increase for sure. But however, L2+ has volume -- a lot more volume than L3. So L2+ definitely is where the biggest opportunity for us moving forward.

Okay. Got it. And could you provide us with an update on the design wins? You mentioned about $13 billion of automotive opportunities. Is Ambarella better positioned with Western or Chinese OEMs right now?

It's Western. We know that for Chinese OEMs, they want to select domestic solutions. But for us, really that Western OEM is our target. However, we still stay in China. The reason for that is we believe that for the AI applications, China innovate fast, and they have a lot of ideas how to use silicon into -- morph into different product lines. And that we want to stay in China to understand how the things -- how the new application popping up, what kind of technological requirement and that will continue to help us to position our silicon worldwide.

So we're still opportunistic staying in China for that. But also, we believe with current supply chain be so divided that China for China, there will be different supply chain for Western. But however, I think for Chinese customers want to do export business, they also realize they need a totally different supply chain. That's our opportunity. So there are 2 reasons for us to stay in China.

Okay. And in terms of like the near-term growth for auto, which region is it coming from right now?

It's all Western.

All Western.

Yes.

Okay. Got it. Could you walk us through a typical time line from design win to revenue production? Is there a sort of design win from automotive to revenue?

Right. So it's different from application, right? For telematics, those are non-ASIL type application, 12 to 18 months is good enough. For the OEM type of autonomous driving car, usually 3 years, 4 years. However, I really think that becomes a problem for Western OEM. China turning the product line every 18, 24 months, while we still stay at 3, 4 years, that definitely create a problem for us. So I think although I still think all the design win on the Western OEM, we're still looking at 3, 4 years to production. I think as an industry, we need to start considering how to move faster.

Okay. Got it. And just to be clear, the majority of the revenue from auto today is coming from Western.

Right.

And so as we think about -- so you mentioned a little geographical, geopolitical risk, China localization, how is that changing relative to a year ago? Is that getting worse? And are there any areas where you think that you're better positioned to grow with?

Yes. I think it's similar to last year. I don't think it's getting worse or improving. I hope we don't get any worse than right now. But however, I really think that when we do business, the only thing we ask is predictable environment. I hope if they stay today -- like today, it's okay because at least I'm dealing with predictable future, not try to worry about all the changes all the time.

Yes. So based on the conversations you've had so far with investors, what are the top 3 things, Ambarella is often misunderstood, unappreciated?

Well, I think everybody understand we are focusing on edge AI. And the biggest question is, what's one vertical or one application that's going to drive -- be the first one to drive tons of revenue for you. We don't have answer for that. And also, we don't have a time line for how fast the edge AI going to develop. I think everybody is looking for an indication of the timing and the application. But unfortunately, we just don't have a solution for it.

Okay. Got it. Any key message for investors to take away from?

I think it's really that we spent 10 years to build the edge AI applications. And we believe we have a very differentiated technology for edge AI and physical AI. And I also strongly believe when the market starts taking off, we'll be one of the few that are going to stay and take advantage of it.

Thank you. Thank you, Fermi, for joining us.

Thank you very much.

Thank you for standing by, and welcome to the Ambarella's Second Quarter Fiscal Year 2027 Earnings Call. [Operator Instructions] As a reminder, today's program is being recorded.

And now I'd like to introduce your host for today's program, Louis Gerhardy, Vice President, Corporate Development. Please go ahead, sir.

Thank you, Jonathan, and good afternoon. Thank you for joining our second quarter fiscal year 2027 financial results conference call. On the call with me today is Dr. Fermi Wang, President and CEO; and John Young, CFO.

The primary purpose of today's call is to provide you with information regarding the results for our second quarter of fiscal year 2027. The discussion today and the responses to your questions will contain forward-looking statements regarding our projected financial results, financial prospects, market growth and demand for our solutions, among other things.

These statements are based on currently available information and subject to risks, uncertainties and assumptions. Should any of these risks or uncertainties materialize or should our assumptions prove to be incorrect, our actual results could differ materially from these forward-looking statements. We're under no obligation to update these statements. These risks, uncertainties and assumptions as well as other information on potential risk factors that could affect our financial results are more fully described in the documents we file with the SEC.

Access to our second quarter fiscal year 2027 results press release, transcripts, historical results, SEC filings and a replay of today's call can be found on the Investor Relations page of our website. The content of today's call as well as the materials posted on our website are Ambarella's property and cannot be reproduced or transcribed without our prior written consent.

Before starting the call, we hope to see you at one of the following investor events that we have scheduled in our third quarter. First, on September 8, we'll host a DNB Bus Tour at our offices in Santa Clara. September 9, we'll be at Citi's 2026 Global TMT Conference in New York. September 15, we'll participate in Piper Sandler's Growth Frontiers Conference in Nashville. September 16, we will host Sanford Bernstein's 8th Annual West Coast Semiconductor Bus Tour. And during the week of October 4, we will have a European NDR with cities to be determined. Also available to investors during the third fiscal quarter will be our booth and presentations at the AI Infrastructure Summit in Santa Clara on September 15 to 17. We hope to see you there where we will lead the Physical AI track with a number of edge AI and robotics demos in our exhibit area.

Fermi is now going to provide a business update for the quarter. John will review the financial results and outlook, and then the 3 of us are available for your questions.

Fermi?

Thank you, Louis, and good afternoon. Thank you for joining our call today. Driven by a new record level of edge AI revenue, we reported fiscal Q2 revenue slightly above the midpoint of our guidance with non-GAAP EPS of $0.18, and with guidance for seasonal fiscal Q3. By product, we are in the midst of a very steep revenue ramp with our 5-nanometer CV75 and CV72 AI SoCs. And by market, we had sequential growth in both IoT and Auto with automotive revenue driven by commercial vehicles.

The market is increasingly recognizing the strategic value of edge AI as well as our edge AI and Physical AI platform leadership. We continue to make significant progress with the expansion of our edge AI platform leadership, including new go-to-market strategies and engineering and market development for a number of new higher-value AI SoCs, some of which extend our reach into entirely new markets. We remain optimistic about the long-term secular growth opportunities in the edge AI market and our R&D priorities are aligned with both the Physical AI markets that represent a vast majority of our total revenue today as well as the robotic and edge infrastructure markets that are in the early stages of developing.

Altogether, our technology, product and new go-to-market combined with the significant secular growth in edge AI are increasing our 5-year serviceable market forecast today. Before I review our new market forecast, I would like to step back and discuss the market environment we are in. Demand signals for the application of edge AI remains strong. At the same time, it is obvious that memory vendors and the entire supply chains are prioritizing AI data center demand, which is resulting in rising supply chain costs for everyone. Surging memory price and the scarcity of supply are impacting the entire industry.

Related to this, we are providing significant assistance to customers who are attempting to create a wide variety of workarounds to the memory situation. Ambarella itself is also facing rising supply chain costs, and we plan to pass this cost to our customers to maintain our long-term gross margin target of 59% to 62%. Returning to our rolling 5-year serviceable market update, I would like to remind you of our methodology. Our SAM for any given year is based on the products we expect to have available for production in that year, overlaid on the total available market projections from a number of third-party research firms.

So our 5-year SAM captures any revenue-generating products announced or unannounced on our road map in the next 5 years. Our prior 5-year rolling SAM was announced in May 2025 and projected a 5-year fiscal year '26 to fiscal year '31 compounded annual growth rate of about 18%, with Auto representing a slightly higher proportion of the terminal year. Our new 5-year rolling SAM from $8.5 billion in fiscal year 2027 to $22.9 billion in fiscal year '32 represents a CAGR of about 20% with IoT markets now representing about 70% of the terminal year.

While there are several factors behind the strong growth and the underlying mix change, I will focus on the most important change. In the last year, it has become clear that operational efficiency or the ability of our enterprise to generate more revenue and/or to reduce expenses is likely to be a key driver of our emerging edge infrastructure business. Operational efficiency at the edge refer to the use of open weight and distilled models running on on-premise inferencing hardware in contrast to the large frontier models that run in the cloud. Benefits of this approach include reduced latency, data protection, privacy, lower bandwidth costs and high reliability.

Target markets include security, retail, lodging, logistics, healthcare and more. The on-premise operational efficiency use case has emerged with growing expectations for sustainable high-volume inferencing and increasingly for agentic AI and Physical AI application that can perceive, reason and ultimately act in the physical world. The key question has become who can help the enterprise lower the cost per useful AI inferencing outcome? This is where Ambarella's superior performance per watt portfolio kicks in, providing the efficient edge intelligence needed to enable this next-generation agentic and Physical AI workload at scale.

With this perspective, in the last year, we have several new products in development targeting on-premise hardware or what is commonly called edge infrastructure. As you know, we already have our N1-655 AI SoC in the market, and we have additional unannounced AI SoCs in development. We also are implementing a stand-alone AI accelerator product line targeting the edge infrastructure market. Together, this new edge infrastructure products, both AI SoCs and stand-alone AI accelerators represent the single most important reason for the upward revisions in our SAM.

Before I introduce our first stand-alone AI accelerator, allow me to be clear about our terminology. We define edge AI SoC as one integrating all of the accelerated computing functions into a single chip, camera perception, AI accelerators, CPUs, encoding and so on. We define an AI accelerator as an AI processor that is not camera specific and targets a wide variety of digital or physical modalities. We believe this type of multi-modality is critical for edge infrastructure applications that target operational efficiency.

While not formally announced, I would like to preview one of the new AI accelerators that will anchor this new product category for us with another well-defined, well-performed product already behind it. We refer to this new AI accelerator as X7. This SoC is sampling now and expected to land initial design wins in edge infrastructure applications where it can serve as an AI coprocessor for host processors such as ARM or x86. Together with our new product thrust, expanded market reach and the SAM, we expect our revenue growth to be supported with 2 incremental go-to-market strategies. First is the multistep establishment of indirect sales channel and the second is a semi-custom chip strategy, both of which will augment our existing direct sales efforts.

As a reminder, virtually all our revenue is generated by our direct sales teams. And today, I'm excited to announce 2 material partnership agreements to develop our indirect sales channel. Combined, these 2 partnership plan to drive a significant amount of incremental revenue over the next 7 years through customers who have largely been unserved by us so far. First, today, we announced Ambarella's strategy partner with CapGemini designed to help enterprise adopt edge AI and Physical AI solution faster by reducing the complexity of moving from evaluation to scalable deployment.

By combining Ambarella's power-efficient AI software and platforms with CapGemini's global engineering, system integration and industry expertise, the partnership aims to help customers improve operational efficiency, enhance real-time decision-making and deploy intelligent system and in physical world environment with greater speed, scalability and confidence.

In our second partnership to develop our indirect channel, today, we also announced a 7-year agreement with Macnica, a leading global technical distributor. Macnica will support both Ambarella's Physical AI and the new edge infrastructure products by developing and supporting an independent software vendor ecosystem, including onboarding, technical integration support and joint go-to-market progress. With this ecosystem in place, Ambarella solution can be offered as individual component or as a complete bundle for multiple edge AI vertical markets, including video analytics, smart city, edge computing platforms, robotics, industrial IoT, intelligent transportation systems, retail analytics, security and surveillance.

I want to emphasize the importance of the indirect channel to serve small and midsized customers and highly fragmented market like robotics. However, the indirect channel is also critical to support our more complex AI SoC targeting the edge infrastructure where a broad network of partners is vital for our long-term success. Meaningful revenue is expected in 2 to 3 years and will grow as we introduce new products for the market. Our second incremental go-to-market is our semi-custom opportunity, which can enable us to gain more share in existing market and reach into new markets.

We have our first semi-custom project underway, the 2-nanometer CV8 SoC, which is expected to generate first production revenue in fiscal 2028. And we are in discussion with other companies for additional semi-custom chip projects. Our representative customer engagement this quarter once again demonstrates Ambarella's expanding traction across a broad set of applications, robotics, automotive, security, trail cameras and smart video intercoms. With a CV72-based quadruped robot validates Ambarella's high resolution, high multi-camera edge AI capabilities in robotics. A major S&P 100 communication equipment company announced an AI-based enterprise video intercom, further extending our reach in the emerging access control market.

We landed another win with Moultrie for AI trail cameras and win with Canon, Suprema, IDS and Sepro further strengthen our AI monitoring pipeline with CV75, CV72, CV5 wins using our own AI ISP software. Through Tier 1s, we had 2 in-cabin vehicle wins with Tier 1s in China, one for driver monitors and one -- the other for more complex camera monitor system used in Audi and the VW vehicles. The breadth of these wins and the wide variety of corresponding AI workloads highlight the programmability and the flexibility in both our AI SoCs and our Cooper Developer Platform. This ease of use is facilitating the onboarding and expansion of our indirect sales channels. Very few competitors can offer this type of proven platform with more than 50 million edge AI SoCs shipped.

In conclusion, I remain very excited about the overall growth opportunity of the edge AI market and our company-specific growth drivers put us in a unique position to benefit. Ambarella is expanding beyond low-power AI SoC to deliver the complete foundation for Physical AI, and we are becoming a full stack Physical AI platform provider.

With that, I will now turn it to John.

Thank you, Fermi. I'll now review the financial highlights for the second quarter fiscal year 2027 ending July 31, 2026. I will also provide a financial outlook for our third quarter of fiscal year 2027 ending October 31, 2026. I'll be discussing non-GAAP results and ask that you refer to today's press release for a detailed reconciliation of GAAP to non-GAAP results.

For non-GAAP reporting, we have eliminated stock-based compensation and acquisition-related expenses adjusted for the impact of taxes. In addition, this quarter, as described in our Q1 fiscal 2027 10-Q filing as a subsequent event, we recognized a $9 million reduction in our GAAP research and development expense due to the cancellation of a customer's development project. We do not expect any impact on our non-GAAP outlook from this development. For fiscal Q2, revenue was $108.1 million, slightly above the midpoint of our prior guidance range of $105 million to $111 million, up 7.7% from the prior quarter and up 13.2% year-over-year.

Automotive revenue established a new revenue record on continued strength as the commercial vehicle adoption of AI remains strong, and Auto revenue slightly outpaced the growth in our IoT business, where our enterprise-driven businesses outperformed our consumer-led businesses. Non-GAAP gross margin for fiscal Q2 was 59.3%, below the midpoint of our prior guidance range of 59% to 60.5%. Non-GAAP operating expense in Q2 was $57.4 million, slightly below the midpoint of our prior guidance range of $56 million to $59. Q2 net interest and other income was $1.8 million. Q2 non-GAAP tax provision was approximately $344,000. We reported Q2 non-GAAP net profit of $8.2 million or $0.18 per diluted share.

Now I'll turn to our balance sheet and cash flow. Fiscal Q2 cash and marketable securities were $272.3 million, decreasing $5.5 million from the prior quarter, but increasing $11.1 million from the same quarter a year ago. The sequential decrease in cash and marketable securities was primarily due to higher payments for IP licenses. Receivables days sales outstanding decreased from 35 to 32 days. While inventory dollars declined 4% sequentially, the days of inventory increased from 145 days to 157 days. Operating cash outflow was $260,000 for the quarter.

Capital expenditures for tangible and intangible assets were $6.8 million for the quarter. Free cash outflow was $7.1 million for the quarter. During the second quarter of fiscal year 2027, we did not repurchase shares of our stock. During the second fiscal quarter, Ambarella's Board of Directors authorized a new $50 million repurchase program valid through June 30, 2027. The repurchase program does not obligate the company to acquire any particular amount of ordinary shares and it may be suspended at any time at the company's discretion. WT Microelectronics, a logistics partner in Taiwan that ships to multiple customers in Asia, was 60.2% of revenue for the second quarter. Hakuto, a logistics and distribution partner in Japan, was 11% of revenue in the quarter.

I'll now discuss the outlook for the third quarter of fiscal year 2027. We are anticipating favorable seasonality in our fiscal third quarter with revenue in the range of $115 million to $124 million or $119.5 million at the midpoint. At the midpoint, we expect our growth to be led by Physical AI demand from the IoT market. We expect fiscal Q3 non-GAAP gross margin to be in the range of 59% to 60%. We expect non-GAAP OpEx in the third quarter to be in the range of $56.5 million to $59.5 million. We estimate net interest and other income to be approximately $1.9 million, our non-GAAP tax expense to be approximately $700,000. And our diluted share count is expected to be approximately 44.9 million shares.

Thank you for joining our call today. And with that, I'll turn the call over to the operator for questions.

[Operator Instructions] And our first question for today comes from the line of Christopher Rolland from Susquehanna.

This is Dylan Ollivier on for Christopher Rolland. So it's nice to see your road map sort of expanding, and I know that you announced this X7 accelerator. I was hoping to hear a little bit more about this new chip. Is this a chip that you can bundle with your existing N1 portfolio? Or does this address a different part of the stack?

So yes, Chris, for the X7, this is -- chip is an accelerator, which can be bundled with any host, including our own chip. So in fact, that some of our customers using a certain part number and when they feel they need to have more AI performance for certain workloads, the X7 give them a flexibility to upgrade the product without redesign the board. So this accelerator definitely is a way to design that. But in addition to supporting our own SoCs, but any other CPU like ARM or Intel chip, Intel CPUs that we can also bundle X7 with that as an AI accelerator.

Great. I appreciate this. And for my second question, I wanted to ask about sort of the Physical AI and humanoid opportunity. Is this responsible at all for this increase in SAM? Are there any new engagements or new designs that you can point us to?

Yes. So definitely, that's a big part of that. And last -- in the last earnings call, we talked about 15 design wins for the robots, including for roughly $100 million. Although we didn't give you another breakdown, but I can say that we add more design wins to that pipeline and a higher revenue target. So from that point of view, we continue to make progress. But in addition to robots, I also think that edge infrastructure and also enterprise security as well as portable video are all the reasons that we are increasing our SAM number.

Yes, Dylan, we did -- Fermi mentioned a quadruped robotic dog with the CV72 chip this quarter. So continue to add on to the robotics wins we've described before.

And our next question comes from the line of Joe Moore from Morgan Stanley.

I wonder, first, in terms of the broader ecosystem, you talked about some of the challenges in memory. What is that meaning for your business? Do you think -- is there a risk of pull forwards or things like that because people are trying to get ahead of memory price increases? Is there pressure on you? Just what are you seeing from that memory impact from your customers?

Right. So we continue to monitor this situation very closely, by talking to customers all the time. So for Q3, we are comfortable with our -- the guidance we provide today. In Q4, we continue to talk to customers to make sure our customer will have -- we can secure enough memory for a Q4 business. That's definitely the uncertainty that we are dealing with.

Okay. That's helpful. And then in terms of opening up to a broader ecosystem, distribution partners, things like that, I think you made the comment about -- that would take a couple of years to inflect. I guess I would sort of think that those customers would act a lot more quickly and would -- that pipeline could build a lot more quickly than what you had seen previously in automotive. Just what do you -- what was the comment that I maybe misunderstand there? And then what is the time line to start to see traction from that kind of broader ecosystem?

Right. So when I say 2 to 3 years, we talk about meaningful revenues. And I agree with you that we -- in fact, we already start seeing a small amount of design wins, which can generate revenue next year. But when we talk about meaningful revenue that will have an impact to our revenue forecast, I think that will take 2 to 3 years. In fact, when we talk to both CapGemini and Macnica, we kind of -- in fact, the range of revenue we are expecting from this collaboration is $0.5 billion with each one of them. So from that point of view, we're definitely looking forward to gradually ramp up the revenue for the next couple of years and start seeing meaningful revenue behind that.

And our next question comes from the line of Tore Svanberg from Stifel.

Congratulations on the Macnica and CapGemini partnerships. I'm curious on those for me. What are some of the early use cases that those 2 partners are going to be helping you with? Maybe you can call it some markets or applications. And how should I think about that in the context of your Cooper platform? Are they going to be working with you on Cooper? Are they going to be providing some of their own software? Just curious how that's going to play out?

Right. So -- let me answer the second question first. Yes, both of them will use Cooper. In fact, that's a key driver for them to select to work with us because they see a very mature software platform they can immediately tackle on and start building around it, generating infrastructure for their own product line. So that our mature AI SoC as well as a mature Cooper software platform is the probably most critical engineering aspect that we offer to our partners.

Go back to the potential market that we are talking about. In fact, there are multiple of them. And in fact, when I talk to Macnica, CEO in that meeting, they are highlighting that they have already started winning design wins with our solution on drones, on retail channels and also manufacturing. So that is definitely -- you can see that it's really a large market that. But -- however, most of the design win is small and segmented at beginning, but can ramp up to -- if they can ramp up to large volume of business, that will take time. But we already start seeing our partners start talking about different applications.

Tore, it's Louis. They can work together as well. As Fermi said, Macnica can serve small to midsized markets that oftentimes are very fragmented. But really for CapGemini, it's large enterprise customers, and you can look at who they've talked about before. Those are the type of customers we'd really go after with them. So they're very complementary to each other.

Very good. And as my follow-up, on the edge infrastructure market, this is obviously a completely new area. It sounds like that's the sort of biggest contributor to your increased SAM. I'm just curious, who's going to be some of your partners there? I mean, are these going to be your end customers sort of building their own infrastructure? Or is there going to be like an intermediary company that's building it? Is it going to be the traditional server guys? Yes, just curious how that's all going to play out?

Well, I think, obviously, we're going to continue to talk to some of the large customers directly. But at the same time, we're counting on CapGemini and Macnica help us to penetrate this because they're already in that market, they're already selling solution to the existing edge AI customer with their existing solution. So working with them will help us to ramp up our revenue much faster than just were talking to direct customer directly.

And our next question comes from the line of Quinn Bolton from Needham & Company.

I just wanted to ask just longer term on the -- sorry, Macnica and CapGemini partnerships. Does that change the long-term gross margin target? I assume that there's probably some allocation of revenue that would be attributed to those partners. And so I'm wondering if that has any gross margin implications as that indirect channel ramps?

Right. So today, I think our long-term gross margin is still 59% to 62%. We are definitely trying to continue to watch because this is just -- we just start ramping up this business. If there's any change, we'll definitely inform our investors. But today, for us, after we talk to CapGemini and Macnica, we don't feel there's any need to change that target today.

Got it. And then I guess just a clarification on the $9 million charge for the project that was canceled. Was that a semi-custom project that was canceled? And does that have any impact on your expected revenue time line for the semi-custom business?

Yes. Thanks, Quinn. It is not one of the semi-custom opportunities that we were -- that we've talked about. It was a development project with I guess, you could say, an automotive customer, auto autonomy customer. And we've been negotiating the termination of that for quite some time. And in Q2, we finalized the agreement.

And our next question comes from the line of Kevin Cassidy from Rosenblatt Securities.

Going back to the shortage on the memory side and you've got near-term visibility. But I'm wondering on the designs, I know a lot of your customers or the market out there is probably dominated by a GPU-based embedded product that uses much more DRAM than yours would. Are you seeing any additional interest because you're more efficient with DRAM content?

Well, yes, first of all, the memory situation is dire for everybody, but some of our competitor who has more money to buy more memories. But however, any customer who come to us for the edge AI or Physical AI, they probably only use GPU for their first-generation product, and they understand. So the memory cost is just one reason, but more importantly, it is power efficiency and other reason. But the memory cost definitely is a driver for people start considering what's the more efficient way to do the product. So I agree with you that some of the -- most -- in fact, almost all the customers who come to talk to us is because our power efficiency solution and the lower cost solution than what they're using.

Okay. And maybe along the same lines with the AI accelerator, you'd be competing against a GPU that uses a lot of memory also. What is the memory architecture inside your X7?

Well, in fact, that we need a much smaller footprint. For example, we only need 4 megabytes memory for the accelerator running large language model. So just give -- show you the -- and more importantly, the accelerator, the power envelope you have to fit in is anywhere between 4 to 5 watts in the current design win. So all of the power efficiency, memory size and also cost is really helping us to penetrate this market right now.

And our next question comes from the line of Suji Desilva from ROTH Capital.

just a clarification for me on the X7 chip. Is that competing really only with edge GPUs? Or is it other AI specialty chips? Or how should we think about the competitive landscape for this new offering?

Right now, well, in addition to NVIDIA and Qualcomm having similar products in this market space, there are probably 50 start-up companies doing similar chips. So it's a crowded space. But however, at the end, it's really about the power efficiency because I just talked about to run a certain workload, you have to have a mature -- not only a power-efficient solution, but mature hardware and software, which I think we are one of the very few that can do that today.

Okay. That's helpful for me. And then my other question is you're talking about customization now projects. I'm just wondering what's precipitated the demand from the customers or your push to provide customization? What's newer versus your standard product history now that's driving the need for that or your desire to do that?

I think you're talking about the optimization for the memory situation. Is that correct?

Well, I think...

Semi-custom.

Oh, semi-custom.

Semi-custom, I apologize. Yes.

So -- yes. For semi-custom customers, in fact, we basically allow our customers give us a spec and we build on the spec. But however, when we negotiate spec with a customer, we need to make sure that we can sell the spec to somebody else. So we -- for the semi-custom chip, we pretty much build a purpose chip for the one customer, which they benefit from this. But at the same time, we can sell the chip to others that are not competing with the key customer. That's the business model and how it works on the engineering side.

But of course -- Suji, it's Louis. We'll try to offer as much of our own IP in those semi-custom chips as possible. For example, we have our own IP for the AI accelerator, the NPU for all the perception capabilities, including ISP, and the encoder, the CPUs, all of those functional blocks are available for a customer to develop a semi-custom or custom chip with.

And our next question comes from the line of Liam Pharr from BofA.

Is there a way to frame how much memory cost inflation you're absorbing this quarter, either in basis points or maybe what gross margin would have been without any memory cost inflation? And is the path back above 60% feasible while memory prices stay elevated? Or does that require pricing to come down?

Right. So first of all, the memory price doesn't impact our gross margin. It really only have a potential to impact how many chips our customers can buy. So memory cost because we don't buy memory, and we don't resell memory. So the memory price has no impact to our gross margin. So that -- I think I hope that answers your question. But the real question for us is how that memory cost can -- because our customers need to increase the price, whether that will reduce the total volume they can sell and therefore, reduce the total ordering to us, that's something we need to continue to observe. In Q2 and Q3, we see a little impact on our revenue because of memory situation. We continue to watch for the Q4.

And then I guess for my follow-up, Q3 is guided up 10.5% roughly sequential versus 13.5% last year. How much of this next quarter is normal seasonality versus underlying end demand strength? And given you flagged Q4 memory supply, obviously changing the demand picture, how should we think about Q4 seasonality and whether the full year 10% to 15% is still reasonable for the full -- for the guide?

Right. So I think the outcome this year is still a little uncertain because of the memory constraint that you talk about. And like I said, we continue to talk to our customer for that to monitor how that impacts our performance in Q4. Barring for any memory impact to our revenue, I think that you should expect Q4 was a regular seasonality.

And our next question comes from the line of Gus Richard from Northland.

Robotics architecture look an awful lot like an autonomous car in terms of what it needs to do. And I'm just wondering, you have a domain controller for autos and you have the CV products. Are you seeing any traction in the domain controllers? And -- and then any clarification on where you're seeing the strength? Is some of this coming out of China?

Right. First of all, you're 100% right that a lot of robot design system architecture looks just like autonomous driving car, which I totally agree. And however, I think the robotic market situation really reminds me autonomous driving 7 years ago when that at that time, all of our automotive customers in trying to just using individual modules and put the solution together and start demoing and selling the first generation product. I think this is how we are at with the current robots. We see a lot of customers are rushing out their first-generation product by putting individual components together to demo their capabilities.

However, we do believe that integration path of the robotic will be very similar to what happened to the autonomous driving car. It is there will be people going to buy perception system, but down the road, people want to buy domain controller. We do see both opportunity today, but I would say majority of our customers today is asking for perception modules, perception solution. But on their road map, they want to have a way that can buy a domain controller in the long run. So I think we have a complete road map. We can sell just perception system to a customer today. In fact, people want to buy brain -- domain controller like for the brain of the robots, we have the solution, too. But our plan is we're going to continue to develop solution for both so that we can cover the total space of robotics.

Got it. And then just if I think about, again, robots, cars are 2D and robots are 3D. And I'm just wondering, is one of the limitations of penetration training and can you help your customers train robots. Thinking about humanoid, but -- go ahead, sorry.

Right. So in terms of training, it's really about how to collect data. One thing we help our customers is we build a platform for people to collect data easily. And also we provide a platform that can provide a service to help people to label those data automatically. So people can use our system to -- reference design to collect data. In fact, some of the, I would say, the people doing mapping, generating the 3D mapping are using our system to collect data. And also, we are providing service to some of our automotive customers that we can -- using our tools to auto labeling all of the data they generate. Those are 2 things we can help to provide assistance on the training side.

And our final question for today comes from the line of Martin Yang from Opp & Co.

Fermi, you sized the potential revenue from CapGemini and Macnica pretty similarly, but they face different variety of customers. Can you maybe talk about the methodology you arrived at those dollar figures? Is a similar methodology or a very different approach to size those potential markets?

Yes. Go ahead.

This is Muneyb just jumping in there. I think both Fermi and Louis were commenting earlier about how complementary they were, right? So I think, one, on the Macnica side, I think Louis has commented, it was is large-scale, medium, large kind of customers we haven't addressed in the past. So think of them as a large volume play where we've typically directly engaged with high-volume customers. These will start aggregating a whole bunch of small, midsized customers that we did not have access to in the past. So it's a volume play, and I think Fermi already indicated that we're starting to see some small design wins come through with these distribution.

And then if you think about CapGemini, it's more of a value play. And I think Louis indicated before, these are large enterprises and customers who will bring complex solutions, deploy at scale to enterprises. So the modeling is on both slightly different. One, distribution channels, reseller scaling with small design wins, so building up small volume. The other ones are large customers and logos, which have much larger opportunity deals, but complex opportunities. So on both sides, the modeling is done on value versus volume. And I think the earlier question was also, you should see different time lines on this.

So we do expect faster time lines on the distribution side and more longer time lines on the more larger complex opportunities. But the modeling has been built out over 7 years of how this will come to fruition. And of course, they are -- some of them new to our products. So initial ramp-up, market making, pilot opportunities is what we are allowing for. But we will keep you updated as we start winning some large deals and meaningful revenue, as Fermi pointed out, in future quarters.

Great. I have a follow-up on X7. Is that accelerator chip primarily targeted for as a channel product? Or there's no distinction between for channel or for direct?

There's no distinguish. And in fact, that I'm expecting that both CapGemini and Macnica will do product rapid design for that and targeting different customers.

This does conclude the question-and-answer session of today's program. I'd like to hand the program back to Dr. Fermi Wang for any further remarks.

And thank all of you for joining our call today, and I hope to see you and talk to you next time.

Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.

Welcome back to this afternoon session at the BofA Global Tech Conference. I'm Vivek Arya from BofA's semiconductor semi-cap equipment team. I'm really delighted to have Fermi Wang, the Chief Executive Officer of Ambarella, join us. And like always, I'll start with a few of my questions, but please feel free to raise your hand if you'd like to bring up something.

Fermi, welcome back. Really delighted to have you at our conference again.

Thank you.

Really looking forward to this fireside.

So before we get into kind of the nitty gritty of the quarter, maybe help us, Fermi, in kind of framing your edge AI opportunity? Which are the areas that you think Ambarella's value proposition kind of best stands out? How are you investing in those areas? Like what is the right way investors should think about Ambarella at this point?

I think we definitely focus on edge AI, as you said, but I think really, what we are offering is a very extensive road map on the silicon side and very mature software stack for our customer whoever -- it doesn't matter what kind of application they want to. They can program their application on top our video processor as well as AI inference engine and quickly turning it to a product.

So that extensive road map, mature software is what really differentiates against our competitor at this point. And we can, of course, go through the details. We have better power efficiency and much better video quality. Those are also important, but just looking at the breadth of our offering in terms of the silicon point of view and the mature software in terms of models and other things, SDK point of view, I think that's definitely our main differentiation.

Got it. Now investors would say, well, there's a lot of alternatives in the market. There are many companies who -- whether it's Qualcomm or NVIDIA or others who kind of play up and down the stack, right, as low power as being in a mobile device to obviously as high power as being in a data center server. So how would you help us kind of rank Ambarella, right, your attributes, your differentiators versus the Qualcomms and NVIDIAs of the world?

Right. So I think for that, I want to emphasize that when we design silicon, we design -- we talk about this, say, algorithm first. When we design just any silicon, we think about application. We think about how to implement most efficient architecture for both silicon and software point of view to address that application. Instead, I think our competitors are thinking about programmable architecture, GPUs or app processor to serve a totally different kind of workload.

So from that point of view, that's where we can really optimize not only power efficiency die size almost all of that around this target embedded AI, I think that's a major difference. Yes, we're much smaller. But if you look at high company with $3 billion, $4 billion market cap can compete with NVIDIA and Qualcomm. It's really about technology we're offering.

Okay. And if you were to help us segment the 2 big areas, right, automotive and IoT that Ambarella plays in, what is the right kind of formula we should think about projecting growth for the next 3 to 5 years?

Right. Maybe I'll talk about this year first. Last year, we talked about IoT growth faster than auto. But this year, although the growth rate is similar, but I think we just talked about auto, we have a solid growth this year. And in fact, that we're going to have a record revenue for automotive this year. And that will show you that although that a lot of people are focusing on our autonomous driving market, but our other automotive market has really become a solid growth engine for us, including financial fleet. In that financial fleet, there are many things, including telematics or fleet management as well as other DMS and type of location. This is an area that really grows nicely to us.

And then on the IoT side, it's really where we have several important growth engine for us. Everybody is talking about robots, well, everybody agreed it's segmented. It's going to take a while to grow. But if you really look at it down the road, I fundamentally believe that when I retire, it's going to be a robot taking care of me, not my son. From that point of view, I really wish and I think I want to do everything I can to make this product become available to everybody. But I also believe with GenAI technology and the right silicon architecture, we will get there. But however, it takes time to get there.

But if you look at short term, that in the robotic application, I think the biggest opportunity for us short term is drones. And we are seeing a lot of consumer drone commercial drone opportunity out there, which really differentiate based on, first of all, we need to be really good at perception, we need to be good at fusion and the decision-making, which is really similar to the technology we develop for autonomous driving. So that is one area we feel excited.

The other thing I will feel excited on the edge AI side is, of course, there are a lot of different applications, but there's one new application we call the edge infrastructure. It's really that people are going to aggregate a lot of different sensors and using a box to really upgrade the existing infrastructure to gen AI level of AI performance.

Just to give you an example, we talked about this before, in this hotel floor, there's, I don't know, 20 different let's call them traditional cameras. Maybe some of that has AI performance, some don't. But to upgrade those cameras to be GenAI ready, it's not easy. You need to change all the camera or you can plug in appliance that aggregate all the video feed into that box and suddenly, all camera can be enabled by GenAI, running the AI on the appliance.

So there are many different applications popping up in the last few years -- just months or few quarters that we think has a lot of growth potential. But I think the key is, today, all the investments focus on data centers and the edge, it really takes time. For example, how you're going to run a really large language model on a chip, which consume 2 to 5 watts and deliver high-performance running 1 billion parameters ChatGPT-type or small LLM models on 2-watt to 5 watt chip and selling for $30, $40. That is a challenge, and we provide a silicon solution and the customer on their way to put their solution on top of that. So we are enabled a market that does not exist before because that the mutation of silicon architecture and models just didn't exist until just recently.

Got it. How would you kind of urge investors to think about Ambarella's growth rate opportunity over? Like is it a 10%, 15% kind of growth opportunity? Is it 15%, 20%? The reason I ask the question is that traditionally, more consumer-based products have grown at a slower pace. And now when we think about AI, people always think about 30%, 40%, 50% kind of growth rate. So help us -- help calibrate us on what is over the next 3 to 5 years, what is Ambarella's growth rate opportunity? .

Yes, 30%, 50% is talking about data centers, but there's no edge AI market that grow that fast. When we look at our SAM, our SAM rental growth at 18%. So that's really the limiting factor. Do I believe that's going to grow faster in the future? Absolutely. But however, we haven't seen that turning curve yet. And I think we are waiting for many things to happen on silicon, software, our customer software and models, everything need to come together and enable new applications. So I think we haven't reached that point. But today, this year's guidance is 10% to 15%. Next year, our Street, I think some analysts looking at 15% as that's reflecting the reality. But if you look at the SAM, we think that can definitely -- we should grow faster than the same in the near future.

Got it. Okay. Makes sense. Are there certain product category -- so auto is easy to understand, right, and you mentioned all the growth in ADAS and AV and whatnot. But when it comes to IoT market, right, because you're larger -- what are kind of the form factors that you think will help you outgrow your SAM?

Well, let me just give you various examples that one of our customers, their form factor is a box, maybe it's just this big, and they need to put in 8K P60 video running AI performance on the 4-watt system power. Just think about that, right? It's challenging enough for any kind of silicon, but we are probably one of the very few can deliver that kind of performance at the power envelope.

But this is not one unique application. This is really -- when you look at any edge AI application, most of them are battery-powered. So it's not just a form factor limitation. It's really that your -- how long you can run an equipment is based on the power efficiency of your silicon system and the battery size, right? The battery size is fixed, and then you have to really control your silicon, which has to run billion parameter perimeter GenAI model, while give others -- for some product have to be 2 to 3 hours. Some product has to be 8 hours, for example, the wearable camera. So this is really the limitation. How you drive performance per watt is going to be the key metric for edge AI applications.

Got it. Recently, on your earnings call, you mentioned that you had signed a long-term agreement with Hanwha. So can you walk us through what kind of led to that element, what kind of products are included? When will we start to see the benefit of that agreement?

Right. First of all, for people who are not familiar with Hanwha, Hanwha is a Korean company that sells USD 60 billion revenue per year. It's a huge company. And they have a retail, they are banking, they have drones, they have robots, you name it. They have quite a diversified product portfolio. I think what they are trying to do is I think their Chairman looked at us and said we need to go to have edge AI, but they don't -- he doesn't want to allow each of their group develop their own edge AI. So they say, how do we have a solution that we can have a unified solution for the edge AI moving forward?

And while the group is called Hanwha Vision, which has been our customer for the last 15 years selling enterprise security camera. So they were assigned by the -- because they are really the first group inside Hanwha trying to deploy edge AI solutions. So they were assigned to evaluate a solution that it can be accepted as the group level edge AI solution. And for us -- so basically, they are asking for product portfolio wide enough to address multiple different opportunities and also software mature enough that can quickly enable different applications I think that's really our strength.

In fact, the first question I asked at the beginning, that was the reason they choose us. But at the same time, we would like to work with them because they present huge revenue opportunity for us. But more importantly, I think that working with a company like Hanwha really give us not only the scale but also how easy to forecast our revenue in the future. We talk about working with different companies. You really have better way to forecast revenue. So all these things is reason that we want to work with Hanwha. So I think there is definitely a strategic value for both sides to come to the conclusion on this.

And the other thing I think is driving this decision is really for anybody who want, any company want to build their own silicon. In fact, Hanwha was doing their own silicon before. The cost just become prohibitive on that. When you go to 4 nano and 2 nanometers for data centers, it's easier to justify. But when you come to edge AI, the silicon price is anywhere between $20 to $50. Trying to build their own silicon paying $100 million just to do 1 silicon for 2 nano doesn't take sense.

So I think there are many reasons. And in fact, this is this is our second kind of agreement that we achieved in the last 12, 18 months. And there are more engagement that we talk about. It's really the same concept. People want to have a differentiation on their product line. They don't want to pay for their own silicon but they want to have some secret sauce in the silicon, and we can help them to build that. And with that, we get a better relationship with the customer and the more guarantee on the revenue. It's a win-win for both sides.

Got it. And I think you mentioned, is it $800 million over 10 years? How do you see that profiling over the next 10 years?

Right. Hanwha Vision is a customer today. I can say that their revenue with us is roughly mid-single-digit percentage of our total revenue. So from that point of view, they are not the largest customer for us but a sizable customer. But however, from there, they multiple ways to scale up our revenue. First of all, like I said, they are doing their own silicon. We only have less than 50% total market share. For us, we can probably assume that we want to get better market share moving forward quickly. And also that all the new chip we're going to get is going to be higher ASPs. And that's something what should happen in a short period of time.

More importantly, I think most of the other Hanwha Group can use Hanwha Vision's camera to plug into their product. For example, today, it's not about the form factor of the camera. It's really about what kind of software and the model you run on it. So any Hanwha Vision camera, if you swap out the models and the software running on it, which we can easily enable, you can enable retail application, you can enable drone security, you can do many things even with simple factor that Hanwha Vision generates. So we think that there's a quick path to revenue also to help other, enable other applications by using Hanwha Vision's existing product line.

But then the last one is, really, this is a multi-generation commitment for both sides. So we will help them to build a solution they can use. When multiple generation, when the second generation come up, we definitely can address even bigger TAM or SAM for Hanwha. So that's really the progression of getting more revenue from Hanwha.

Understood. So just for the near term, do you think that is kind of already reflected in this and next year expectation? Or could that be upside, too?

I don't think that's been factored in just because consuming -- we just talked about. But we probably start thinking about how to reflect that in our future. But next year is really about really gain shares on the existing product line, right? So I won't say immediately reach to higher level but next year, we should some positive impact to our total revenue.

Got it. The 10% to 15% growth rate for me is that constrained by supply in any way? There's a lot of cost inflation, right, in many of the end markets that you serve from memory or other things, right? So how much of this 10% to 15% is gated because of supply or other inflationary aspects in your business?

It's definitely not limited by our supply. It's really the market condition, particularly on the memory side that continue to bother almost all our customers. Our annual guidance from Q1 to Q2 didn't change. That reflects that we consider all of possible scenarios. We think that's a prudent way to give the guidance. But we continue to tell the customer, in fact, customer gives a very consistent feedback they still can buy memory. It just became extremely expensive. What that means is, in the future, they probably have to pass those actual cost to the customer. So from that point of view, others impact the demand and so whether people can continue to secure DRAM supply moving forward. I think those questions we don't know the answer yet. So that's definitely something we factor into our consideration.

Got it. This is a little bit of a very near-term question. We saw that when your results were kind of in line, you guided somewhat better, you outlined this LTA. There was a volatility with the stock the next day. The market has 50 different reasons for being volatile. What do you kind of ascribe that to? And what do you think is kind of underappreciated and missed by investors right now?

Right. But if you look at what happened before the earnings in the previous 40 days, our stuff jumped $40 also, right? So in kind of that people build out expectation, I think everybody expecting that semiconductor company go to raise their guidance, right? So when we didn't, I think that's probably reflected on the people buying short term, selling short term. So I really don't think those people who trade in that period reflect our long-term investors' view.

I really think that for the people who are buying our stock really appreciate, first of all, people need to appreciate GenAI because it's not -- if there's a home run in that GenAI story today, but they have to believe that with our technology, we will become one of the few players that survive through this -- well, we continue to play in this market when they become big.

Got it. As you look at next year, Fermi, are there opportunities for you to outgrow your SAM? Or is it too early to kind of think about that?

Well, I think there are a few things we should consider. First of all, we talk about our 2-nanometer chip, CV8, which is much higher ASP than our company, corporate-wide ASP. So from the ASP point of view, it should provide a growth areas. So definitely, our CV7, our first 4-nanometer chip, will ramp up next year, too. So from that point of view, it should help, right?

But however, at the same time, the biggest challenge for me is to forecast how fast our customers will ramp up those new products based on CV7, CV8, how the memory impact to that. So I won't -- we haven't provided any guidance for next year yet, but I just want to give you the pluses and minuses that we are considering right now.

Got it. That's fair. And I think you have another LTA or semi-custom, right, at 2 nanometer. Could you help us understand what that is and when that start to impact your financials?

Right. So we add a little bit more information this time. The chip is our first 2-nanometer chip called CV8. It was taped out in January, and we'll be back to our office in a few weeks. The first 2-nanometer chip, we're going to -- all of the indications so far is that no process will be able to ramp up in production in second half of next year. In fact, second quarter next year. And that's how we want to ramp up or test out our chip and make sure that our customer can -- if they want to, they can be put out in the second quarter next year. So that's the plan. And that 2-nanometer chip is related to one of the customer willing into sign up a long-term commitment with us. And so that -- they definitely help on the NRE side but also get some benefit from that, which we haven't disclosed the detail yet.

I see. Got it. Okay. And then recently, we saw your inventory dollars rose quite a bit. So how much of that is just because of supply chain is tight and you probably need to take product whenever you can versus preparing for some ramp over the next few quarters?

I think it's definitely both, but I would say one thing. Samsung just informed us that their foundry supply will be tight next year, particularly on the 4-nano and 5-nano process node. The reason for that is definitely you see their announcement, they are seeing extra customer comes in at. So from that point of view, Samsung basically informed us that with a good intention, hey, I know that we have a partnership, we want to really give you the best service. So give us your PO commitment so we can secure those commitments for you. So that's the conversation we have.

So from that point of view, we think it's prudent to build up a little inventory just in case there's upside from our customer and making sure that we can support some upside. But I really don't think that our inventory growth on the current point, you'll probably drop a little bit slowly, but definitely, that's the range we're going to stay at for a while.

Got it. Okay. In general, are you seeing any other inflation pressures? I think your gross margins are kind of back towards the lower end of the 59% to 62%. How much of that is a reflection of kind of the cost inflation? How much of that is a reflection of just pricing, mix or any other product factors?

Mix is always the biggest one for changing our gross margin. But I think we guide to 59% to 62% of long-term guidance. And this quarter, we guided 59.75%, which is the same as the previous 2 quarters. I really think that from the cost side, foundry hasn't raised a price, but other supply chain raised the price, not in a significant way, but definitely is consistent with the feedback you heard in the market. Supply chain definitely tightened up and they are raising place. It's just something we deal with. And our thought is that if the price goes to a point, we need to pass that to our customer, but we haven't reached that conclusion yet. But eventually, if the trend continues, I won't be surprised that our supply chain is going to continue to be very tight in the next year.

I see. Beyond video as a modality, how do you think about other modalities as you look at edge AI? Because I assume that as you think about robotics, something that could help all of us in our old edge, it will require a lot more modality.

Absolutely. In fact, we support many modality already. Other than video, we support of audio, radar, GPS, LiDAR, we support all of that. But I just want to point out, there's no other modality required as much passing performance like video, right? So that's why if you use video as one of the modality, the performance of system is dominant by video requirements, particularly on AI performance point of view. Nothing else even come close.

So from that point of view, we basically trying to address -- and because our customer come to us for video processing, although we didn't talk about our support for other modality that much, but in our HD cable support almost all of that. In fact, even for the -- all the AI models, we support a lot of the non-video models like audio model, radar models, LiDAR models and as well as the non-video LLM models. All of that are supported by our SDKs today.

Got it. A few months ago, we had heard the U.S. institute a lot of restrictions on Chinese drone manufacturers. Is that a source of upside? Have you already started to see some of that upside? Or is that benefit still to come?

We haven't seen a lot of movement on the volume yet. But definitely, this is going to help the U.S. drone market -- I mean, U.S. supplier in this market. So we're looking forward to see impact at this point. But so far, we haven't seen a lot of major move because the reason for that is this regulation change only impact to the drones that haven't been shipped yet. There are a lot of drones that in the market already get FCC approval, they are not impacted by this new regulation. So maybe the next generation we'll start seeing an impact.

How large can that market be? And I mean, I would imagine that they may have started to at least engage with you, right, from a design win perspective?

Absolutely. In fact, in our script, we talk about this time, we have 15 robot design wins, which includes several drone design wins in there. And cumulatively, we talked about $100 million of revenue pipeline in there. So definitely a significant momentum that we're seeing there. But however, just like I said, getting a drone into a production, particularly from a consumer drone or commercial drone, the design cycle is a little long, and we are just in the middle of that.

And also a lot of shipments today are not at impacted. So we are watching the effect after the regulation gets really becoming implemented in the market, it will be some significant impact, but we haven't seen the results yet.

I see. Fermi, maybe talk to us about what your software strategy is, right? Because you have made a lot of investments on the software side. How is that providing stickiness to the platform? Or do the customers just want broad interoperability and then they are the ones who decide which models to run on which platform?

Right. So that's a great point. Let me spend a little time on that. We spent 5 years to develop some platform called Cooper platform. It's a platform that's almost 100% of our customers using today. And why is that? Because the feedback is really pretty positive for a few reasons. Why is -- that's a platform really help our customers easier to put not only the model but our software onto our platform. For example, we continue to add function to it. We just add agentic function on that so people can use agentic tool to help to write software on that platform. And just because of that, it becomes much easier. So you can see that we'll continue to help our customers to get easy to pull software over. That's one.

In that SDK, we support 200 model architectures, not 200 models. Like HULO, we treat that as a 2 different architecture. So we support 200 model architectures. So let's just consider how much from our customer point of view. Really, if they have any specific model they are considering, they will find similar model or model guidance to compare. Just to give you an example how complete and through software stack is. But more importantly, any customer using Cooper SDK to build one product on one silicon, when they try to move the same software to different silicon, maybe higher performance, lower performance depends on the performance they have, it's extremely easy.

It's almost like we're proud how the software works. And of course, you have to get a different performance based on the silicon you chose but the software development is clearly, the effort is really significantly reduced. So from a mature point of view and from a scalability point of view, our customers really appreciate the software that they become the biggest -- that's one of the reasons Hanwha appreciate because they have uses Cooper SDK for the last 5 years, and they see the progress. They really think that's the way to go.

So from that point of view, that maybe that's another point we have to touch on is really we believe that we need to go to indirect sales model because with a robotic application we cannot go direct support to all the possible robotic implication out there. The only way we can do it is enable our software partners and distributors or system integrators to integrate and support customer from. And then our software need to support them and they use our software to put that. So the mature software of Cooper is critically important for that.

Got it. In the last some time that we have left, how do you think about your M&A strategy going forward? Like what would help Ambarella double your growth rate, right, over the next several years? Are there white spaces? Because edge AI is such a fast-growing market so how do you think about your opportunity to accelerate via inorganic means?

On the technology point of view, when I look at all the space we are in today, there's no really one big shiny object I need to have to really complement to my technology. I don't think that exists, but we continue to look for algorithm that can help. But however, I really think that if we have to reinvest heavily, that's indirect sales channel, anything that can speed up that because that's where we can scale revenue quickly. That's definitely the direction we are trying to get more support and help if there's a possibility.

Makes sense. Terrific. Thank you so much, Fermi. Really appreciate your time.

Thank you very much.

Good day, and thank you for standing by. Welcome to Ambarella's First Quarter Fiscal Year 2027 Earnings Call. [Operator Instructions]. Please be advised that today's conference is being recorded. Now it's my pleasure to hand the conference over to the Vice President of Corporate Development, Louis Gerhardy. Please proceed.

Thank you, Carmen, and good afternoon. Thank you for joining our first quarter fiscal year 2020 Financial Results Conference Call. On the call with me today is Dr. Fermi Wang, President and CEO; and John Young, CFO.

The primary purpose of today's call is to provide you with information regarding the results for our first quarter fiscal year 2027. The discussion today and the responses to your questions will contain forward-looking statements regarding our projected financial results, financial prospects, market growth and demand for our solutions, among other things.

These statements are based on currently available information and subject to risks, uncertainties and assumptions. Should any of these risks or uncertainties materialize or should our assumptions prove to be incorrect, our actual results could differ materially from these forward-looking statements. We're under no obligation to update these statements, and these risks, uncertainties and assumptions as well as other information on potential risk factors that could affect our financial results are more fully described in the documents we file with the SEC.

Access to our first quarter fiscal year 2027 results press release, transcripts, historical results, SEC filings and a replay of today's call can be found on the Investor Relations page of our website. The content of today's call as well as the materials posted on our website our Ambarella's property and cannot be reproduced or transcribed without our prior written consent.

Before starting the call, we hope to see you at some of the following investor events scheduled during our second fiscal quarter. June 2 will be at the Bank of America's TMT Conference in San Francisco, June 23 at Northland's Virtual Equity Capital Markets Growth Conference June 23 and 24, we'll be hosting investor meetings in Baltimore and Boston on August 18 at Rosenblatt AI event.

For your calendar planning in our third fiscal quarter, please note, we are a sponsor at the AI Infrastructure Summit in Santa Clara on November 15 to 17, and we hope to see there where we will lead the physical AI track with a number of AI product demos in our exhibit area. Fermi will now provide a business update for the quarter John will review the financial results and outlook and then we'll be available for your questions. Fermi?

Thank you, Luouis, and good afternoon. Thank you for joining our call today. During our first fiscal quarter, we delivered on our key financial guidance, revenue, gross margins and operating expenses. Most importantly, we continue to extend our HA platform leadership with technology and product innovation, addressing existing and emerging use cases.

As a recognized HAI leader, we are entering a new and a significant phase for our market development with the execution of long-term customer agreements, which can drive a more predictable revenue stream while also offering lifetime revenue potential far in excess of what we have realized in the past. Let me provide a few comments about the current market environment. In Q1, we delivered revenue at the high end of the normal season range and slightly above the midpoint of our guidance.

Demand signals and the long-term secular growth outlook for HAI remains very strong, and I'm very optimistic about our ability to serve it. In particular, as AI workloads become more complex. LT applications were about 3/4 of our total revenue and was seasonally down with our enterprise security camera market growing in the high single digits sequentially offset by a double-digit sequential decline in our consumer IoT business. Our automotive revenue established a new all-time revenue record with very strong double-digit growth, led by the rapid emergency of AI within the large and growing commercial vehicle telematic markets as well as automotive safety applications.

After a multiple year build-out of AI training capacity in data centers, the AI market is increasingly focused on AI influencing. And we think the inferencing market, the processing is becoming more distributed. In ours, processing is moving to the edge and the fiscal AI layers of the network hierarchy. As the age market evolves to Gen Ai and Agentic AI, in particular, our positioning become even stronger, and I would like to explain more about this.

First, before talking about Ambarella's unique positioning, let me remind you of the Advantage HAI efforts relative to the data center. HAI processing reduced latency lower power consumption, minimize commission expense and improves privacy and security. So why is Ambarella's HI platform so well positioned. First, Ambarella's HAI platform is comprehensive and well established, yet expanding and under constant evolution to adapt to new AI trends.

We believe a broad and highly programmable AI platform is required to address a wide number of use cases, enabling customers to be more efficient by using software and scaling their business. Our software platform is now open and easy to use and supports a wide variety of AI models with more than 200 different AI model architectures reaching production. We have cumulatively shipped more than 46 million AI SoCs, and we have 12 AI SoC already available with up to hundreds of top life performance.

Another reason in [indiscernible] we are so well positioned for Gen AI and Agenetic AI of age is that our software tools and AOCs integrates all accelerated computing system functions into a single platform. In the data centers, the functions such as data aggregation, acceleration, CPUs and other system functions are usually a collection of discrete SoCs from different vendors. However, at age to be successful our AI SoC integrated all the functions. Fusion perception, AI acceleration, CPUs, including and other system function into one single chip.

And our differentiation is not just in proprietary processing elements and advanced via [indiscernible] integration, but also in the proprietary algorithms, full Hasta software and AI agent frameworks by [indiscernible] entire systems together as workloads become more complex, such as Gen AI multi-model reasoning and autonomous agent-based workflows, our deep expertise across the full accelerated computing stack optimized basically for a deployment become increasing rare and strategic value in the industry. In other words, as our customers need most performance in their AI applications, there are an extremely limited number of companies that can do this and even fewer like are proven and established.

We are now becoming recognized as one of the very few companies that can tie this all together of AI workload game more complex. We are entering a new phase of HAI and the fiscal AI market development, where we are engaged in multiple discussions with customers who want to enter deeper relationships, including multigenerational commitments. This can take the form of long-term agreements or [indiscernible] that involves our standard products and/or our semicustom AI SoC optimized for customers' particular workload.

Relative to our current customer relationships, LTAs will enable long-term partnerships that may include a structured contract involving volume and pricing typically over 5 years or more. Over the long run, we expect the LTAs to be an important driver of revenue growth, improve visibility, resulting less volatility and improve the predictability of our revenue. Our first on TAA example involves our first 2-nanometer chip and the semi-custom HAI SoC, which we tape out in January. And this product is named CV, this AI SoC will serve both consumer and enterprise applications in the IoT endpoint market.

For this long-term agreement, we agreed to develop semi-custom ASIC for a customer who want to support of certain complex AI workloads. We will sell this AI associate as a standard product to a variety of other customers in other markets. And this afternoon, we announced another material this time with Hanhwa in South Korea for the enterprise CapEx side of IoT market.

With Hanhwa this LTA is for the sourcing and co-development of Ambarella's AI technology across Hanhwa's product lines and the industry, including physical security operational automation, life sciences, robotics and other industry market. The agreement has a potential revenue in excess of $800 million over a period exceeding 10 years. and represents one of the largest agreements in Ambarella history and one of the first agreements of its kind in the HAI semiconductor market.

The multi-generation nature of this relationship is expected to enable both companies to plan jointly cross technology road maps accelerated product development cycles and bring new category AI-enabled products to market at scale. This relationship will involve standard AI SoCs we will sell in a variety of markets to our customers. Beyond these first two LTAs, we are engaged in discussions with other companies. Today, I will also provide an update on robotic edge infrastructure and automotive markets, which represents material market opportunity for us. I'm very pleased to share that we now have a 15-plus robotic design wins, including aerodrones, with lifetime revenue exceeding $100 million with more than 30 customers in our robotic pipeline.

Our AI SoC combined high-performance AI influence advanced computer vision and ultra-efficient power consumption into a single age optimized architecture and represents the foundation platform for robotic system to run vision language action VRA models in drones. CD5 enabling platforms such as the antigravity A1 enabling capability, including 8K imaging, real-time perception, accounts navigation, upscale on land and on-device AI inferencing without relying on constant cloud connectivity for these functions.

Strong evolve from flying camera into autonomous areas of robots, Ambarella's CV flow AI crater architecture allow manufacturers to deliver lower latency decision-making. Improve latency, longer flight time and more advanced economy at the edge. The robotic market is fragmented, and we are realizing design wins across a variety of other robotic applications including industrial automation, autonomous mobile robots or AMRs, delivery robots, our AI SoC involves from providing perception, sensor fusion and AI processing to also offer decision-making and the full autonomy needed for real-time robotic awareness and action. This convergence to high-quality imaging and AI acceleration and a autonomy rounding VRA models efficiently position us as a key enabler of the broader physical AI and embody AI ecosystem.

As I mentioned earlier, our automotive business established an all-time quarterly revenue record in Q1 and is on pace to establish a new fiscal year record. Third-party research firms indicate global automotive product is expected to decline 1% to 2% this year. But with semiconductor content per vehicle rising market research firms also anticipate the automotive semiconductor market to grow 10% to 15% this year. We expect our automotive revenue growth to outpace these figures due to our success in commercial-free [indiscernible] and safety applications. The commercial fleet panomatics market offers continue and exciting growth prospects as there is an installed telematic base in excess of 100 million vehicles growing around 10% CAGR, but only about 10% of this installed base is so far I -- we are aligned with industry AI market leaders who are also increasingly demanding AI SoCs that can take now to take on not only more sensors, but more complex AI workloads.

And our platform of 12 HAI SoCs is very well suited to help them scale in this market. I will also provide an update on the build-out of our indirect sales channel that we announced to augment our existing direct-to-customer business. The development of our indirect sales channel is important to not only help us address fragmented robotic market, but also to provide support for our emerging edge infrastructure business. We have already onboarded half a dozen ISVs in vertical industry like retail, industrial automation, transportation, health care and smart cities since our launch of our developer zone at the CES in January with more ISV expected to be onboard by the end of this fiscal year.

In March, for the first time ever, we have pushed an embedded award in Nienburg, Germany, where we did live demonstration highlighting how Embraer AI SoC, software stack and developed tools deliver a competitive advantage across a wide range of applications. from AI agent automation and orchestration to physical AI system deployed in real-world environment. One of our existing design partner to demonstrate a real-time industrial quality inspection solution on CV72 and 1655. Multiple new ISV partners will present our booth including one who demonstrated retail AI solution for in-store and drive-thru optimization. Another ISV demonstrated continues trending for high-speed rail network and the third demonstrating warehouse robot solutions.

In March, we also posted an invitation-only expiration at ISC West showcasing how HAI is powering the next generation of intelligent security and physical AI systems. At the center of our exhibitor was our newly launched CV 7 HAI vision SoC, delivering advanced imaging and on-device AI processing alongside the 1655 HAIC age infrastructure for low-power, high-performance enterprise security applications. One of our ISP partners demonstrate a smart city security solution based on 75 and 165. I will now briefly summarize our representative customer engagement in Q1, and it is notable for the first time, all of the examples are based on our HAI OC, 3 from our CV2 family and 8 from our new CV7Xpund.

In the enterprise security while fiscal security remains a principal driver of this market while seeing our customers develop AI application software that enable their product to provide operational efficiency to our business. Examples including predictive maintenance supply chain optimization and automated customer support, we expect operational efficiency in the long run become an important new growth of shoot of what we refer to as enterprise security today. In particular, as Gen AI and Agentic AI is deployed at age. We achieved an important milestone in March on apron Panasonic announced the first endpoint camera to range AI locally based on the transformer capability in our CV72-AISOC. We also have the number of other CV75 and CB72 wins in the quarter, including [indiscernible] in South Korea, access in Sweden, [indiscernible] in Germany.

And with major communication equipment company in Americas, notably, we had an additional CV2 with [indiscernible] South Korea that also utilize our AI imaging signal processing software. We also won a CV22 platform with [indiscernible] that had another CV win with the major communication equipment company in the Americas. In the industrial market, we earned another AI-based barcode reader project based on CV28, this time with [indiscernible] we expanding its reach beyond the traditional physical security market. In the automotive market, our safety and panamotic customers' engagement activity remains strong. For example, we are pleased to announce Lite an industry leader in the commercial and public sector telematic market is designed in CB75 and CV72 into multiple platforms.

For in-cabin pre-installed safety market, we have 2 B72 wing with South Korean-based Tier 1 Era and the CV22FS win for Western OEM in China. Our new product momentum remains very strong, both in terms of fiscal '21, revenue generation as well as new products that have not started to generate revenues. While our 10-nanometer CV2 family of HCA processor for CN location continue to land design wins and grow our new 5-nanometer CV75 and B72 capable both of the [indiscernible] and the transformer-based gene as well as Agenetic AI -- sorry, as well as Agentic AI [indiscernible] a production ramp and are expected to drive material incremental revenue this year.

Then on top of this, we continue to expect our new CV 7 AI processor to enter production by the end of the year. And in the first half of fiscal 2028 or less than a year from now. We expect our 2-nanometer CVA AI SoC to commence production. All these new products I have described as well as all the new unannounced AIA SoCs, we have in development, target more sophisticated AI workload and command average ASP well above our currently 15 SoC ASP in Q1. As you can tell, we have a lot of technology product, market and customer development activity going. I would like to summarize this quarter's call with 3 observations.

First, the AGM market is just getting started and the momentum is building in multiple areas. Second, Ambarella is clearly an AI technology platform and a product leader, and we are already and we are already well established. I think our positioning is getting even stronger as AGM workloads get more complex, and there become fewer and few companies capable of integrating all the edge accelerated computing functions into a single chip. Third, customers are recognizing the first 2 points and now want to engage with us more broadly and more deeply.

For example, LTA agreements can build stronger relationships and get us designed into new markets like robotic while the indirect channel sales ecosystem bring us more scale. In conclusion, as all this comes together, we intend to drive shareholder value with strong revenue growth and a more diversified and predictable financial models that offer material operating leverage potential for our shareholders. With that, John will now discuss our Q1 results and Q2 outlook in more detail. John?

Thank you, Fermi. I'll now review the financial highlights for the first quarter fiscal year 2027, ending April 30, 2026. I will also provide a financial outlook for our second quarter of fiscal year 2027, ending July 31, 2026. We I'll be discussing non-GAAP results and ask that you refer to today's press release for a detailed reconciliation of GAAP to non-GAAP results. For non-GAAP reporting, we have eliminated stock-based compensation and acquisition-related expenses adjusted for the impact of taxes.

For fiscal Q1, revenue was $100.4 million, slightly above the midpoint of our prior guidance range of $97 million to $103 million, down 0.5% from the prior quarter and up 16.9% year-over-year. On a sequential basis, automotive revenue driven by commercial vehicles experienced a strong above seasonal double-digit percent increase, while IoT revenue was seasonally down. Non-GAAP gross margin for fiscal Q1 was 59.9% and slightly above the midpoint of our prior guidance range of 59% to 60.5%. Non-GAAP operating expense in Q1 was $56.4 million, slightly below the midpoint of our prior guidance range of $55 million to $58 million. Q1 net interest and other income was $2.1 million.

Q1 non-GAAP tax provision was approximately $740,000. We reported a non-GAAP net profit of $5 million or $0.11 per diluted share in Q1. Now I'll turn to our balance sheet and cash flow. Fiscal Q1 cash and marketable securities were $277.8 million, decreasing $34.8 million from the prior quarter, but increasing $18.4 million from the same quarter a year ago. The sequential decrease in cash and marketable securities was primarily due to an increase in our inventory levels to better service our customers in the face of a number of new product cycles. Receivables days sales outstanding of 35 in Q1 was flat with the prior quarter, while days of inventory increased from 99 to 145 days.

Operating cash outflow was $25.6 million for the quarter. Capital expenditures for tangible and intangible assets were $4 million for the quarter. Free cash outflow was $29.6 million for the quarter. During the first quarter of fiscal year 2027, we repurchased 47,798 shares of our stock for a total consideration of $2.4 million or an average price of $51.04 per share. During the second fiscal quarter, Ambarella's Board of Directors authorized a new $50 million repurchase program valid through June 30, 2027, replacing the program that expires on June 30, 2026.

The repurchase program does not obligate the company to acquire any particular amount of ordinary shares and it may be suspended at any time at the company's discretion. We had 1 logistics company representing 10% or more of our revenue. WT Microelectronics, a fulfillment partner in Taiwan that ships to multiple customers in Asia, came in at 6.7% of revenue for the quarter -- for the first quarter. I'll now discuss the outlook for the second quarter of fiscal year 2027.

We forecast a seasonally strong fiscal second quarter with revenue in the range of $105 million to $111 million or $108 million at the midpoint. Sequentially, both auto and IoT revenue are expected to increase with growth in both consumer and CapEx-driven market. We expect fiscal Q2 non-GAAP gross margin to be in the range of 59% to 60.5%. We expect non-GAAP OpEx in the second quarter to be in the range of $56 million to $59 million.

We estimate net interest and other income to be approximately $1.9 million, our non-GAAP tax expense to be approximately $800,000 and our diluted share count to be approximately 44.3 million shares. Thank you for joining our call today. And with that, I will turn the call over to the operator for questions.

[Operator Instructions]. Our first question is from Ross Seymore with Deutsche Bank.

You mentioned earlier about the auto side growing faster than the end market itself. I think you said that end market would be 10% to 15%. In the past, you've given year fiscal year revenue guidance, I think you said 10% to 15% on your last call. How are you thinking about that for this year now?

So yes, I think for the whole year, we're still thinking it's probably 10% to 15%. We're not changing that. And automotive is stronger -- grows faster than the other market.

Okay. And then on the LTA side of things and maybe the one that you announced tonight in the 8-K with the $800 million over time, but more conceptually, how are you thinking about those? Are they going to be guaranteed revenues? I think you said there were potential revenue. How do we build that into the estimates as we think forward for the company?

Right. First of all, we already have a run rate with Hanhwa for the last 15 years, and we know we only take a percentage of their current market share. So we expect with this LTA we're going to gain market share on their annual run rate as well as this is a multigenerational commitment on both sides so that this is -- so we're going to talk about and these 2 generations of silicon that will be codeveloped between these two companies. So I think from that point of view, we believe that it's long period time of commitment as well as gaining market shares from the Haha. But on the same time, if you look at our current ASP, although our corporate ASP is $15 but our CV ASP is a lot higher than that. So if you calculate put all the things together, that's how we calculate this potential $800 million.

Ross, just a little background on Hanhwa it's a major multinational conglomerate with more than $60 billion in annual revenue. It's involved in aerospace, defense, robotics, physical security, life sciences, industrial, Ocean solutions, chemicals, a lot of different things, retail services. And so an important part of this relationship in the intermediate to long term, is moving beyond what has just been the physical security relationship that Fermi described as our current run rate. So we can gain share on that business. But at the same time, the press release talks about, in addition to physical security, things like operational automation, life sciences, robotics, other industrial markets. That's another very important angle of this relationship.

Thank you a moment for our next question. It comes from Tore Svanberg with Stifel.

Yes. Maybe a question to follow up on the LTA and not the Hanhwa one specifically, but how should we think about these sort of folding in here over the next few years Obviously, these could be quite large. I'm sure you can say yes to all of them. is there also going to be potentially some OpEx sharing with some of these customers that you signed LTAs. Any more color you could offer us as far as how we should think about the magnitude and how it's going to be funded over the next 3 years?

So first of all, I think I want to go a little high level and saying a lot of -- most of the LTA discussion is based on two things. One is really when you look at all the fast AI trend and our customers continue to look at the new AI model, the info the higher performance requirement lower power efficiency. So to meet this kind of AI demand, while for the most AGI plication power efficiency is probably the most important thing. So it's getting harder and harder to build a platform of silicon that can address all of these new applications.

So I think that trend really helped our customers think about how to partner with somebody that can build a platform of silicon that can help not only on the security enterprise security, but also other associated market they are trying to address. I think that's probably how the LTA started. And in this kind of LTA, most of the discussion will involve and -- but also we'll develop a product mutually beneficial, right? So I think those are two things that we definitely want to make sure that we develop a software platform that can go across our current silicon platform that offer a complete road map to our customer. And in exchange, they were willing to help us to fund those platforms with , particularly not only on the silicon side but also on the software side.

That's very helpful. And as my follow-up and maybe related to that platform approach for me. So just thinking about the competitive landscape, obviously, there's big processor companies. There's obviously analog companies and so on and so forth. But how flexible can your software platform really be? Because obviously, the use cases at the edge are going to be quite different from use case to use gas.

Yes. So I think our unique architecture, you remember that we have been talking about [indiscernible]. So our -- that out how we accelerate like our image processing and our CVflow AX reader has been -- really passed many battles and have been proved to our customers is not only powerficient, but also program enough to adapt to many different applications, different AI models. For example, which I talk about our CVflow architecture can do 200 different AI -- sorry, model architecture, not 200 models. We're talking about 200 model architectures, and we took all of them into production. Just show you [indiscernible] our accelerators. But more importantly, with the latest agenting AI approach, you need to integrate everything together.

So in that, including not only the ISV and the CV flow or impact income H2 for encoder. On top of that, you need to integrate the complete SoC for CPU and also IO and DRAM activity to provide very -- not only powerficien but also cost-efficient solution to our customers. So that's where we are basically having a reputation to deliver those kind of products consistently in the last 20 years. And when we look at the competitive landscape, with the DLC quota, obviously, but I really don't see any other people coming out with one complete silicon platform that we have 12 HAIC can go from very low performance to few hundred top performance with power efficiency.

On top of that, all the silicon was covered by one unified software development SDK, that our customers, if they develop one product on one so can easily go to different SoC, providing a different price performance point on their product platform. And this kind of flexibility and the width of our product format, I don't think there are money many people can match.

It comes from the line of Quinn Bolton with Needham & Company.

I guess I wanted to follow up for me on the Hanwha LCA. In the press release, you talked about the internal SOC that Hanhwa continues to design. And so I wonder -- it sounds like you're co-designing multiple generations of ships. Is there an opportunity to get a bigger percentage of share away from that internal SoC or do you think that internal SoC continues to hold a portion upon was requirements? And then I've got a follow-up.

I think it's a mutual intention that they're going to use more of this codeveloped platform into more product line. So we fully expect we're getting more market share from Hanhwa with this new development.

Excellent. And then, Fermi, you guys have talked about the fleet telematics market for a number of years, and it feels like it may finally be starting to inflect on the script, you mentioned an installed base of over 100 million units or plus one of that being AI. But what do you think is driving the inflection? Is there anything you can point to in particular that's driving the pretty strong growth here in the first and second quarter?

Yes. Quinn, it's Louis. Just in terms of the market opportunity, telematics, there's about 100 million subscribers and third-party research firms like ABI or Gartner see it growing maybe 10% CAGR. Within that $100 million subscriber base, maybe only 15%, 20% of the market is using AI and AI video as an additional ARPU-generating feature in their platforms. And so what you have is the overall telematics market growing, you've got AI video growing into that.

And at the same time, there's demand for more sophisticated AI workloads and multiple sensors, which is causing our ASP, the demand for more sophisticated AI chips to go up. So those are the dynamics that we're facing. And we're doing a very good job with share. For example, this quarter, we announced [indiscernible], as Fermi mentioned, multiple platforms with CV72,/CB75, which is they're one of the leaders of this market. So any follow-ups on that?

Our next question comes from the line of Joe Moore with Morgan Stanley.

Wonder if you could talk about are you guys impacted at all by shortages of DRAM or storage in any of the surveillance markets or consumer markets? Just are you seeing any impact from any of that?

Yes. So first of all, obviously, just like everybody else, we are impacted, but not directly, we are impacting directly. In the last quarter, when we talked about this, we definitely talk about our customer-facing much higher DRAM price and flash price as well as potential shortage in the second half. So that situation didn't change. The biggest direct impact to us is most of our customers are handling this shortage by doing -- first of all, they are trying to source different DRAM supplier or to cut down the DRAM utilization per product, we basically try to optimize the DRAM utilization in a chip so that they can use smaller DRAM side. And all those activities require we put in our AP support to help our customer. We're happy to do that because that's definitely something we need to do to help our customer. But at the same time, I think that's probably the biggest direct impact to us on the engineering resource requirement.

Okay. And then as a follow-up, the inventory, you sort of described it supporting product ramps, but it was a pretty big increase. I just -- can you give us any more color, make us feel comfortable with that amount of inventory build.

Joe, yes, I mean if you look back in the last couple of quarters, we were running kind of lean. And over the last few quarters, we have looked at kind of the opportunities that are coming to us with the strong growth that we had last year, and we wanted to position ourselves in the right product categories to be able to continue to serve our customers. And so going from a fairly low number of days of inventory at points in last year, we wanted to build a bit of inventory heading into this year to be able to do that.

And Joe, it's Fermi, just want to add another color on this Sensor has officially informed us that their supply is getting tighter. Obviously, we have secured our supply, but we feel it's prudent for us to build up a little inventory just in case we run into a different supply issues. So from that point of view, I think I think you also heard that supply chain become a problem for everybody. We just try to be prudent to build up some inventory just in case.

Our next question comes from the line of Kevin Cassidy with Rosenblatt Securities.

Congratulations on all the great progress. And along those lines, you're building out the indirect sales channel. I think you mentioned 6 groups hired. Do these -- maybe if you could describe these a little better geographically? And then also, do they have application engineers Yes, I'll just stop with those questions.

Right. So first of all, there are many in the United States right now, obviously, but we are definitely trying to go to Japan and Europe to expand our ISV. But today, all six of them are U.S. based. In terms of size of the ISVs, it's really from large to small but more importantly is really what their current engagement with the end customer. We're looking at ISV can add value can easily pull their software onto our platform, and so they can go to their existing customer with our solutions.

So those are ISV we're engaging. And more importantly, we are really enabling applications that we have not touched in the past. So overall from that point of view, we are happy with the progress that we've made in a short period of time since we announced this at CES. But I think 6 of them is just the beginning. I think -- our goal is to double that in this year and hopefully, we can continue to build on momentum based on that.

Yes, Kevin, it's Louis. In addition, for this indirect effort, in addition to the ISVs we just talked about, there's significant effort building out channel partners and even system integrators as part of this effort. And this is really important to help us serve markets like edge infrastructure that need a lot of technical support, a very fragmented markets like robotics.

So a lot of efforts going into the ISV channel partners and even system integrators.

Okay. Great. Yes, that was going to be part of my follow-up question of how much of a -- I hate to use the term, but the cookie-cutter approach can it be? But someone can go in with a lot of the solution already and help out a customer. Is that the idea?

Yes, you might have multiple ISVs working on one project. So if I understood the question right, it's not just always something right off the shelf, cookie cutter, as you said, but you might have ISVs. You might have channel partners and you might have system integrators all involved on one project. So I think it's pretty far from being cookie cutter. It's very sophisticated software as these workloads get more complex. You need that whole indirect ecosystem to play a role in many of these designs we're pursuing.

And Kevin, just to add, part of the decision-making engaging with some of these ISVs is seeing the expertise that they have in different verticals. And some of the -- some of them are -- they have expertise across multiple verticals. And so it's really trying to take a holistic approach to that engagement and cover as many verticals as we can with folks who specialize.

Our next question comes from the line of Christopher Rolland with Susquehanna.

My question is, I think a few quarters ago, you talked about doing some infrastructure, putting your chips basically into servers. I think it was first security camera applications, but potentially other -- have you had any other further engagements there on the infrastructure side?

Yes, we do. Aged infrastructure continue to be a focus, and we have we have not only engaging with the customer, we also have design wins that we're working on. The first product will probably come out, I would say, second half of this year. And in the same time, we are building definitely building a road map to continue to address this opportunity. So we haven't talked about our next-generation chip and also the potential updates on that. But I think next time we should give you more updates on the age infos. That's remain to be a very important direction for us.

Yes. Just to put it in perspective, based on the products we have now it's a couple of hundred million dollar SAM for things like AI Vision Box, but as we come out with more products and build out this indirect ecosystem that, as we mentioned, will help the edge infrastructure business, you'll see us update the SAM appropriately when we have more products here. So definitely high focus area and it's part of that indirect channel we were just talking about.

Great. And then for my second question, just the robotics opportunity. I think last quarter, you mentioned warehouse robotics any other engagements there? And then if you could talk about humanoid and engagements there, that would be interesting, too.

Right. So we talk about different applications. This time, in fact, last time we talked about this, the warehouse design win and now we are definitely in an extensive engineering development cycle with them. But more importantly, on top of that, we have multiple design wins in this quarter in the span of many different applications I talked about in my script. But I think the most important thing is now we focus on that for any robot, there are multiple different solutions today.

One is just purely video, the other one is perception. Three, the third one is using Fusion to put a different sensor together and both one is really because of the controller or decision maker for this whole robust system. Our design win is in the all these areas, but more focused on the perception side because of our expertise in on the video and also fusion side. So I think it's across different applications and also focus on perception and also the decision-making part of the robotic solution.

There are some of the humanoid type of applications in that design wins but we are not in the place to talk about just yet.

One moment for our next question that comes from Martin Yang with Oppenheimer.

First, a couple of questions on LTA. Are you able to say or quantify how many other potential partners you have in your LTA discussion?

We only announced two. We're talking about potential, but there's no other concrete information we can disclose at this point.

Got it. And then also on LTA. So for example, the one with Hanwha spending over multiple years, do you need to also secure wafer supply for your foundry partner regarding those LTA agreement?

First of all, I believe that our relationship with Samsung, although we only try to secure wafer commitment every year. But in the last 18 years, we never have any problem to secure the wafer we want for our customers. So I expect that our wafer supply for 5 or 2-nanometer from Samsung will not be an issue for us. Although there is a lot of discussion that Sensors also gaining momentum on those process node. And obviously, we don't have any long-term contracts with any suppliers, and that might be a question you're asking, but I don't think that's a contract we're going to sign with any supplier anytime soon.

Got it. And last question, also on LTA, do you view your relationship with Samsung and your design capabilities as a key value prop when you try to secure long-term agreements with other customers?

Absolutely. Because one loan discussion with any customer on LTA is how you secure your wafers particularly on the 4-nano 2-nanometer process node. In fact, some of the time, we have to really bring our supplier partners to -- into a conversation to make sure that our customers feel comfortable. By the way, I can say that we -- today, sensor announced their Central has their official company-wide event, talking about the tuna process, and they made an announcement that VDI and Ambarella 2-nanometer customers that come in to their process node as official price release from Samsung.

[Operator Instructions]. Our next question is from Suji Desilva with Roth Capital.

So a quick question on robot robotics as well. I'm curious, are those -- you have a breadth of wins there? Are those across the board on your product portfolio? Or does that category lean towards the leading-edge products, the leading-edge nodes? Just want to understand where robotics is intercepting your product portfolio?

All the products that we have design win is our CV product line. It's really -- and most of that is our 5-nanometer products. There are some 10-nanometer maturities of 5-nanometer products. In fact, they are also our 4-nanometer products in there, too. So it is really covering our CVflow architecture. That's the main reason people are using us, but more focused on the falter for the performance efficiency.

Got it. Okay. And then a question perhaps for John. Should we expect typical seasonality this fiscal year, this calendar year? Or are there factors that might be swinging it differently than prior years?

Suji, I think the expectation is to continue to see the seasonality that we've seen in the past at this point.

Our next question is from Ross Seymore with Deutsche Bank.

Just one follow-up for you. Given the change in the business more automotive this year relative to the IoT side and then longer term, some of the robotics and IoT and Edge AI, physical AI, everything you've talked about on the call, how do you see that impacting either benefiting or being a little bit of a headwind to your gross margin in both the near term and long term from those dynamics, please?

Right. So Ross, in fact, we since 3 quarters ago, we started really watching our gross margin carefully. I think that today, what I can say is based on all the things that we just discussed, we believe that we are going to stay with our current long-term gross margin model that's between 59% to 62%.

And this concludes our Q&A session, and I will pass it back to Dr. Fermi Wang for closing comments.

Thank you for joining our call today, and I really hope I can see you at some of our events this quarter. Thank you. See you next time.

This concludes our conference. Thank you for participating, and you may now disconnect.

Great. Thank you. Welcome, everybody. I'm Joe Moore, Morgan Stanley Semiconductor Research and very happy to have with us the CEO of Ambarella, Fermi Wang.

Thank you.

Fermi, thanks for being here.

Maybe we could start out the AI focus. We talked a lot about edge AI a year ago, but the focus has really shifted to data center, but there's a lot of activity in the edge space, and you guys have pivoted nicely to sort of attack that. Can you talk about that? How -- before we get into the details, just the enthusiasm you have for the edge AI opportunity and what it means to you?

Right. So in fact, that edge AI really developed fast, although data center take all of the investment and focus for AI side, but edge AI definitely really developed very quickly and very nicely for us. First of all, everybody talk about robot today, which is a portion of the physical AI we talk about. We all believe that's going to be a huge market down the road. But today, most of the volume that we are focusing on is still from the enterprise security to the portable video to the drones, those we talk about a lot.

But at this earnings call, we talk another thing, which is very important for edge AI is really we believe. In the past, we talk about edge endpoints, cameras. But this time, we talk about building an AI box, not sitting in a data center, but sitting at the edge, which aggregate all of the different sensor input and apply GenAI type of model on top of that to serve totally different purpose. Just to give you one quick example. For example, at the CES, we demoed with one of our software vendors that they collected the security camera feed to this AI box at a retail store. And with the new GenAI, they can turn this security camera into operational tools that can monitor how the customer comes in, what they buy and collect the customer data. So suddenly, security camera become your operation tool for your efficiency.

So those kind of new applications that you can think about almost all retail store or any other operation can take advantage of the installed base of video feed and using AI box to turn into a different application. That excites us because that really take a very little investment, you can turn on a huge market opportunity. But the requirement is for each application, you need a software vendor to write software on that. We need to have a system integrator to integrate that box into existing infrastructure of the retail stores. And those things need to be -- we need to find partners to that.

So for us, we enable this, we call edge infrastructure business. We are working with ISVs, with system integrators to enable. That's another market. So we feel excited because we're starting more and more edge AI application popping up, and we believe eventually the robots will be there driving huge volume. But before that, we see continuous of new application popping up, and we're going to continue to serve that.

You had some really interesting demos at CES, which is not new for Ambarella. You've had great demos for a decade. What's new about it is the pace of implementation seems really rapid. We talked to guys where you made your software available to them months ago and they're productizing it already versus automakers that have taken many, many years to get to production. Are you surprised by that? Are you seeing opportunities that you weren't aware of as you've opened that up?

First of all, I'm surprised how fast our software vendor can really adopt our platform, but it's by design. 5 years ago, when we start looking at talking to a customer, customer biggest complaint is every time we switch from one silicon to next silicon, I have to port your software over. That's time consuming and waste a lot of resource. Why don't you make your software is portable? And that's where we start coming to do this Cooper Development Platform that will allow us -- our customer to pull application from one of our chip to another chip without too much effort. That takes time for us to build the extra layer so to separate out the hardware layer, but keep all the software layer open so that our customer doesn't need to spend a lot of time to port. It takes a few years to reach that. But as soon as we reach that flexibility, this ISV demo you saw, we give our Cooper SDK to them 3 months before CES, they port the software over from our competitors' platform to us in less than a few weeks and start demoing at CES. So that just shows you when you have this not only a very powerful silicon platform, but a portable software platform that really help our customers to enable new application quickly.

And I shouldn't really say that's new in the sense of Conti and Bosch said the same stuff. It was really fast for them to productize that. Of course, it's taken a very long time for that to turn into revenue, which is a different question. But you've always had that kind of software flexibility built into the system and kind of a software-first mentality to development.

Absolutely. In fact, today, we fully recognize that building 2-nanometer silicon takes time, but it takes even more time for our customers to port software because that's out of our control, make their life easier, it's helping us to get time-to-market.

And what other IoT applications do you see emerging? I mean you've always been near dominant with surveillance camera companies on the professional side, you've always seen a lot of opportunity to do stuff with those kinds of image data. Consumer has been a little harder because of the margins and things like that. But as you see new opportunities, do you think those can open up? What -- and are some of the new opportunities seem like there are opportunities before they are coming back like drones. But just how are you thinking about IoT now as a market?

Well, in fact, for IoT, there's also a market we talked about in the past is a called wearable cameras. In the past, wearable cameras for policemen only. Today, we start seeing the wearable camera goes to the clerk at the retail stores, right? Because everybody wants to document what happened, interaction between the service people to the customers, and that becomes a really big market. But in fact, I want to mention another important application that we never talked about 2 years ago. For example, the edge AI technology going to fleet management. In the past, fleet management is just using GPS to identify where the car is. Today, with AI-enabled camera and the telematics information, not only you can enjoy using AI to help you to detect the environment so like more safer driving environment, but also use that to provide more information about the condition of the drivers, the cars, the products and the storage in the truck. And with those information, they can build even better service to the fleet management customers. So that, for example, Samsara is our biggest customer in that space. We see huge growth from them just in that space, and we're also seeing that almost all the people in that space start adopting edge AI. That's just one simple example that edge AI can apply to almost all the business to help them to get more productivity out.

Okay. Great. Maybe we could talk a little bit about the business. You had a nice growth year in 2025, good product cycles with CV5 and now starting CV7. You sort of started this year with a little bit more of a moderate view of this year's growth. Just how are you thinking about those growth dynamics?

So when we look at fiscal year '26, which is last year, when we started Q1 last year, our first guidance of the year was high teens, but we end up growing 37%. So what happened was that at the beginning of the year -- we were very confident about our own product ramping up growth, and we know that we're going to ramp up the product on time. But what comes to surprise, pleasant surprise to us is our customer ramping up faster than we expected. So that's where we get this really outstanding growth last year. This year, we're standing in a very similar position. Our -- we are very confident in our new product ramping up, including CV75, CV72 and CV7. And in fact, we know that our first 2-nanometer chip will ramp up first half of next year, which we've already seen that. So we are quite confident about our own product ramp-up. Now what we are trying to do is understand our customers ramp up with the new product second half of this year. This will determine how well we're going to perform this year.

So you might be conservative again, but you want to wait and see how that pans out.

You know me. We try to be conservative all the time.

Yes. For sure. Okay. And then maybe just speaking of earnings to clarify the Insta360 lawsuits. Ironically, the night you reported the stock was down a lot on a lawsuit that doesn't seem to have much fundamental impact.

Well, in fact, that somebody -- one of our investors texted us when we're still in that conference, think about that. Fortunately, nobody asked that question because I don't know the answer. But fortunately, after that, we start talking to our customer and also they made a public announcement that there's absolutely no impact to them. Therefore, there's absolutely no impact to us because of lawsuit. They -- obviously we have done a lot of work to work around the patent getting -- the product that involve the patent. So from our point of view, there's absolutely no impact to us.

Okay. Great. That's helpful. I guess going back to IoT, as you think about consumer opportunities, you've had a lot of ramps of drones and then you sort of got priced out of the market a little bit. Same thing on doorbells, things like that. How do you think about that? I mean, on the one hand, the technology requirements in those markets are going to grow as you actually incorporate AI, so you can get those kinds of premium margins. On the other hand, is there a merit to thinking about a lower gross margin model so that you can participate more in those opportunities?

That's absolutely true. We definitely want to play on the lower gross margin to -- as long as they give us better leverage on the operating point. But I also want to point out one of the reasons that people can -- the low-end product or lower-priced product come in because there's no other feature differentiation. Then the price become only matter.

You start putting resources into holding a declining margin.

That's right. So what we are trying to do is we are trying to invest on the market that we continue to see differentiation. AI, particularly edge AI, when we start doing a CV2 family, you need, I don't know, 2 TOPS, 3 TOPS performance. Today, our customers may say need 20, 30, maybe 50 TOPS performance because the applications require more and more AI performance. One is because of application, the other one is because of GenAI model, which is really big, and that requires a huge amount of AI performance. So as long as we can continue to offer differentiated technology and the ASP continue going up, I think that's market we can continue to maintain not only the growth, but also ASP and gross margin. Any market when it happens that the price is the only thing matters, that's where I think we should consider maybe minimize the investment in the market and try to milk it through. But AI, edge AI way ahead of that point yet.

Yes. I mean it seems like when you're incorporating AI features, there's a lot of headroom. It's not like your 4K, 120 frames per second and trying to get to the next thing.

Let me use an example. Our new chip CV7, which is our first 4-nanometer chip and the CV5 is our first -- was our first 5-nanometer chip, CV7 have 2.5x AI performance than CV5. Just within 2 years, our customers are demanding that kind of performance increase. And I can say that for sure that our design-in activity with CV7 is huge in terms of multiple different applications. So you can see that -- we can see that there are definitely applications want to have more and more AI performance and the CV7 hit the right spot.

Great. And other categories within IoT, this portable video and others, and obviously, Insta360 has become a biggest enough customer that we had to worry about that stuff the other day. How do you think about that part of the market? It seems like there's a lot of really cool stuff on display at CES around that.

Right. So first of all, before this kind of the video product coming out, cell phone kind of take over that they become the best video capture devices. But if you look at the product built by Insta360 or DJI for the drone, for the 360 cameras, I think we can argue that they have a much better video capabilities than cell phone can be. So from that point of view, for anybody who are interested in capturing the highest possible quality video or for the YouTuber that want to capture their own production video, that become a popular device. So that's where we see huge growth for the total TAM of this kind of video capture device.

So we feel comfortable that the market is going to continue to grow because the differentiation is on AI on the video resolution side. So we definitely think that's a market we want to continue to focus on. But we also realize that it's having one really concentrated customer might not be the best for the company. And the way to do is not to lose revenue from that, but try to create more revenue source on the different market. And that's exactly what we're trying to do, get more AI application that we can address our current silicon and software solution and try to really dilute that concentration.

Great. So I want to pivot and talk about cars. I continue to be a big believer in your technology for ADAS applications and particularly advanced levels of autonomy. And the market just hasn't really gotten there yet. So I still feel like the opportunity is there. And it seems like you guys also are still investing for some of those opportunities even if you're sort of having the burn not talking about it as much. But just how do you think about it? I mean, ultimately, it seems like we can't just ignore what Tesla FSD is doing, what China cars are doing in the autonomy space and just not start to follow down that path.

Right. So first of all, we definitely continue to work on those autonomous OEM design wins. One of the important reasons is that not only we have technology, but also we -- throughout the process, we learned how to compete in that space. Although that we lost VW, but after the VW case, we get more recognition on the space, almost all the new RFQ, RFI tend to invite us to bid on. So from that point of view, we get better visibility even we lost the VW case. And in fact, that reflects on our auto opportunity we disclosed this time, right? We talk about $13 billion total auto opportunity we identified for the next 6 years compared to what happened last year, this number goes up. That just -- that's a fact reflects that we continue to bid on more projects.

I also believe that autonomous driving is going to be a very important occasion. I think that any design win can drive meaningful -- sorry, a meaningful revenue change. But more importantly, I really think that the software and the silicon investment we put autonomous driving, now we start seeing that can go to robotic, will go to drone application. Any autonomous drone will require similar software and hardware requirements as we develop for CV3 as autonomous solution. So from that point of view, we are convinced that if we want to stay any kind of mobile robot market, our current investment is important. So maybe the combination of that existing business we need to bid on and also future opportunity, we need to continue to have investment on that.

I mean it still seems like the technology that you showed with Continental and Bosch is better than anything that's been implemented in cars so far, at least outside of Tesla and China. So that was 7 years ago, no? It's been a long time.

5 years ago.

Really big breakthrough kind of technology that has yet to just be adopted. And I guess what -- and you referenced the VW situation specifically. What is the challenge? Is it that people in that world are reluctant to bet on smaller companies from a supply chain standpoint? And what needs to happen to close that gap?

Before VW, I really think that our company size is a problem. And for any design win, we need to go in, not only talking to their purchasing people, engineering people, but also the top management to explain why a company that like Ambarella can compete with NVIDIA and Qualcomm in terms of technology, sometimes hard to convince them. But that -- I think we overcome that problem. For example, VW at the end, we talk about this. We lost the deal because our competitor put a financial deal in front of our customer to offset the decision at the end. So we think that if that happens, again, we know how to counter-offer that. And also that we continue to see more and more deal. I think we are in a better position to address this before. But I still believe that working with Bosch and Conti, those large Tier 1s continue to be important for us.

But if you have progress where you feel a high percentage chance that you're going to win something, you're probably not going to tell us about it until it becomes revenue. Is that fair way?

Actually we learned from the lesson. Until we confirm that, we don't...

So the fact that you're not necessarily pointing to the funnel anymore, it doesn't mean that they're not -- there isn't still a vibrant funnel opportunity.

Well, in fact, that our funnel growth -- actually not funnel, the auto opportunity growth, that's just indication that we continue our effort to make sure that we can win something in the near future.

Yes. Okay. That's very helpful. And then this idea, I mean, you can teach a car to drive itself. There's a lot of other things that you can teach robots to do and devices to do. I guess how much have you thought about those opportunities and the focus on the humanoids robots and all the super futuristic stuff. But when you talk about military applications, defense applications, it seems like your technology would have really viable use cases in those markets.

Right. So anything -- in fact, any mobile robots can take advantage of our technology, both on hardware and software side. So all the applications you talk about, definitely that can use our technology. I want to emphasize is that moving forward, if you want to do any like humanoid, although I think it's going to be far away from the high-volume production, you need to have a very powerful GenAI type model, end-to-end model to control robots. And just from that point of view, my personal belief, humanoid is more complicated than Level 4 cars. Level 4 car drive in an environment is well controlled. You have traffic light, you have sign, you have things. humanoid, you have no limit working environment. Anything could happen. So from that point of view, the model you need to develop so to guarantee not only it works and functional, but also has safety concerns, security concern, all of that, it makes things very difficult. So what I try to say is that's a road map. I think we talk about what kind of market we want to invest. We want to invest -- there's a road map that we can continue to offer differentiated technology. That's where we want to be.

Okay. And your software-first approach is helpful at driving those opportunities?

Absolutely. In fact, that you really don't want to put a water cooling system into your robots, right? You have to have be power efficient. From that point of view, that you have to make sure that you design a chip for this application, right? Obviously, that people are going to using whatever they can access to, most like GPUs to prototype or generate first generation, second-generation product. But when you come to really mass production when the power efficient matter, when the cost matters, you need to have a chip that designed for this application so that it can minimize not only the power, but also minimize the cost. And that's -- and also minimize the DRAM, which has become so important this year. So all of that is the reason I think that a chip that designed for specific application will win at the end.

And you talked about a warehouse robotics win. That's a market that actually historically hasn't used as much vision maybe as what we're starting to see. And you've seen cobots and things like that don't even have vision capability. Is that changing now? Are you starting to see the need for that?

The particular design we mentioned in our earnings call is using a video as a -- and basically, the chip aggregate multiple video camera sensors and along with other sensors and do a sensor fusion for the perception. So that's exactly the usage for this particular application wins. When I talk about robotics, we offer three types of different products. One is just one single camera using to detect object, which is very simple. The second product we are offering is a perception box that taking multiple inputs and doing a sensor fusion, then you identify the environment object. That is the product we talk about in this. The third one is really, I think, the final solution, which is domain controller using one single chip to control the whole robot, not only perception, but also path planning, controller movement, functions of the moving arms. All of that within one controller, that's where is not only you need a powerful AI processor, but you need an end-to-end software model. That's where I think it's...

It's sounds a lot like CV3.

It sounds like CV3 but having more higher performance than that. So we are talking about the design win we just talked about is really about perception box sitting in the middle.

Okay. And does it require -- I mean, you sort of talked about a full stack approach in automotive, but this is going to be a much more fragmented customer base. Does that require allowing them to do more of the heavy lifting on the software side? Or how do you go to market in these kind of more fragmented spaces?

First of all, I think we want to demo. We need to have a software that we can demo the basic function like perception, planning, movement, all of that. But however, like you said, each robot requires different controllers. We -- there's no chance we can develop all applications for all the robots. So what we need to do is really working with our software partners, system integrators so they can help each customer to optimize the software and the models. And that is a new business model we talk about at CES. We want to work with a partner like ISVs, system integrators or the OEMs to build a complete solution for customers, and they can maybe even add their own software on top of that. This kind of overall development platform need to be ready, not only for robots, but also for edge infrastructure box that we just talked about.

Okay. And can you talk about semi-custom ASIC types of products in the context of all of this?

This discussion started with our first 2-nanometer chip that we taped out at the end last year. And when we started this project, one of our customers approaching us, hey, I'm interested in doing a semi-custom chip with you and by adding some of the special sauce into our chip, but they don't want to pay for the whole.

I was going to say they know what a 2-nanometer [indiscernible] cost?

The reason they came to us because nobody wants to afford that 2-nanometer chip payout. However, so the trade-off is they pay for a portion of the fee, but allow us to use that chip to sell to people -- to customers not competing with them. So those arrangements definitely is a win-win because that cut back our R&D cost to develop the first 2-nanometer project. And since then -- since we started talking about this, many customers and many new customers come to us say, we are interested in similar deal. But then they become us is what's our ROI, how do we determine who to engage, who not to engage and how we -- is that a real business for us? We are at the stage now we need to make a decision by engaging maybe 2 to 3 customers to size up the opportunity, understand how difficult it is and also go through the ROI to make sure that we build a business that we can justify in the long term. If we decide to do -- this is going to be a long-term business for us, then obviously, we can ramp up our go-to-market team as well as VLSI team. But most important thing is that we need to make sure any customer come to us, they want to leverage first, our IP, our video processing IP, our AI inference engine, our low-power technology as well as 2-nanometer technology. So if anybody wants to leverage those combinations, it's a good customer. But at the end, it's ROI decide whether that's going to be a business we get to engage or not.

And how are you thinking about 2-nanometer in the sense of CV3 was a really expensive chip that didn't ever get the traction that you wanted? And 3-nanometer, do you have to replicate that investment at 2? Or are you able to sort of focus on these more project-oriented or sort of defined application-oriented 2-nanometer products?

For 2-nanometer, we definitely try to -- when we tape-out any 2-nanometer chip, we need to identify not only just a customer but potential sales, right? So we have been very careful about defining what kind of 2-nanometer chip we want to build. There's a guaranteed customer and guaranteed market for us to enable 2-nanometer. So from that point of view, we are not going to do another automotive chip at 2-nano anytime soon. But we are going to focus on the areas we know there's a customer demand and also have a high-volume potential already. So I think from that point of view, 2-nanometer has become I think important for the power-efficient solution, if you really care about lowest possible power for robotic application for any other edge AI application, 2-nano become must-have.

And the state of your 2-nanometer comes from Samsung, you're very confident in the process technology underlying it.

We are getting a lot more confident than before. First of all...

Does it help that Elon is putting a lot of volume through...

Of course. When Elon announced it, I feel so happy because I don't need to be the only one defense [indiscernible] anymore. But however, I think throughout the last 12 months, working closely with Samsung Foundry on this process technology, we get to a point that we believe their yield will be not only acceptable, but also good enough for going to production. We are going to go production first half next year. I think that will happen. But more importantly, it's not only us saying that, there are multiple other large companies are saying that. So I think from that point of view, we are getting very comfortable that, that will happen. And more importantly, working with Samsung become an advantage to us because that TSMC has been telling people they don't have enough capacity for everyone. Samsung not only have enough capacity for us, we secured all our capacity that required for 2026. And we also secured the 2-nanometer capacity for 2027. So from that point of view, I think that Samsung definitely help us to give the confidence to our customers that they don't need to really worry about the supply chain for many different reasons.

Yes. That's great. I just have one more question and open it up to the audience if there are any. I mean just you've reinvented the company, I think, really for the third time on all of this, starting out is sort of video processing and then incorporating AI capabilities and now sort of pivoting more towards the edge. I guess what is that -- where is the endpoint of that? What do you think the business mix looks like a few years out? I think automotive, given the SAM's commentary is still going to be a big part of it. But how are you approaching resources to automotive versus these other edge AI markets?

So even for automotive, I want to make clear that we are focusing on the project that can generate revenue. In the -- at the beginning, we focused on developing technology and invest heavily. But now we kind of reposition that for automotive, let's focus on the project that we know is going to generate revenue. That is a different -- huge change for us. But I think it's more important to address your first question is what's -- how do we see this company going to evolve in the next 5 years? I really think that edge AI will become a lot more visible in the next 5 years. Today, 95% of AI investment go to data center. I think that will change. And that will change because that you need to deploy real-time location based on AI. Everybody talking about even OpenAI says they want to build the wearable device. Guess what, that's an edge AI device, right?

So I think that we still don't know what's the most -- the killer app for edge AI, yet we start seeing a lot of smaller market, but we believe that will happen. So from that point of view that we will continue to invest heavily on the edge AI side, not only just on the video. Video is definitely our cash cow and we're going to continue to milk -- develop project on that. But we also see that on the digital AI going to AI box to aggregate. That's also a huge investment. But edge AI will be the key platform and key area we're going to continue to play.

That's great. Let me see if there's questions from the audience. If not, we can wrap it up there. But Fermi, congratulations. It's been a really successful year for you and everything you've achieved is really putting the company in an interesting place.

Thank you, Joe. Thank you very much. Thank you, guys.

Thank you.

Good day, and thank you for standing by. Welcome to the Ambarella Fourth Quarter and Fiscal Year 2026 Earnings Conference Call. [Operator Instructions] Please be advised that today's conference is being recorded.

I would now like to hand the conference over to your speaker today, Louis Gerhardy, Vice President, Corporate Development. Please go ahead, sir.

Thank you, Michelle, and good afternoon. Thank you for joining our Fourth Quarter Fiscal Year 2026 Financial Results Conference Call.

On the call with me today is Dr. Fermi Wang, President and CEO; and John Young, CFO. The primary purpose of today's call is to provide you with information regarding the results for our fourth quarter of fiscal year 2026. The discussion today and the responses to your questions will contain forward-looking statements regarding our projected financial results, financial prospects, market growth and demand for our solutions, among other things.

These statements are based on currently available information and subject to risks, uncertainties and assumptions. Should any of these risks or uncertainties materialize or should our assumptions prove to be incorrect, our actual results could differ materially from these forward-looking statements. We're under no obligation to update these statements.

These risks, uncertainties and assumptions as well as other information on potential risk factors that could affect our financial results are more fully described in the documents we filed with the SEC. Access to our fourth quarter fiscal year 2026 results press release, transcripts, historical results, SEC filings and a replay of today's call can be found on the Investor Relations page of our website.

The content of today's call, as well as the materials posted on our website or Ambarella's property and cannot be reproduced or transcribed without our prior written consent. Before starting the call, we hope to see you at one of the following investor events scheduled for our first quarter of fiscal year 2027.

March 3 will be at Morgan Stanley's TMT Conference in San Francisco. March 10, at Loop Capital's [ seventh ] Annual Investor Conference in New York, March 10 to 12 will be an embedded world in [ Nienburg ], Germany, and we're offering a limited number of investor meetings. March 11 at [ Amtor's ] Global Technology and Industrial Conference in New York. We'll be hosting bus tours at our Santa Clara headquarters with Instinet Nomura. Pate, on March 12, 18 and 20, respectively. March 16 at Bank of America's 2026 Asia Tech Conference in Type A and March 24 at the Roth Conference in Dana Point. As a reminder, we'll enter our first quarter quiet period on April 16, 2026.

Fermi will now provide a business update for the quarter John will review the financial results and outlook, and then we'll be available for your questions. Fermi?

Thank you, Louis, and good afternoon. Thank you for joining us on our call today. Fiscal 2026 established a new revenue record for Ambarella. Revenue increased at 37% year-over-year, well above the growth in the mobile semiconductor industry and most of our semiconductor company peers. Our 5-nanometer new product cycles, together with our customers' new product launches combined to drive 50% year-over-year growth in our Edge AI revenue. About 80% of our full year fiscal 2026 revenue is Edge AI, all of which is also defined as a fiscal AI.

Overall, [ auto ] and IoT revenue both grew with company-wide growth in both units shipped and averaging selling price. Our fourth quarter revenue results follow a seasonal pattern with revenue down 7% sequentially, slightly above the midpoint of our original guidance. Our new third-generation 5-nanometer [ CV 75 ] and the [ CV72-AIoCs ] are rapidly growing, reaching a high single-digit percent of total revenue in Q4 and these new products are poised to be an important source of incremental revenue in the new year.

Looking further into fiscal 2027, we anticipate the total revenue growth in the 10% to 15% range, with non-GAAP gross margin within our long-term model over 59% to 62%. For the year, we expect our new product cycle to continue to drive both units and [ average ] selling price increase with revenue growth in both [ auto ] and IoT. In addition to the anticipated revenue ramp from CV75 and the CV72, the recently announced CV7, our first 4-nanometer chip is expected to begin to generate revenue in the fourth quarter of this year. By a variety of measures, our team's achievements in the last year have strengthened our Edge AI leadership and we continue to enhance our market position.

Financially, in fiscal year 2026, we continue to commercialize our AI investments and deliver premium revenue growth returning to full year non-GAAP profitability. Fiscal 2026 was our 17th consecutive year of positive free cash flow with free cash flow for the year of a $58 million or 15% of revenue. We executed to both our operational and R&D priorities. While facing a variety of industry-wide supply chain constraints, we shipped more than 25 million units across more than 15 SoCs with many variants. And we taped out our first 4-nanometer chip [ in ] our first 2-nanometer gate or around AI SoCs while successfully bring CV75 and the CV73 mass commercialization.

Our Cooper Development platform, while already powerful and well established is in a constant state of enhancement, including new agenetic capabilities. Strategically, we announced during our [ CES ] 2026 product and pilot technology briefing, we are augmenting our direct-to-customer go-to-market with incremental initiatives, we expect to materially contribute to our long-term revenue growth.

First, we are incrementally building an indirect sales channel during including independent software developers, distributors and system integrators. We expect this to improve our ability to address the aged infrastructure market, as well as the highly fragmented robotic market. Furthermore, in the long run, our existing portfolio should benefit with long-tailed revenue from small to midsized customers, we have not directly supported in the past.

The second strategic development is the establishment of a semi customer [ custom ] ASIC business, where we have strong interest from a variety of companies. Our deep intellectual property, perception engines, AI accelerators, software development platform and advanced [ VSI ] capabilities and established position in the Edge AI market are increasingly valued by companies considering [ semi-custom ] or customer [ ASIC ] projects.

Stepping back for a minute, [ there ] continue to be significant industry development with AI agents, applications, content, models and the services that, when combined with our enabling AI SoCs create an environment where more age and physical AI use cases can practically emerge. Technics developed in the industry such as distillation and a mixture of experts are enabling age models to become smaller, yet smarter, which we expect will enable complications to evolve from early [ adopter ] adopters to the mainstream.

Altogether, we see a variety of enterprise and consumer Edge AI system become real time, proactive and able to make closed-loop decisions autonomously for the end users through [ gentex ] of course, with all of breakthroughs, our customers have a lot to learn and evaluate as they consider new AI business cases. The various components of our comprehensive corporate development platform together with our engineering support are enabling customers to implement new technologies.

For example, a power constrained application may need a hybrid AI workload split between cloud and age. But in other cases, where no latency is acceptable, we need to support a vast majority of AI processing on our [ silicon ]. Overall, you can see there are many different Edge AI applications, use cases and the trade-offs we must support and our broad Edge AI products portfolio and established powerful development platforms are must [ haves ] to drive the proliferation and diversification of the age and the physical AI market.

I will now discuss some representative customer engagement during the quarter. I want to start by highlighting our industrial automation robotic design win as [ the ] warehouses of a large U.S.-based e-commerce provider. They leverage our N1-655 AI SoC to develop a perception hub for the warehouse floor. A fleet of this system is being deployed to enable high-speed, accurate, and efficient storage and the retrievable system at their large-scale warehouses across the country. We are seeing several such physical ad designs starting to emerge on our SoCs.

In other IoT applications, we were awarded several projects in the video conference market this quarter. Insta360 launched their Link 2 Pro and Link 2C Pro high-end web camera based on our H22 SoC in the [ QHC ], a cloud-native audio, video and control ecosystem company based in California announced their [ QSI ] high-definition video conference in PTZ camera designed on our CV72LC. They are leveraging our AI ISP for enhanced video quality and use AI for face detection and intelligent presenter tracking -- in Enterprise Security, Idis, a leading security technology customer announced their [ DCD-31 ]6a security camera based on our CV72SoC. And our customer Dallmeier based in Germany launched their [ DomeraRDF- ]6140 done camera based on CV25 this quarter. They leverage our AI accelerator to offer several AI features like motion detection, temporary detection, intrusion detection and the line growth.

Finally, one of our leading customers, IQ site previously known as [ Bosch ], announced two new AI products, both based on CV72. The flagship on [ 100 ] noise, the [ image ] inside the camera will enhance privacy and compliance and the deep [ 100 ] detect people and the vehicle accurately with maximum detail in dark low-light conditions. In our automotive safety, ADAS and telematic business, I would like to share some key customer wins during the quarter.

Ford recently launched the dealer fit truck [ bed ] camera last quarter, it's a smart security camera for the truck [ and ] built on our CV25. It provides real-time truck bed monitoring, leveraging AI-powered intrusion monitoring and threat detection. [ Thinkware ] system in South Korea launched their [ QXD2-incortage ] to video recording system, which is the first of the kind to leverage our AISP new network on our CV25 SoC think we also use Ambarella's ADAS software stack to enhance perception capability for their forward-facing ADAS. Garmin announced their innovative due view based on CV25, it's a rugged [ 2 ] camera systems that enable professional truck drivers and adds in situational awareness.

In summary, this 11 representative customer engagements represent the implementation of a wide variety of applications and AI workload. Inherent in these wins is the high degree of programmability and the flexibility in our SoC and software platform, enabling us to serve a wide variety of applications with minimal incremental investment, while the customer benefit by having the ability to reuse their software and scale.

While we are seeing Edge AI [ green ] shoes emerging a very diverse range of Edge applications, we currently see the largest long-term growth opportunities in the robotics, automotive and edge infrastructure markets. The robotic market is a diverse market in a variety of [ applications ]. Fixed factory automation, humanoid, mobile texture, aero [ Jones ] and more. We are already shipping into the fixed factory automation market and Q4 was our first full quarter of a production revenue from the aerial drone market, which we believe is one of the highest value mobile robot market today.

With our industrial automation, robotics engagement announced announced today, we are establishing ourselves in yet another form factor in a diverse and nascent robotic market. In the automotive market, we have two business. One safety [ panematics ] ADAS business, which represents most of our revenue and a majority of our near-term growth opportunity in autos and also our [ autocom ] business starting at [ Level 2 ]+, which offers long-term growth opportunity.

At this time, the [ auto ] opportunities we have either won all being invited to bid upon in the next 6 years, from fiscal year 2027 to fiscal year 2032 is approximately $13 billion, with 1 proportion similar to the last year. In the aged infrastructure market, we are observing early customer opportunity with 2 different design architectures, one fiscal AI and a second digital AI.

First, enterprise buyers want to run [ physical ] AI year influence on a local edge gateway to aggregate multimodal data from multiple sensors preprocessing is in real time for use cases such as fleet management, physical security, industrial robots that typically designed fully self-efficient agent solutions to process data locally on devices for real-time low latency and secure decision-making that can be summarized and sent to data centers for training and analytics.

Second, we see early customer opportunities from enterprise IT buyers for digital AI application that push centrally trend and high-capacity models to be distilled, quantized and deployed in [ age ] nodes to enable low-latency closed-loop automation for secure digital applications while still maintaining centralized control in the cloud.

In summary, we are an Edge AI market leader across a broad set of criteria. First is our credibility. We have an installed base of 42 million Edge AI SoCs with more than 370 unique customers customer products reaching production and approximately $1 billion in cumulative Edge AI revenue primarily from our second-generation CV2 family. Next is our portfolio breadth. We have a 12 Edge AI SoCs, supporting models ranging up to 34 billion perimeters. We support up to 100 billion parameters in the future covering the full branch of Edge AI applications.

Finally, our development platform is established a critical enablement tool. The Cooper Development platform scales across our Edge AI portfolio and multiple applications with customers implementing and reaching production with more than 200 different model architectures.

In conclusion, I'm very proud of the resilience, commitment and execution of our team in the last year. I'm very excited about our prospects in fiscal '27 and the years ahead. We are committed to our Edge AI strategy and driving earnings growth.

With that, John will now discuss the Q4 and the fiscal year 2026 results, as well as the first quarter outlook in more detail. John?

Thanks, Fermi. I'll now review the financial highlights for the fourth quarter fiscal year 2026, ending January 31, 2026. I will also provide a financial outlook for our first quarter of fiscal year 2027, ending April 30, 2026.

I'll be discussing non-GAAP results and ask that you refer to today's press release for a detailed reconciliation of GAAP to non-GAAP results. For non-GAAP reporting, we have eliminated stock-based compensation and acquisition-related expenses adjusted for the impact of taxes. Fiscal year 2026 revenue increased 37.2% to $390.7 million. Automotive revenue led by telematics increased in the high single digits and IoT increased almost 50% year-over-year, led by portable video and a continuation of strong growth in physical security.

For fiscal year 2026, non-GAAP gross margin was 60.7% versus 62.7% in fiscal 2025. Non-GAAP operating expense increased 12.9% for the year versus 6.5% in the prior year, driven by higher costs related to employees and SoC Development projects. Ending cash and marketable securities totaled $312.6 million, up from $250.3 million at the end of the prior year, driven by free cash flow of $58 million for the year or 14.8% of revenue.

For fiscal Q4, revenue was $100.9 million, slightly above the midpoint of our prior guidance range of $97 million to $103 million down 7% from the prior quarter and up 20.1% year-over-year. Sequentially, automotive and IoT both experienced a similar seasonal decline. Non-GAAP gross margin for fiscal Q4 was 59.8% at the midpoint of our prior guidance range of 59% to [ 60.5% ].

Non-GAAP operating expense in Q4 was $56.5 million, also at the midpoint of our prior guidance range of $55 million to $58 million. Q4 net interest and other income was $2.3 million, Q4 non-GAAP tax provision was approximately $551,000 and we reported a non-GAAP net profit of $5.5 million or $0.13 per diluted share in Q4.

Now I will turn to our balance sheet and cash flow. Fiscal Q4 cash and marketable securities reached $312.6 million, increasing $17.3 million from the prior quarter and $62.3 million from the same quarter a year ago. Increased cash and marketable securities benefited primarily from operating cash flow associated with increased revenue. Receivables days sales outstanding of 36 in Q4 was flat with the prior quarter.

Days of inventory increased from 76 days to 99 days to support our current level of business. Operating cash inflow was $18.9 million for the quarter and $73.5 million for the year. Capital expenditures for tangible and intangible assets were $3.9 million for the quarter and $15.5 million for the year. Free cash flow was $15 million for the quarter.

During the second quarter of fiscal year 2026, Ambarella's Board of Directors approved an extension of the current share repurchase program for an additional 12 months ending June 30, 2026. In the fourth quarter of fiscal year 2026, the company did not repurchase shares. During the first quarter, we repurchased 24,152 shares of our stock for total consideration of $1 million.

As of today, there is approximately $48 million available under our repurchase authorization. We had one logistics company representing 10% or more of our revenue. WT Microelectronics, a fulfillment partner in Taiwan that ships to multiple customers in Asia, came in at 73.1% of revenue for the fourth quarter and 69.7% for the year.

I now will discuss the outlook for the first quarter of fiscal year 2027. We forecast Q1 revenue to be seasonal and in the range of $97 million to $103 million or $100 million at the midpoint. Sequentially, auto revenue is expected to increase with IoT revenue expected to be seasonally down. We expect fiscal Q1 non-GAAP gross margin to be in the range of 59% to 60.5%. We expect non-GAAP OpEx in the first quarter to be in the range of $55 million to $58 million.

We estimate net interest and other income to be approximately $2 million. Our non-GAAP tax expense to be approximately $800,000 and our diluted share count to be approximately 44.1 million shares.

Thank you for joining our call today. And with that, I will turn the call over to the operator for questions.

[Operator Instructions] Our first question will come from the line of Quinn Bolton with Needham & Co.

Congratulations on the results. I wanted to ask for maybe a little bit more detail on the e-commerce warehouse robotics win that you discussed in your script. Can you give us a sense, is this already in production? If not, when would you expect it to go to production? And how many warehouses or perhaps how many robots could you guys be participating in for this customer? Is it a meaningful opportunity?

First of all, it's in production, [ although ] the low-volume production right now, but we definitely expect that will continue to grow and we think it can be meaningful, it depends on how wide they [ go ] to their warehouses.

In terms of the function that we're doing is really, in fact, I said, it's a perception in the warehouse to help them to [ do ] automation for the -- from the production and also the product movement. I think this is significant because that's the first such a design win for us, although we are not allowed to talk about the NIM and also the size opportunity, but we think this is definitely an indication that our perception system that have been respected and used in this large organization.

Yes, I imagine it could be a nice flagship customer that could lead to some other wins as well. So congratulations on that. The second question I had is you gave us sort of the update on the auto pipeline now standing at $13 billion. I believe that, that's sort of an on probability-weighted number. In the past, I think you've given us a $2.2 billion probability weighted forecast. I'm just wondering if you look back at the last forecast that was profitability weighted, if you unweighted it, could you give us sort of an apples-to-apples comparison as to whether that auto pipeline has grown over the last year?

Yes. So first of all, we call automotive opportunities now just trying to differentiate what we have been doing in the last several years. So first number we call is the total size of $19 billion. That's involved all of the business opportunity that we see in the next 6 years that we have either won or being invited [ to ] bid.

And compared to last year, this, we do see the growth in this category. On the one business that we see the numbers similar to last year was an apple-to-apple comparison. And -- but also I want to highlight one thing, although that the one business is flat, but considering the weak automotive market in 2025, we are very happy to see the end result because I should [ not ] we see more opportunity in the total automotive opportunity side, but also we continue to add new design wins to compensate for the -- for example, a lot of customers cut their forecast or delay the production, but we continue to maintain a healthy design win momentum in automotive.

And sorry, just a clarification for me. Did you say that the total pipeline is [ $133 ] billion or [ $199 ] billion

It's $13 billion, sorry, I it must be a -- so [ $13 billion ] with a total opportunity.

Our next question comes from the line of Tore Svanberg with Stifel.

Congrats on the record revenue year. For me, I was hoping you could maybe help us look for, I guess, guideposts on two particular topics. One is just your channel strategy. How is that going? Are there certain things that we should look out for 2027 and then on your semi-custom [ ASIC ] business, again, any specific things that we should be keeping an eye on and what are perhaps some of the early applications you think where you would potentially get an ASIC design win?

I think the CS, we we talk about our new go-to-market strategy and also highlight several milestones we want to achieve. I think the first year, our goal for go-to-market -- this new go-to-market strategy to focus on buildup with our partners, particularly in our ISVs and as well as assisting the greater and distributors. We think that we are shooting for get a dozen of ICV committed to our platform and by the end of the year so that they can help us to drive multiple different applications, different customers at the same time.

So also, we are targeting at this [ state ] certain milestones with distributors and the system integrators, the milestone for the first year. So you should expect us to continue to make progress on that. I think that -- but however, revenue probably -- I'm not expecting any meaningful revenue this year from this new business model, but we expect to start seeing maybe ramping up a little bit in the next year.

In terms of [ Consecana ] AC business, we only talk about our [ first ] 2-nanometer chip is in this business model and it's in the IoT space. And we are current engagement show multiple companies are interested in this model. And I won't be surprised that we continue to announce new design win in this category.

But so far, only the first one is being confirmed and we'll announce it. So you -- what you should expect is when we get new design wins, we will give you a hint that we definitely win something, but maybe we won't disclose the customer name or the business, but we should give you a hint that we continue to make progress in this business.

Very good. And as my follow-up, and on the 10% to 15% growth guidance for fiscal '27. I know in fiscal '26, obviously, IoT outgrew automotive by quite a bit. Just wondering how you think about the mix in fiscal '27, and I assume that 10% to 15% assumes both unit growth and obviously, also continuous ASP growth.

First of all, your assumption is right. Both ASP and unit growth is there and also that we believe both IoT and auto [ will ] grow. But I want to add a little bit more color on this -- our growth rate [ when ] we look at fiscal year '26 growth of 37%, it comes from two areas. One is our new product ramp up at the time and also that our present surprise a strong customer new product ramp up in fiscal year '26 that combined generate this growth.

In this year, we are very confident that we're going to continue to write down this momentum, and we are confident about our own new product ramp-up like [ CB72, ]-CV75 and CV -- but we are trying to understand is working with customers to understand their new product ramp-up and how they're going to impact our growth in this year.

Our next question is going to come from the line of Kevin Cassidy with Rosenblatt Securities.

Congratulations on the good year. Just what are you seeing in the competitive landscape as you're getting into the [ drones ] and I guess we passed the point where companies are trying to build their own devices, and we'll prefer to work with you. for the AI capabilities and just what I'll see as competition say coming from China.

Right. You're talking about China specifically. So first of all, in the drone market, [ continue ] to build their own [ silicon ], but they also use external silicon solution to complement their product portfolios. And outside that, I think the majority of other drone market will be -- they don't plan to -- a list, we don't know anybody plan to build their own [ solanefinitely ] to try to use external silicon, particularly that if you look at our offering to drone market, is from 5-nanometer to 4-nanometer then will be 2-nanometer. And from that point of view, I think that will uniquely position us one of the few that can provide to the Chinese market.

Great. And interesting, with the ASIC market with AMD and Meta announcing a partnership earlier this week, part of the discussion was that Meta had certain models that they want to run on a semi-custom version of AMD's [ 450 ] and to me, remind me of your design where you have algorithm first type of application. [ So ] or the way you made your CV design in the first place. So is that where you're finding applications for the -- for semi-custom version? Is it for certain models for running what the customer is looking for and optimized SoC?

I think that's one of the areas our customers want to leverage on, but I want to highlight most -- in fact, most -- in fact, all of the customers that we are engaging for this business model is trying to leverage either our CV4 AI accelerator because of performance and performance efficiency.

All our ISP, which is using a lot of AI performance; third, or our software platform that they can easily leverage to quickly go to market with a new part of our new models. Fourth, and also as important is our capability to take out a 2-nanometer ship. I think our customers are trying to take advantage of a combination of these four factors as the reason to talk to us. By the way, we are not targeting at all for the data center design, and that's not a warehouse strength. Our strength is some of the customers want to build AI associated with their own algorithm that creates our sweet spot.

Our next question comes from the line of Joe Moore with Morgan Stanley.

I heard you reiterate the 69% to 62% long-term gross margin. I just wonder if you need to rethink that at all with the focus on different markets, anything that would pull you out of that range one way or the other? Just any color.

So first of all, we repeat to say this year, our loan -- our gross margin will be within our long-term gross margin of 59%, [ 62% ]. And at [ CES ], I also mentioned that when the [ Hobson ] semi-custom design become more mature, if we need to change the model because of that, we will come to talk to our investor world about this. But today, I think that because that new business model is still at the early stage, and we're still talking to customers from a different business model. I think that's premature to talk about this in terms of gross margin for the impact for business. for our existing ongoing business, we will continue to feel comfortable that we'll be at 59% to 62%.

Our next question is going to come from the line of Vivek Arya with Bank of America.

This is Liam Pharr on for Vivek. Wondering, are you seeing any or expecting any impacts or benefit from the recent restrictions of a Chinese competitor in the drone market?

Well, we're definitely watching it. So definitely, I think that's something we are talking to our customers. I think that it's not clear also our current design win with that all in production is not impacted by the new regulations. So whether the next-generation impact is really depends on that they were going to file for FCC review. So there's a possibility that will be impacted. However, I want to point out that outside the U.S., there's still huge drone market that we can tap into, not in China, but in outside U.S., that's still a very big market that we can work with.

So I think overall, the answer is no direct impact right now, but we're watching the potential impact in the future.

And then are you seeing any impact on the overall demand environment from component cost inflation?

You're talking about DRAM. So yes, first of all, there's obviously no direct impact to us, but we talk to a lot of the customers. In fact, all of the customers about this issue. It's very clear that majority of them have concerns about the price increases rather than the shortage of the component.

In fact, I think most of the companies that we talk to still can fine supplies, but at a much, much higher price today. So indirect impact, in fact, in my opinion is for the product, which has a very low gross margin, which cannot sustain the DR cost increase what will impact the most? If you look at that from our customer portfolio, that means, in fact, it's very low-end business, which we don't have much at all. So I think for our point of view, we don't expect huge impact because of DRAM price at this point. But we remain to watch this because it changes so quickly and so dynamic, and we want to make sure that we don't overlook this potential impact.

[Operator Instructions] Our next question will come from the line of Martin Wang with [ OpCo ].

My question is on seasonality in relation to your [ CD7 ] launch in the latter half of the year. Do you think that initial launch could change for seasonal patterns a little bit? And also how that -- how should we think about the overall ASP uplift for the year versus FY [ '2 ]6.

Right. So first of all, we expect the ramp-up in the fourth quarter this year, but we don't expect material revenue generated by [ C7 ] this year. But however, we highlight C7 because for 2 reasons. One is is our first 4-nanometer chip and higher [ 2.5x ] higher performance -- AI performance than CVV.

So from that point of view, that we see huge interest and in fact, many design wins already engaged and some of them will be ramping up in production later this year is significant for us. That means that [ conform ] thesis that our customer has huge demand and appetite for higher AI performance for the applications, which is very encouraging to us.

So in terms of ASP, we expect there is a premium ASP compared to current [ ASP ], but we haven't finalized all the negotiations. So I think as an indication, that's just an indication of what we are looking at in terms of total ASP for [ CD7 ].

Our next question comes from the line of Gus Richard with Northland Capital Markets.

As you move into the ASIC business and an [ Inderachannel ], I was hoping you could discuss a little bit about how that's going to change the P&L. Indirect channel, you're going to have [ slightly ] lower volumes, higher gross margin and maybe higher D&A to go with that. In the ASIC business, do you get paid for the NRE because that necessitate lower unit cost, the lower gross margin on the units. Can [ you ] just kind of talk about how you think that's going to play out over time.

So first of all, I think it's a little too early for us to talk about the business model for the new go-to-market strategy. we equally need to come back to you to talk about this, but considering there is no revenue generation from that this year. I would like to delay that discussion a little bit. But your question on the ASIC side is important for us.

First of all, it has to have LIE associated with those kind of projects. Otherwise, it doesn't make sense to us to [ disc ]. But however, also there are all kinds of different variables we play with, for example, some customers [ want ] to integrate their black box IP into a chip, somebody wants to have a special IO design for their application. So it's a huge variety with demand, but -- at the end, we need to have we are willing to look at different ASP structure to make whole overall business making sense for us and for our customers.

And the first product that we talk about. We always talk about the significant amount of [ NRE ] that they are paying out right now. And the first revenue generated for silicon for the first [ semicontiASIC ] is going to be early next year. So in terms of the gross margin impact, I think there's small, we still believe that overall, if you average out the whole business in that first silicon that our gross margin is still within our long-term gross margin.

But I also believe that to exchange for more aggressive [ NIE ], that business model might change for others in the future. So because it's really uncertain. I don't want to talk about it, do we don't want to give you an indication just yet. I just want to tell you that it's a variety of possibility, and we are willing to talk with the customer that want to work with us. Obviously, at the end, it has to be beneficial for me, for Ambarella as well as for our customers.

Got it. Thank you for that. And then just a housekeeping question. Did you take me a sense of in the IoT business, how much of that was industrial, how much of it was consumer?

Maybe if we divide it by CapEx-driven businesses versus consumer-driven. Gus, this is Louis speaking, by the way. It's roughly 50-50. It didn't change much from the prior couple of quarters. If we break it down a little bit and IoT for the year was around 80% of revenue and security, which is mostly enterprise security for us, obviously, that's enterprise CapEx. There's a little bit of home there. But then in portable video, things like [ wearables ] or enterprise video conferencing. And I think we had three announcements in that category this quarter. That's enterprise CapEx, but then you have 360-degree cameras. Things like aerial drones, which did go to production for us in Q4. Those are all consumer or [ prosumer ] type related. So that's how you get to the roughly 50-50.

And I'm showing no further questions at this time. And I would like to hand the conference back over to Dr. Fermi Wang for closing remarks.

Yes. Thank you for joining our call today, and I hope to see you at some of our numerous events this quarter. Thank you. [ Talk ] to you next time.

This concludes today's conference call. Thank you for participating, and you may now disconnect. Everyone, have a great day.

Well, good afternoon, and welcome from all of us at Ambarella to our CES Briefing and Technology discussion. I hope your first day of CES was productive. I'm sure for many of you online, in the room, one of your New Year's resolutions was to learn about at Edge AI and how Ambarella fits into it. So we're definitely going to help you check that box, so you can move on to your other new years solutions.

But some of you visited us earlier in the day. We have in the adjacent rooms, more than 30 demonstrations of Edge AI at work. And if you're not registered to visit us later in the week, you're welcome to contact me afterwards and we'll fit you into one of the groups. Also for the online participants, we'll need some time to prepare a virtual CES experience, but we should be running that in March, and you can also send me an e-mail if you'd like to participate in one of those events.

So Ambarella has been, I thought -- I was going to say 10-plus years, and I was just talking to Fermi. Ambarella has been at CES at this location in a much smaller footprint initially for almost 20 years now. So this is a very important event for us to share with customers and partners, our new products and technology and to talk about our Edge AI platform is evolving. And that's what we're going to achieve here. But we felt that in addition to all the demonstrations that we can provide to you, given all the developments in the AI market that providing some additional perspective would be very helpful in terms of what's happening in the market, our view of it and our strategy to tackle this space. So that's the purpose of the meeting today.

Before we proceed, I do need to read some forward-looking statements. I'll make it very short. I apologize. I wanted to get through it quickly, but I do need to read this part. Today's discussion will contain forward-looking statements that are based on currently available information and subject to risks, uncertainties and assumptions. Should any of these risks or uncertainties materialize or should our assumptions prove to be incorrect, our actual results could differ materially from the forward-looking statements. These risks, uncertainties and assumptions as well as other information on potential risk factors that could affect our business are more fully described in the documents we file with the SEC.

So we got through that. In terms of the agenda, we're going to run with management presentations for about 75 minutes. We're going to try to save 15 minutes for Q&A. One thing that's unique about this presentation is that you'll be hearing for the first time from 4 Ambarella executives that have not been accessible to the investment community before. So you'll hear them present today. But also after this event, will be hosting a reception immediately outside this room with some refreshments and I think some snacks. And you'll have a chance to meet with some of these executives as well as some of the partners in the room that we'll introduce later in the presentation.

As a reminder, this is not a financial event. So Ambarella's fiscal year, fiscal '26 ends in 3 weeks. And in our earnings call at the end of February, we'll be providing financial updates at that time.

So with that, let me turn it over to Dr. Fermi Wang, Ambarella's Co-Founder, President and CEO, to go through the rest of the deck.

Thank you, Louis, and good afternoon, and thank you for coming to this important event for Ambarella. As you know that CES is always important event for Ambarella to show our new technology products. And this year, with the fast evolution of AI technology market, we feel critically important for us to provide a deeper insight into our Edge AI strategy platform road map as well as demonstrate the capabilities of our new products.

Compared to the AI data centers, AI at Edge serve different purposes and calling for a fundamentally different silicon architecture. And on top of that, the AGM market and products needs unique requirements, like, one, power consumption is most critical; two, low latency and the privacy or must haves; three, much less bandwidth is available, not including the communication bandwidth as well as DRAM bandwidth; four, this market is supported not only by the business CapEx, but also the consumer spendings.

With all of the silicon hardware and sulfate optimization needed to achieve those market requirements as well as drive meaningful volumes at Edge, we believe that this Edge AI market still at the very early stage of commercial development. And this is the opportunity for Ambarella. And this is where we are focusing on today.

And at Ambarella, we are very proud that we offer a very comprehensive Edge AI platform, featuring both AI silicon and the software designed to support our customers not only for the design for the customer not only for their -- to meet the market requirements, but as well as help them to scale their business. And we already shipped that we will ship more than 50 -- sorry, 40 million units SoC across different application is the best proof for that to that statement. And today, we are happy to show you that we're going to expand this platform in a way that we have never done before. and also helping us to tap into new opportunities.

As you know, Ambarella is recognized as the best high-quality video perception system at age, original design for human viewing. And later on, this proprietary video perception system, which get a significant amount of data from various sources fit into our customer cost and AI accelerators. And this setup helps machines to operate autonomously, sometime partially some time fully on 1 single AI SoC. And this evolution between the video perception system as well as AI technology really help umbrella define multiple generation product cycles already.

In 2004, and we found the umbrella with one simple idea that the digital video content will become popular. And consequently, that we develop our first-generation video perception system. And in 2012, with all the AI paper, research paper published such as AlexNet, we came to the conclusion that the CN type new network will become the foundation of AGI, and we develop our CV2 family chip to seize that opportunity. Today, 80% of total revenue came from CV2 family chips.

In 2017, the transformer-type new network is published, which led to the development and creation, our third-generation architecture, which support both and transformer type of network. And with that, we create our CV3 family of chips, of course, based on our third-generation architecture, which really help us to address new applications both new physical AI application like aerial drones, auto driving and other robotic locations.

So that was the past history, how we get here. Today, we're going to show you how we're going to improve and improve the product offering in 5 different ways. First, we're going to really provide the update how we're going to improve our Edge AI strategy and platform. Two, we're going to announce our first 4-nanometer AI SoC being sampling to our customers. Three, we confirm that we on time tape out our first 2 nano meter chip to Samsung foundries. Fourth, we're going to discuss a brand new go-to-market strategy that will improve and increase our revenue generation. And fifth, describe the significant evolution our Cooper software development platform that will help us address the market need at the edge infrastructure market as well as the endpoint market.

I think many of you have seen this slide at the bottom of the slide, the last row of the chip is our video processor, which is average ASP of single-digit dollars. On the second row, which is our second-generation AI silicon and UC CV2 family like I talked about, the average ASP is ranging from $10 to $75. And third generation is on the third line and the average ASP is from $200 to $400. And today, we'll add 3 new members into this family.

First one, as I said, we are sampling our first 4-nanometer chip we call CV7. 48 hours that we received the sample from Samsung Foundry, we successfully brought up of 2 demos in our lab. One is 4K P60 video and AI processing for video streams. Second demo is 4 channels of video input at 4K P30, both for video processing AI. Both of this configuration has a performance level that nobody else can reach in the market. And those 2 configuration is going to be critically important for many different applications for our customers.

And we can achieve this kind of engineering the deals within 48 hours of receiving the chip, show 2 things. One, Ambarella continued our tradition of excellent engineering execution. But also throughout years, our silicon architecture and silicon road map plus our software platform becomes so mature that we can easy to tape out a chip and even though it's a complicated chip, that we can deliver those kind of demo in real time. We expect CV7 will be in production at the end of the year into certain different applications.

Second family we add to this page is our first 2-nanometer chip. We talked about this before. This is a chip going to be -- a tape out to Samsung foundries. And this is a -- the first customer is really helping us to pay for this chip is a semi-custom design for letter application that we haven't talked about, but we're definitely going to provide you more updates later on. And this chip will be used by our first customer into production in first half 2027.

And the third thing I want to talk about today is in the future -- in the past, Ambarella focus on edge endpoints. But just a few quarters ago, we announced our 1 family chip to address each infrastructure. We talked about that we had our first design win 2 quarters ago and the first design win were turning to mass production in first half of this year. And on top of that, Muneyb going to show you, talk about how we're going to evolve our go-to-market strategy for this age infrastructure market by including our GSIs and ISV partners to help us to expand the software offering to customers quickly.

And this is a new strategy, go-to-market strategy going to implant for not only for the aged infrastructure, but also for age end points. And I think this is important for us because in the history of Ambarella, most of our revenue is still driven by direct sales. And this channel sales, I think, will be beneficial for us by increased our potential revenue creations I think throughout the development and customer engagement of CV2 and CV3 family cycles. One thing become very clear that all of our customers, all of them telling us that they need more and more AI performance while they don't allow us to increase the power consumption numbers.

So first of all, let's talk about how we -- what kind of workload will trigger this kind of AI performance at edge. First of all, the CP workflow be comparable AI processing. And this fundamental change really pushes us to implement our silicon architecture in a way so that we can really have to increase our AI performance and horsepower in a big way. The second thing I want to point out is due to the need of low latency and privacy at edge we have to move some of the workloads, particularly for the application like physical AI, drones, driving car. You have to move those workloads from the data center to the edge. It's not just limited by the physical location. There are many, many more application networks start seeing our customers, enterprise customers, moving their AI workload from the cloud to the edge.

The third thing is probably one of the most important in the explosion of the transformer type of the workload, particularly for GenAI and the vision language model being widely used by our customers today. And they are trying to figure how to implement those new models into their products, and we are helping them to do that. And throughout the process, in the past, at the C level, we're talking about maybe a few hundred K parameters today, we're talking about $2 billion parameter as a minimal running on our 2 out chip. That just shows you how much more AI performance we need to add to our silicon to address that need.

And in addition to that, all our customer have said we need to attach more and more channels of sensors into our silicon, not only just camera sensors, but radar sense all the other type sensors. All those new channels of sensors requires new AI performance. And to make it worse is in the past, we are handling 2 megapixel cameras of the input. Today, we are addressing 8K P30 camera as a video input and the data among the data that comes into our chip increased significantly over the years. And because of the 5 reasons, you can see that our new products continue to add more AI performance, therefore, bigger die size, therefore, bigger, better ASP for us to address our customer needs.

One thing I can say is, for example, in Q4 last year, the average corporate ASP is $15. Today, every chip that we are developing or sampling right now have average -- higher average ASP than $15. That just show you how we're going to capture more value per AI unit.

In the last several quarters, we start introducing brand-new AI applications, take -- helping our customers to take them into mass productions. And we get asked many times by our investors what's the additional R&D cost to enable 1 extra new applications with our platforms. Today, I want to address this question by showing you the flexibility, programmability of our silicon and our software architecture. On the left-hand side of the picture shows this is a software stack and some of them done by us, some of them done by our customers. And MunibMuneyb going to go through the detail of that.

I just want to point out the core of the software stack is our corporate development platform. which is identical for every product that we ship in the market. And let me allow me to use the right-hand side, the CV5 as an example. In the last 2 years, we used CV5, our first 5-nanometer chip, take them into 7 different type of applications from automotive to enterprise security, drones, portable video cameras, perception system for robots and so on. see totally different applications. But if you open up those products, you will found that a bottom of that product is CV5 silicon plus a corporate development platform, identical -- the only difference on top -- for those 7 applications is the application level software that our FAE help our customer to develop.

So the extra cost to enable 1 new application is really that small FY '18 dedicated for each customer for their application. So you can see that the extra cost to -- for us to enable new application is limited -- and therefore, we can start showing significant ROI when we increase the number of new locations. And that's why we never fear to talk about new locations because the ROI from our investment. But I also want to point out -- this benefit of programmability and flexibility applies not only to Ambarella, but to our customers. For example, any customer who use CV5 to implement one application. If they want to go to lower end products, they can port their application level offer easily to achieve like CV75r for the low-end product porting to CV7, our new 4-nanometer product because they are sharing the sen silicon architecture and the Cooper development platform.

So our customer by using our system they can significantly reduce the R&D dollars and improve the ROI also -- and that's why -- and the reason I mentioned the previous slides is because we start seeing many, many new AI use cases. 5 years ago, when we get together, the most significant application for Edge AI for us was enterprise security. But today, all of the boxes I show here, we have either engagement or design wins in those spaces.

Let me use the last year as an example. Fiscal year '26, we not -- while our enterprise security market continued to grow in a very healthy way. The growth, our revenue growth last year also country by 2 new markets. One is a portable video market, the otherwise telematic market that we have not -- we never addressed in the past for the HI applications. And with this new application, it really helped us to significantly increase our revenue last year.

And in addition to the 2 new applications, we think there are 3 large opportunity for us, in front of us that we can tap into in a short period of time. One is robot. In this particular market, we focus on aerial drones because this has a huge volume opportunity for us. We also focus on, for example, MRs or manufacturing automation, human noise. They are all important markets in the future, but I think all of us still at early stage of the revenue generation. And also, we think the edge infrastructure like we talk about is important. We have first design win. We're working on more design wins. We think that potential is huge moving forward. Of course, the third one is autonomous driving that we have been working for a few years, and we continue to commit to this market and working with OEMs and Tier 1s for the -- to secure our first major OEM designs.

And beyond those 3 big opportunities, there are many green shoes here. All of them has shown some potentials, but it will take time for them to develop their revenue generation. But I will point out that all of these new opportunities that we put out here -- all of them are using Cooper development platform and leverage an silicon architecture so that to enable each one of them won't take a lot of new investment on the R&D side.

With all the discussion so far, I think it should not be a surprise to you that fiscal year '26 was our record revenue year in the Ambarella history. Here, we show you our revenue for the last few years and the blue bar is our -- the revenue generated from our traditional video-only product line. And the green bar is the revenue generated by our -- if you look at the overall -- our revenue performance, we generated 12% CAGR in those period of time. But if you discount, if you remove those company revenue generated by the company who impacted by the entity list.

For example, in fiscal year '20, HiVision Dahua, DJI was 45% of our total revenue. And they are almost 0 today because the entity list. If you remove that, our company revenue CAGR was 18%. And -- but that's -- there are nothing compared to our Edge AI revenue CAGR at the same period of time, which is [ 60% ]. So I think the only conclusion you can draw from here is the transition from a human viewing company for Ambarella into a machine perception, plus autonomous decision-making company is complete.

80% of our revenue come from our market -- our Edge AI solutions. That also shows the most important IP in this company today is our AI accelerators, which is really well defined and has better performance per watt than all of our competitors out there. And we believe that this is going to be the focus for us continue to leverage our traditional on the perception side but focus more and more on the AI accelerator to make sure that we can continue to deliver the performance per watt for our customer in the future.

At the end, I want to wrap up my presentation to make one statement, which is Ambarella is a leader of Edge AI market. I don't think that's an empty statement because there are support -- the statement is supported by the following numbers. We shipped more than 40 million Edge AI SoCs so far, cumulatively. And we're putting $1.3 billion of investment, R&D investment in the last 10 years into this product line. We helped 370 unique AI customer projects in production and also report over 200 unique AI model architecture, not AI model architectures for our customers. So all of this I challenge anybody outside who trying to come to AI to provide a similar matrix and provide the comparison. And with those numbers, I'm proud to say that Ambarella is the leader of AI.

So with that, I would like to introduce Muneyb to give you a more deep presentation into our strategy on the marketing and customer engagement. Munib joined us as the executive team of the customer growth office 6 months ago. The first time we met, we came to -- quickly came to a conclusion that we both believe that the HCM market is at early stage and worth our time and effort to make it bigger. And because Muneyb was -- spent his first 20 years building data centers and cloud products at a company like VMware. And he joined us from Intel as a Chief Marketing Officer for their network and the Edge IoT business.

With that, please welcome Munib.

Thank you, Fermi Thanks, Fermi, and thanks, folks, for taking your time and coming out here. I appreciate your spending here. And folks online. Hopefully, you can watch and follow us. So I have a longer agenda here, but let me kind of get in and thanks for -- as Fermi said, when we first met, we talked about the opportunity at edge about to explode. And I've spent 25, 30 years of my life just working in the data center and cloud and arrive at that conclusion that edge market is going to stand up and proceed just like data center and cloud was 3, 4 decades ago, right?

Because I think -- in the data center market, initially '90s and 2000s, it was about transforming companies. There was data automation and people who are building clients or architectures. And I see a lot of investors in the room top companies in the market, in the top where system providers went from silicon CPU as the market cap to system providers like the Ciscos and the world then came the cloud native market started moving towards the architecture was very much more mobile apps and developers building this.

Of course, market caps of SaaS companies and cloud companies went to the roof. Where we are looking at, we believe, will be where Edge will stand up, and you're actually now transforming not people now you're transforming things because everything is going to talk to each other to the AI market. And we believe that is going to explode and at the same level. It's not one against the other. There will always be a continuum, right? So there will be a continuum where there's still 30% of the workloads in the data center, 40% in the cloud, but new workloads emerging at the edge will be 30%. That mix is about to happen over the next decade, and that's the opportunity we're talking about, right? And how do you go capitalize on that, and that's a huge opportunity.

But let's kind of double-click into that because the segmentation is not clear. As you have data center folks moving towards the edge, they're building edge servers, which are in the scale of servers and platforms and racks. Whereas Ambarella, as you can imagine, IoT endpoints, we're seeing adjacent markets come in 2 ways. One, physical AI, where you're taking multiple sensors, vision, audio, infrared, radar and putting together cars or robots, which multi-sensors put together on one end. On the other end, up the network, we're seeing more and more kind of sensors and inputs to be processed, but not at a server level, more at an appliance level.

Now add AI to the mix that starts getting interesting because AI processing and model sizes vary. If you're in the cloud and data center, now you're looking at hundreds of tillions of parameters and models. As you move those towards the edge and you have big edge AI inferencing solutions and chip providers who look like wafer size chips, we're not talking about that, right? So a big, huge still a AI inferencing market, but then you're looking at more simple, smaller. So these are more parameter sizes of $500 million to $50 million or in the multimodal language, it's about 50 billion parameter model that you're going to do with.

So that kind of sizing is where we think that edge AI market comes up. And that -- but designing for edge, going on what Fermi said,it's a very unique architecture. You can't just take data center silicon software and move it very quickly to the edge. You have to spend a lot of different cycles because privacy and security very important. You can't stick a huge load balancer in front of an edge endpoint right? So that's not that easy, real-time processing.

You can't deal with 30, 50 milliseconds of delay in the cloud and data center, you have to be really real time because you do it real with real-time things. Reduced network is not like a land with a backplane of 10 gig you have to deal with really hard network at the edge and low power consumption. I don't have massive power facilities like data centers, it has to be embedded low-power and I think these are the synergies kind of help Ambarella position ourselves because we're designed for all these elements and capitalize on this, right?

So now what does that do for opportunity? Now I'm making a hypothesis that workloads are going to move to the edge, and they're going to run on these different devices rather than in the big data centers. Therefore, that infrastructure build-out is going to look from first our TAM, and we always talk about our current SAM. But our TAM in '27 is about $12 billion. But as you look at this new edge infrastructure market that's expanding, hardware, software services, that's a big pie. And our silicon can scale and capture some of those early wins in the network and gateways, which are already seeing the design wins we've shared about, but that's a potentially large TAM that we are able to tap into as we go after this market space, right?

So that's kind of a big opportunity for us to kind of scale and products that you go into the infrastructure market. But you do need a full stack solution. You need silicon system, software applications all coming together to kind of deliver on this. And the good news, as Fermi has been talking about is we have been as Ambarella for the last several years building that stack. We've not come forward and presented that stack as a whole solution. We've been going after vision and different aspects of that. But we have that stack and then just what you'll see in a lot of our demonstrations is as Fermi articulated, we have a full kind of range of silicon first, second, third generation but also designing the systems from endpoints.

Now we're starting to see the infrastructure being built in. We do so a lot of time building software. This software supports multiple operating systems from real-time operating systems, robotic operating systems, embedded Linux, QNX, all of these, but a common set of SDK in tool chain. That's super important because now you can write an application for 1 chip and easily move to wherever you want. And that's super critical for applications in the outbound interface in the software. And that ability for us to do that has been tremendous on back of which we stood up Cooper Developer platform, functional safety, ASL kind of very compliant platform and real-time platforms on top of that to support our customers.

Now we're also embracing open source because you see open source framework. And this is an important part because we believe open source will take off and agentic layers kind of helps the scaling and automation for those at the edge. So we have can announce today, and we'll get into the details of some of our developers one, but we're embracing an entire can open source framework and publishing, be on a set of algorithms we've already built.

So at this point, I probably want to invite our experts on stage from the company. So I want to have a panel discussion with Bob, who's our Chief Architect, come on to Bob and then [ Mohar ], our software principle. And Alberto Broggi, who is our General Manager of his lab to kind of start getting into unpacking that stack because they spend a lot of time working on these and you want to kind of understand how this works.

So thank you, gentlemen. Thanks for coming up. Let me kind of start with Bob and say, Bob, you would have talked about AI acceleration evolution. My first question always comes when I can always on discussing with you is, are you really an AI kind of architect? Or are you an environmental scientist?

Well, I think that's a funny question. Truthfully, I'd rather be a computer architect. But honestly, at the rate that power is scaling for these AI applications, I don't really have a choice. I mean, there's a fundamental scaling problem here. When the AI, there's a lot of exciting things that we can kind of imagine that AI can do and a lot of people are just throwing MAX at it. So a number of parameters and an AI may be doubling every 6 months. But if you look at the actual power and that requires more power to both train and deploy those type of systems.

But globally across the world, like our ability to generate enough power doesn't scale at those rates, maybe global power generation despite all our investments in clean and abundant energy is maybe only growing 1% or 2% a year. So fundamentally, those scales don't match. And so the consequence of that is something has to be done. The edge in that sense is inevitable.

I think that -- so let's look at from -- I'm a computer architect. So let's look at where power is actually spent. If you look one in of these devices, the amount of energy that it requires to compute is actually like fairly small. What actually is expensive is actually moving the data around. So maybe a Mac might cost like at most like 1 pico joule. But then once you need to move it either in some large kind of rectal limited chip or off to some DRAM memory DRAM or off-chip or some high-speed interconnect network, that can be 10 to 50 pico joules per second. So much larger kind of power just to move data.

And so then the natural consequence is, hey, let's get that computation, let's get that intelligence closer to where it's actually perceived, closer to the video of the image sensors, closer to the perception, closer to where we have some compute to generate some local intelligence. That's really critical. And I've been working at Ambarella for a long time. One of the things that's really exciting is we've been building edge products from the very beginning, all back to A1. So we have expertise in image processing and video compression and those are really exciting. And I think based with this energy challenge, one of the things that I'm really passionate about is architecture really matters and looking forward to future chips.

Yes, having been an architect myself, I'll tell you that's always constrained and balanced, like how do I architect something that is wonderful, but then I have all these little limitations. Well, it's reticle size is too big, it's too small. How do you kind of trade off between constraints and balance really?

Yes. I mean computer architecture, a lot of times are -- you talk about constraints. I mean power, we've just talked about is a really important constraint. -- latency is a constraint, compute is constrained. All of those are really interesting to kind of think about. But those are really inputs and constraints alone don't actually make a good chip, actually balance does.

One of the things -- when I see Ambarella chips, what I get really excited about from a computer architect point of view is when you look at all the different components that are inside our SoC, the image processing pipeline, the AI, CV flow as an example, DRAM controller and even ARM processing. When all of those are running at full capability and are fully utilized, then you know you have a balanced chip that's quite beautiful and very effective and also attractive from a business point of view because you don't have any kind of slop in that.

Inside, if you have an unbalanced chip, you can have a chip that looks very good on the slide. It hits like a benchmark, it hit some peak number. But then when you actually look at real kind of workloads, most of that time, it's not used. You don't have enough of any particular resource to actually use it. And especially when you get into the image processing pipeline or any kind of real-time application.

When you overload a constraint, the system fails. If we have to drop a frame, you've lost the shot. So that's critical. Hitting balance is really important. So actually, I think one of the -- an example of a well-balanced chip is CV7. We've looked at that, and we're looking forward to seeing it deployed. So Greg?

Thanks, Bob. And I think the natural question then I'll get this all the time is, hey, you're pulling accelerators or SOCs. It's like there's a lot of folks building accelerators very dedicated for the market, high speed, but the balance and constraint leads directly to -- should we be doing for the edge market, accelerators or SoCs. , what do you think we should be doing, right? Well, we're an SOC company. So obviously, the question is the answer should be SoC, right? But why I understand that, right?

So look, once balance is really important, then SOC is really the only answer because when you -- balance is really required, if you don't actually achieve that, then we leave here's the problem. If we don't actually build an SoC, then we leave the balance problem to someone else. And really, that's not where Ambarella provides value, right? We want to be able to balance all of these heterogeneous components. Those components that I've talked about, the image processing pipeline, CV flow, DRAM controller, all of those have different requirements and all of those have to be in balance. So if we just do an AI accelerator, then we're just leaving problems for somebody else to solve.

And I think the interesting thing is when you look at these different markets that you identified things like physical AI, AI bought edge boxes, smart camera applications. All of those are going to have slightly different type of balance points. And I think when you look at the number of chips that we've actually designed, maybe 15 or more based on our CV flow generation. All of those represent a very important and different sort of balance point that we've been able to achieve.

And we do that from an algorithm-first approach. Basically, we study where does that balance point need to be? What are the critical applications that we need to solve how do we do those effectively and at low power and then build a chip around that. So in an architecture sort of but it's really -- when you see our Ambarella chips that are working properly in the full SoC, it's also very beautiful.

Last question for you. I'll have 1 more after this. But really, are we are compromising between flexibility and scale in here? Or is it a choice because of all the flexibility, am I losing scale or are we able to do the boat, right?

Yes. So I mean, flexibility comes at a cost, and I think that's important to keep in mind. -- a lot of times in these large kind of cloud systems, maybe the flexibility that you get is fairly small. It doesn't even show up on your kind of balance sheet. But then once you build these sort of edge -- smaller edge devices then building general flexibility is not for free is not free. And everything has to be in balance and be cheap in order to make sense from a business perspective.

So we -- how do you build flexibility in your system so you can cover a larger range of balance on -- and I think Ambarella, we have an approach of building flexibility into the architecture family, meaning we build CVflow. And then once we have CVflow, we can scale that up or scale that down to hit some specific balance point that we actually need. Now there is some local kind of flexibility. Obviously, we want to -- we're looking at algorithms first, as I mentioned, -- we're looking ahead and seeing the road map of where things are going to go, and we build enough kind of local flexibility, but it comes at a much cheaper cost than it would be to build general flexibility into our chips.

An example of this is when transformers and LLMs and ChatGPT networks really kicked off, we're like, "All right, we need to demonstrate this. So we look around at the different chips that we have, we picked N1 that was at a very good balance point. And then with the power of our software stack and the tools that we have available, we were able to go and execute that demonstrate that you could run N1 at LLM inference on N1 very effectively and do that in a short period of time. So that was -- that's sort of an exciting kind of prototyping aspect and then we can use it effectively.

Great. Thank you. So kind of to summarize what you're saying, is Bob is our huge differentiation is not these trade-offs that a lot of other vendors have to do. Our huge differentiation is we can actually do an amazing AR detecture with power consumption, have the constraints, but balance it right and SoCs and accelerators, we can get that blend ride.

And finally, the flexibility and scale giving us to almost 400 million chips that we have shipped with that type of balance. And Fermi's slide especially takes B5, go into 7 different application markets, very easily and programmable. I think this is a big differentiation for what Ambarella has. A lot of people trying to understand where that fits, but this is our key differentiation from an architecture perspective. .

Thanks, Bob. I kind of move over to you, Mohar. Now I think we also talked about the full stack. Now going from the silicon architecture, we talked a lot about Cooper development platform. But before we get in there, can you tell us like what are the trends we're seeing in system software identical, right? .

Right. So as Fermi mentioned, transformer explosion happen everywhere. So we can see transformers even in our traditional like a task like object detection and everywhere, that's already there. We have a CNN Gen tool, which is quite proven. It has been there for 10 years. And we had this thought that put metric multiplication into our third generation, which is what Bob and our architects and team did. So we were able to quickly bring up the transformer model in our CNN tool set. And that basically enabled us a lot of new applications like all this multi-model application that even the traditional ADAS or all the software being and with the transform or the new AI kind of -- the models came up. The AI is kind of slightly different because they are auto aggressive models. So we came up with the new tool LMG, that's part of the Cooper platform also.

So we were able to quickly bring up the VLM Gen, and we can bring up all kinds of models with our PLM software. The N1 example that he gave is basically that's how it is. The other thing that I wanted to actually mention is Bob also mentioned about scale up and scale down of our architecture. With our tool chain, we can actually do it automatically. So it's not -- even though we see a lot of chips, the CNN happens to be one tool that can scale up and down easily, okay? So that's about the current thing that we have the foresight of having this metric in multiplication, which enable all kinds of transformer applications, right?

Now if you see moving forward, the inference time scaling is becoming very important now. Now inference time clinging all these reasoning models are coming up, all this chain of thoughts are coming up or all this things that even the agent in workflow, if you see the context size keeps increasing and the amount of data that is being generated keeps increasing. So inference time scaling is a challenge that we are working on. that's what Bob and the architecture team is actually looking forward in our new architecture. So that's what it is.

People outside also know and the bottleneck that you guys mentioned earlier about the DRAM bandwidth and all that. There are new techniques to people at have core and a lot of other companies about like 4-bed conization to controlling, all these things, we also keep a watt. And that's basically some of it we can apply today with our VLM. Some of it, we are actually going to -- looking at it to do it more efficiently in the future director.

So if you look at -- you also mentioned earlier that we are embracing the open source community. So AI as such is in still nascent format. And new things are coming up every 6 months, you will see new framework. So agent workflow is very popular right now. So going from one LLM, another LLM, defining the memory, define the contact. That whole thing like LLM is an application by itself, right, having all these things to work together. So we are keeping a watch on that one. And if you look at our VLM Gen tools and all that, we are actually -- we are embracing hugging phase, Hugging phase, and actually most of you probably know, hugging phases where almost all models are released nowadays, and we can quickly convert those models from hugging phase to run on our chip.

And with our tools being common, we can go from C72 to even C7 demo that you see outside, we already can actually execute some of those things with our tools. So it's a scalable architecture, right? So with the new optimization techniques we can incorporate, thanks to being embracing with the agent to hugging phase.

And the last, but not the least, so physically I that for me and you were referring earlier, we cannot ignore the security, the cybersecurity and the safety, right? So there are techniques in the -- that are coming up in the LLM world even to address that. And we also -- I mean, with our automotive, we also have experience with ACL and other stuff that we have also worked on -- that experience helps us in the cybersecurity and all that. So that also is -- we can leverage that also today.

So that's kind of our Cooper dev platform like I know you kind of threw a lot of the tool change, I wanted to kind of in there so you can talk to it, right? Yes.

So let's look at the holistic one, the Cooper development platform, right? If you look at the foundation is our metal is our chip is what Bob and the team has developed. And then we have these core libraries. The core libraries are actually based for the vision, like what we are doing, the streamer libraries and all that we released. And the critical part of it is the tools that I mentioned, CNN Gen tool and the VLM Gen 2. We are also -- today, we announced model garden. So some of these models that are converted and optimized with our tools, we are already publishing on our website on the aging phase as well as on our website, right, so the customers can directly use that optimized model, so they don't have to go to the exercise of optimization.

On top of it, we also have a run time environment to run all these models, right? Now one thing I wanted to stress, I think we also mentioned earlier, we have one as Cooper platform that runs across different chips. So once we invest for 1 chip or writing an application, it can easily transform from CV22, let's say, CV75 or let's say, CV7, depending on the balance point that you choose for your application, we quickly can develop have an application. Then we have all other software stacks that are necessary for robotics like Ross and all those things. that also we have developed and those are part of the Cooper development platform.

So Cooper development platform is a holistic, which includes all the libraries that are the drivers, which are interacting with the silicon, which has some open-source model on the model garden, on the hugging phase on the cloud. We have the tools, which are going to convert it and then the application example that we have for various markets. So we have one platform that can address different needs as such.

Thanks, Mohar. I think you go to Alberto. Now the opportunity to kind of double-click into -- well, we talked about a horizontal silicon and horizontal stack with Alberto who's been our General Manager in Italy for Vislab, an acquisition, like how are we seeing -- let's go deep into one vertical, like autonomous driving software stack, but go ahead, Alberto, yes?

Yes. Actually, the trends -- the more the trends in autonomous driving are mainly going towards real end-to-end. And that doesn't -- it means that the full pipeline, so the data acquisition, processing, the perception, the fusion, everything is based on AI. And that does not necessarily imply that we're talking about as one big network, which receives pixels and radar echos and delivers just set points of speed and steering because training that big network would require very large data sets -- and when you talk about large data sets, you mean a lot of energy for the training, a lot of time for the training.

Large data sets also implies in these data sets. So moving towards end-to-end means having big data, big data sets and our solution is in the network that I was talking about in multiple networks that are just one. So instead of having just one network, you have multiple networks, you can train them more easily because you can use data sets and more focused data sets just for that specific purpose. And the network should be overlapped and sharing some latent space. And by doing that, you're ensuring that you are preserving and you're propagating the features which are actually intrinsic feature, non-explicit features, and you can propagate them throughout the whole stack.

So that's the real interesting part for the end-to-end system. So that's what we do. And for example, in this slide here, you can see our current version of our stack, which is based again on multiple networks all connected together that deliver the actual autonomous driving. But actually, the stack itself alone is not enough. So you need more than that. So you need a complete ecosystem around it. So again, the stack itself and the architecture is so important, it's so critical, but you need something more. And because training a network or multiple networks, again, it's required a lot of data sets, diversified data sets.

You need to have large different scenarios covered. And you need to have also grown truth connected to those data sets. So again, the more diversified data sets the more scenes are connected are covered. So that means that you're in large your ODD design domain. And if you enlarge the ODD, that means that you are increasing the maturity of your stack. So if you really want to scale up your in to have lots of data sets, ground truth for them and selecting the data sets.

And this is what we're doing actually right now if you go to the next step. So this is what we're having. So we have a complete tooling. And the magic here is that all the tooling needs to be automatic. So if you really want to scale up, you do not have to have any human intervention in the pipeline. So starting from the acquisition of the data, we have vehicles that drive around acquire data sets and the driving behavior of the person driving the car. And then you pull this data into our automatic annotation pipeline that attaches the precise ground tooth to these data sets.

Then once you have the ground tools, you need to select the data sets. You need to create a right balanced mix of data sets in order to be able to cover the whole ODD. And this is done by another tool that we have in the pipeline, which is this VLM base. This is based on LLM and is able to classify the data sets. So data size whether night day or pedestrians, tracks and so on. And then based on that and the statistics, you can create the right data sets for the training.

So once the training is done, you put these networks, multiple networks on your car or on your simulator and then you get statistics about the performance of the system and these statistics are also used for prioritizing the next acquisition campaign because you understand how much data you have of some specific scenarios if you are lacking some scenarios. So these information are also used in the full stack. So this is actually the full tooling, which we have been using in our project in L4 autonomous trucking project we have with [ OMV ], which is formerly Continental that is going to be SOP next year.

No, I think that's great news. So I think it's good to see these things going into production pretty fast.

Now taking a tangent with you, a lot of the learnings, so there's a lot of nice about robotics and movement in AMRs at CES this week. How can we take some of these learnings and apply it to robotics?

Yes. Actually, we started with robotics at the very beginning. So we started with vehicle automation for big vehicles like heart moving vehicles and agricultural vehicles. So we started with that. And actually, there are strong commonalities between many fields in the robotics and the autonomous driving that we've been working on. And the basic architecture and principle is the same, is very similar.

And the core functions that you have in a robotic system, well, they are core functionally mean the perception, data fusion, planning. These are more or less the same fundamental blocks that you have in a robotic system. Plus, I would also say that maybe some differences could be in the kind of sensor data that you're using. So maybe in robotics systems, you have a strong usage of cameras and maybe other sensors like textile or pressure sensors that you don't have in a module course, maybe some difference in the precision of the planning that you have to do. But despite these differences, I think that, again, the underlying structure is exactly the same. It's largely similar.

And one can also benefit from the well architecture that we have been developing for the autonomous driving, which is kind of similar. I didn't mention before, but actually, our architecture is based on 2 stacks. So the first one is what we call the fast stack. So that means the one that received the input data and provide the steering position and gas position. And then we have on top of that, an LLM based or ELM, stack, the slower stack. That actually synthesizes higher-level information about the seeing the context and drives the lower level. So even in this way of moving towards higher level of automation. That's -- I mean, can benefit from that.

That's like human brain. Like there's a popular book called like in slow think fast, which is about how the brain works, there's a natural thing you do fast and something that you learn and get fast. So that's great. Yes, that's pretty close to what we're thinking.

So well, I think that was kind of our kind of panel section and coming back to the stack. Before I can let you folks go short 30 seconds, what does the future look like, Bob, for architecture?

I mean that's kind of a dangerous question because maybe I know a little bit too much the that I can really say. What I would say is I think that we've had a lot of experience now looking at deploying all of these transformer-based networks and on our architecture. And we have a VLSI team that works very, very fast. And so the important thing is to make sure that they have interesting things to do. So we're looking at what are those opportunities. Now that we've understand and see how the systems operate on our hardware, what does the next generation look like? And I think that's what we're focused on and in sort of even more efficient and more power -- low-power devices in the future.

You're an environment.

Yes.

That's right. Alberto, what about you? Where is the trench going?

I'm going echo what Bob just mentioning, actually having the possibility of having these 2 layers architecture going together, mixing the 2 things together instead of having just 1 slow stack in 1 at fast stack, having them together, thanks to the higher power that we can get. That's the future.

So t 2 brains, 1 brand. Finally, Mohar, why don't you -- where does this agentic and where is the software going to go, right? .

Software as such is how quickly you can enable customers to bring up this SLM or SMALL Language model that you are referring to, the VLM and LLMs, quickly through agenticworkflow, we can bring it to the edge, the small model. That's what it is -- the key part is going to be moving forward.

Thanks, folks. And I'll let you back in as you can head back. Thanks for your time. Yes. .

As the folks head back, I'm going to give you a quick update on our kind of business and portfolio updates. And I think so you get a quick view. I know we're going to run on time. So I'll give you a fast kind of part big parts of our business. IoT, which is kind of one of our largest kind of business portfolio in 25 is almost 70%. As Fermi pointed out, there's more and more AI that's happening.

We got a nose presence in enterprise security. From enterprise security, we're seeing it move into public home safety kind of markets and smart home markets and a lot of kind of announcements along those but also going into portable video. You see drones that are flying outside. You're seeing robotics, as we've been mentioning that there's adoption of these edge IoT kind of environments that are happening on one end, which is end points. But on the other hand, we're starting to see emergence of this edge infrastructure applications that are also saying, "Oh, I need a small box of compute where I can aggregate all of this and compute at the edge without a massive server environment."

So we're seeing huge uptick of this business that is going with the range of articulated from first gen, second gen and third gen and the attributes just that Bob and Mila kind of mentioned it from a unified SDK, a family with a common set of software One thing I know we take for granted, but we have some of the best image quality, like leading AI ISP in the market. and I'll talk about CV7. But that's huge, but we take it for granted, but a lot of people don't realize how good and high quality and how advanced we are.

Power efficiency, I think Bob's job continues to be an environmentalist, but we have a very power-constrained. And the GenAI models, like I know like last year, DeepSeek came out within 2 months, we had it running, right, because of the flexibility and balance that we're looking at and doing this. So a whole range of chips C7 is something we announced yesterday.

So CV7, as an example, goes from CV5 with 2.5x more AI and the CVflow 3, 2x more encode, 2x for CPU, of course, is 4 nanometer. So lower power kind envelope as well and supporting all the applications, Fermi that slide with applications on CV5, we're already seeing the designs of CV 7 in a lot of these kind of environments. And what's kind of really interesting is we announced yesterday with our CV7 and the performance, we had competitors announced a similar chip and new -- like new chip from them, but it's still -- we're 2x performance at launch with those folks.

So I think that's what some of the -- a powerful thing for us is being able to keep up and keep launching at the same time with competition, but we're already 2x ahead of them and what we can do with 8K Vision and 60p that other folks can't do it yet, right? So that's differentiator. But me talking about it, we wanted to also have one of our customer testimonials couldn't be here. But from IQSite, we have Sabrina the CEO talking about it.

[Presentation]

Hello. My name is Sabrina Stanburn, CEO of IQSite. IQSite was recently launched as the driving force behind Bosch branded intelligent video solutions. As a spin-off from Bosch, we have a global footprint and a solid foundation of proven reliability, quality and a relentless drive for innovation. By using AI, we help build a future where safety and security incidents no longer disrupt our lives. Our cutting-edge video solutions combine world-class hardware with intuitive software to protect people and assets. Bosch cameras are installed across the globe in a broad range of applications. from buildings and public spaces to critical infrastructure.

At a IQSite, we will continue to lead the way in AI-enabled video. We are on a mission to advance predictive security to help our customers know what's coming next, act faster and unlock insights beyond traditional security. This is an exciting period of transformation for physical security driven by innovations and AI technology. And we at IQSite, won't leave the way forward. Today, AI-based video systems can classify and tie capture image data to streamline monitoring.

As we embark on the next generation of AI, we are providing customers with technology that will help them overcome the inefficiencies of human monitoring. Our new portfolio with GenAI models allows us to automatically monitor and generate scenes, descriptions and the images recorded by our cameras. For example, our cameras understand the difference between a person on the phone walking close to a car and a person actually vandalizing a car based on the same natural language GenAI technology that has taken the world by storm. -- these visual language models help ensure that no potential threats goes unnoticed and at the same time, avoid false alarms.

Over the last 2 decades, our partnership with Ambarella has been the key ingredients of our innovation and physical security. Ambarella's impressive video processing heritage and their systems on chips incredibly efficient AI performance per watt enables us to run our AI software on edge and low-power devices like our cameras. Our joint innovation continues. We are expanding our edge AI portfolio with more advanced mutual capabilities. These will allow us to generate reliable alerts. And thanks to ASP for Ambarella, we will now enhance video quality in real time. at the edge, it's important to have everything integrated into one chip. The perception, AI and general processing blocks.

That's what Ambarella's single system on chip does. It makes employment of visual language model so much more efficient. And since we run Gen AI-based monitoring directly on our cameras, we reduce human error and cut down on cloud processing costs for our customers. But there's more to it. Everything happens locally, which means faster response times plus sensitive data stays on the edge, which improves privacy, security and reliability.

Looking ahead, as we continue to further improve the efficiency of our Edge AI technology, we see a huge potential to unlock new revenue streams from these advancements. Our collaboration has been a key driver to our shared success. Together, we've combined cutting-edge technology with deep market expertise to deliver a meaningful impact for customers deploying Bosch video systems. Ambarella's world-class support has consistently spout no matter how complex or challenging our requirements have been. We at IQSite, look forward to continuing our joint collaboration with Ambarella in the security and AI markets.

Thank you, Sabrina. I think it was a pleasure kind of working and work with some of our top customers. I just want to highlight, if you haven't had a chance to go look at some demos and robotics, but robotics is being adopted from different chipsets all the way from just vision in Level 2 to stereo vision Level 3 as well as the brain autonomy that we just discussed with Alberto over there. We have a whole bunch of kind of demonstrations. But one of the first ones you will see flying around is the antigravity drone that's if you haven't visit, you should go check it out with the vision goggles. It's -- we call it the first type of robotics as these robotic drones that are flying around on wise commands and following item stuff.

So let's quickly kind of switch over to auto side of the business as well. So that's -- auto is the second part of our large business, 30% of our business. But we're seeing, again, huge AI adoption auto safety, telematics business is growing significantly. We've had a lot of kind of design wins and announcements with key customers in that area. And we see AI utilization inferencing and a lot of adoption in fleet and aftermarket like drive recorders, e-mirrors, driver management systems, occupant monitoring systems as well as fleet and telematics.

At the same time, we had this good discussion with Alberto on where we're going with our AD family and software stack from L2+ to L4, starting to get some design wins and software can be built out. So we have a whole range of chips. And this is very important is people have a lot of chips on different kind of markets. But to Fermi's point, we also have this whole range of these legacy chips, all of them ASIL compliant, all of them ECQ100 compliance. So all the way from vision, viewing, sensing as well as the RADAR family for Oculii acquisition, which is our CV3 family, which is available, and you should take that demo out.

But beyond vision and perception of CV2, CV7X and 5 range of series, the CV3 AD, which also acts like the domain controller, like central domain controller for all of that. you need that brain, that aspect of that as well along with the viewing and the sensing aspects of our portfolio. And performance-wise, you can benchmark, and if you see some of the demos, we have amazing performance against competition that you can see in real time as you go check it out, right?

So again, enough said, we'll have one of our customers' testimonials from Kodiak.

I'm [ Jamie Haffacker ], VP of Hardware at Kodiak Robotics. I am responsible for our hardware platforms, including our Class 8 big rig trucks. I've been fortunate enough to work with Ambarella for almost 20 years now, including developing an Ambarella ASIC basin coder, as some of you may remember, that was used by NBC to broadcast the 2008 Summer Olympics.

So what do we do at Kodiak? At Kodiak, we're a leading provider of Level 4 AI-powered autonomous vehicle technology. And today, we're focused on tackling some of the biggest jobs in trucking. Kodiak has built a single integrated software platform designed for deployment across 3 main verticals: long-haul trucking, industrial trucking and defense. Take our industrial deployment with Atlas Energy in West Texas or we've been contracted to deliver 100 AI-driven trucks. These trucks are operating today completely autonomously.

We designed the Kodiak AI driver to operate in challenging driving environments, and this is definitely 1 of them. We operate every day of the year and truck uptime is everything. We can't stop delivering even in heavy dust or rainstorms. So this is where we turn to Ambarella to customize the solution with Kodiak. We need to do the best camera SoC available on the market that we needed it to be robust across a wide range of conditions. Today, each of the AI-driven trucks in West Texas is running with 4 Ambarella CV2 providing best-in-class performance, especially critical in low light and high dynamic range driving conditions.

Beyond our West Texas deployment, we are particularly excited about our current work with Ambarella on the C3 platform. Not only does the CB3 provide best-in-class camera performance, but we're also able to utilize the additional processing power for RADAR and LiDAR processing as well as running our time-critical neural networks at the edge. Using the CB3 for sensor processing also materially reduces complexity and the power needs of Kodiak's overall solution. Instead of running individual cabling to the central compute node, for example, we're able to do core processing next to the sensor, which also reduces central computing demands. Ambarella continues to be a great partner to Kodiak, delivering practical solutions and top support, allowing Kodiak to stay on the cutting edge.

Thanks, Jamie and the team at Kodiak. So I think I'm going to wrap up the auto section, as we discussed with Alberto, there is both that online software stack from deep sensing across images and imaging and radar but also the deep planning capability for us to do motion forecasting, maneuver and trajectory planning as well as pet control. as well as an offline data pipeline, right? So which is basically data capture autoanotation that Alberto touched on and our data selection, ODD, which is very important and how do we run future data acquisition campaigns.

So that's kind of the -- our auto part of the business. Now Fermi mentioned, we're also thinking about expanding our go-to-market. And to touch on the go-to-market aspect, which Here's our current go-to-market as it stands, right? We take and have our silicon, it goes to ODMs, Tier 1s, goes through OEMs and Tier 2 kind of partners and scale. I call it a typical push motion, right? So it's just basically, hey, we take a design win, we go into and gets involved. And what you're seeing as we kind of evolve our approach in our stack and going to a full stack solution by the time if you leave -- remember one thing, you're going to remember this full stack, right?

So this full stack solution then requires us to actually also create what I would call a pull motion. A pull motion happens when you start building beyond your design wins to start having software, which is built an application which are built some affinity to your silicon. So that is usually built with vertical use cases. with ISVs, software vendors, who kind of solve what's a particular problem and how do you kind of scale that and then get that deployed through system integrators in the global kind of landscape of doing that.

So with that in mind, that's the kind of expansion and growth we want to see through our channels and scaling that we expect to kind of go in our next phase of growth. And I think that's definitely why today we launched in our developer one. developer zone allows for folks to come in and build applications, applications, test out our models in the cloud before they go and actually go and do this evaluation.

So yes, we've talked about our edge AI software stack and the Cooper is developer platform, where we have our SoCs, our operating system, our kits and our SDK. But today, as part of that developer zone, we're also launching our capabilities of model garden. I think Malhar touched on this. It's dozens of models which are tuned for our second already. genetic blueprint. Agent gives us a level of automation before if I'm just going to go and enroll developers, have to publish my SDKs, APIs is a cumbersome process for people to learn how to code with our APIs, but fix it.

Agentics allows us to provide that level of automation very quickly. and be able to -- so we're creating agentic blueprints, which are published on our website with several applications. And at launch, we announced it 2 of our ISV partners and [Cogniac and Mel CX ] we have the CTO of Cogniac, Sandeep in there. We'll have a wave, and then we have the COO for Mel CX [ Thor ] in the back there. So they're launch partners. What's amazing is within a matter of weeks and this we're able to actually bring their applications in retail market, in hospitality market in transportation, like railway network applications to be ported in our platform very quickly.

Shows the programmability of the software aspect, how fast we can go and get this done. So we're kind of opening this early access program for developers to kind of come and start building this. applications have affinity to our silicon and we're enabling that because that will give a level of scaling that's kind of new to us. So when the developers -- I think Malhar was trying to articulate this. So we're able to kind of bring any type of data, rear all simulated data and take foundational models. We published a few eventually, we expect there to be any model.

And then bettering it to us in a platform to a software stack to put and train that but then test those models in the cloud or eventually in the kids where they can attach any type of modality and deploy that to different types of any edge. And then, of course, you want to fine-tune that model as you kind of do this. So our software stack is really kind of providing that capability for any model, any type of data, any modality to any edge, really, what we expect is like I think Malhar said this, AI is happening super fast. So we don't know what comes in the next 6 months or a year, what type of new models and technologies. But as Ambarella, we want to be ready for any. So any type of model, any type of modality. We are ready, and our flexibility and architecture is really available for that. And that's -- we'll prove that through our kind of ecosystem and how we go to market.

And I think the timing is also critical because we're seeing this explosion of as we saw that transition. Agentics provides us a level of sophistication and automation that hasn't existed in the past. So the timing is really important that we're doing this now because the onboarding of ISVs takes days and weeks, not months or years, right? So that's really the speed at which we can go. And our installed base our type of model architectures that are already solved for our ecosystem is ready to kind of go and expand on this at this right point in time because the time to market with us is going to be super fast.

And to close up my section is really our growth trajectory will have a chunk of business still growing, expanding our IoT and auto customer expansion. We'll see additional revenues go through this channel scale motion because now suddenly, you have distribution resellers, ISVs, SIs will start picking and taking a whole range of customers, which we haven't tapped into in the past and in citing that of new opportunities for our AI and infrastructure customers happening. So really excited with the growth that Ambarella has in forward.

With that, I want to probably pass it on to Chan to talk about -- Chan's our COO, has been here a long time and could talk about skilled operations and our VLSI sophistication. Over to you, Chan.

Hello, everybody. My son watches YouTube and the guy so impatient would play everything at 2x or 4x talk, it sounds like it's running like that. So maybe I need to speak like that, huh? Yes. So thank you for coming.

Okay. I will introduce you to my team, what we have been doing and maybe just a glimpse of the future. So from the beginning, we have taped out over production associates. And Bob just mentioned, he said AI. And it just hit my head. It was our 130-nanometer first silicon. Yes. Bob looks the same except he had maybe more hair back then. So the latest as we take out just recently with Samsung, it's 2-nanometer. It took a lot of work, but it was beautiful.

What's really remarkable about our silicon team is not just how many chips we take out, but it's about success rate. 97% of our take out actually went into production, which is a single spin or less. Out of those, 70% of them actually went into production with no spin, E0 production. And this was -- I used to work at Intel and there was this dream of Intel to go into production with E0. But I was reading some news about Intel. I guess they're taping out E0, stepping to go into production, which I was quite sad to see this kind of thing, but at least we are doing a much better job here, and it is truly exceptional in my mind.

Okay. microphone is not working. Sorry. So recently -- most recently, we have been working on these more advanced node products, 10 nano, 5 nano, 4 nano, 2 nano. We taped out 14 production SoCs as recent as a few days ago and 2 nano and 12 of those are actually in production already, demonstrating our technology leadership in every sense. So Ambarella has been in this business for quite some time, 22 years. We have gathered some domain expertise over time. We have very mature proven vision and multimodal AI IPs and deep system-level expertise.

Ambarella team is known for very efficient execution as Fermi was referring to. And at the same time, high-quality silicon. So as you can see from the number of spins we do on the silicon -- we also do rapid process node migration, moving down the node process node very quickly as a reliable semiconductor partner to our customers.

And lastly, we are very proud of this $400 million -- actually more than $400 million shipped to date. This showcases our product quality. We can scale and our scale of operations, we can ship these products to our customers' product and shipping everywhere, which leads me to my next topic.

So as a semiconductor company these days, supply chain and operation becomes very important. -- ever more critical now with the current environment, you are all well aware of in the backdrop. So to that end, we have maintained decades long strong cooperation with our partners. Samsung Foundry, I recognize some faces here from our partner.

So I'd like to tease and call out some name just for fun. There's Peter I see. And Sean, and -- oh, there's Kelvin, yes. Samsung Foundry partners are here to support us. We worked together for 17 years. Fermi and I had a bet actually, are we the only company that's been working with Samsung exclusively for so long. And I lost the bet actually. He was right. We are the only one. So -- what this long and strong relationship gives us is we are able to get early access to the most advanced node like 2 nano process GAA technology and able to get stable wafer supply through so many supply chain shocks. And we have to grant those over and over recently. Our supply chain partners are global, geographically diverse and resilient to supply chain shocks.

And Samsung Foundry, for example, have mega fabs in South Korea and Texas. And our OSAT partners like ASE, CGuard and others, we have locations across Asia, Taiwan, South Korea, Southeast Asia. So with our operations team, very flexible, and we can rapidly reallocate resources, relocate our resources to remediate any kind of surprises and shock and stress in the supply chain very quickly. So that's been our strength.

Next is Samsung Foundry's video clip speaking is Margaret Han, who is our -- who is a U.S. Head of Samsung Foundry.

Hello, I am Margaret Han, Executive Vice President and Head of U.S. Foundry and Samsung semiconductor. In the AI era, Samsung Foundry delivers customized solutions across a full semiconductor value chain, from advanced process technology and design IP to full turnkey services and advanced packaging. Our state-of-the-art manufacturing fab spent from Korea to the onshore facilities here in the United States. These capabilities enable next-generation AI system to process a massive amount of data faster, more efficiently and with great preceding.

One of our most valued long-term partnership is with Ambarella. We have worked together for more than 17 years. Starting the 45-nanometer node for their first video processor SoCs where we help them become leaders in HD broadcast video and sports cameras. Over the years, our strategic partnership has delivered more than 40 products. This includes Ambarella's second-generation AI SoCs built on our 10-nanometer process between 2019 to 2021. And today's third-generation AI SoCs implemented on our events 5-nanometer, 4-nanometer and 2-nanometer nodes.

Most recently, Ambarella announced its new Edge AI CV7 SoC family here at CES. These devices are built on Samsung Foundry's latest 4-nanometer process. Within days of receiving alpha samples, Ambarella is true casing the CV7 at CES, highlighting its readiness and real-world impact. The CV7 delivers a major leap in AI capability. In able intelligent processing of multiple live 8K video streams with improved energy efficiency.

We truly value our long-standing partnership with Ambarella. Through close collaboration, we have helped ship more than 39 million Edge AI SoCs and more than 385 million SOCs in total. The newly announced CV7 family built on this success Using our 4-nanometer process and advanced architecture, it delivers higher AI performance per watt. This enables support for Lantis VOM, VOA and agentic AI models across both aged infrastructure and physical AI applications.

Looking ahead, our joint innovation continues at full speed. We are excited to have Ambarella as 1 of our lead customers to enter production on Samsung Foundry's most advanced 2-nanometer giga around process technology. Samsung Foundry will continue to empower Ambarella to exceed customer expectations, together, advancing the future of edge AI and semiconductor innovation.

Okay. Thank you. We love the phone nano actually. It's a beautiful process node. So one last thing, one more thing. So Ambarella since founding, we have done in a very limited way and selectively some semi-custom joint product SoCs with key customers. very few. More recently, we are seeing very strong interest and our customers' voices are getting louder and louder, asking for custom products, customer SCs or semi-custom products to differentiate.

And it's sold out. It's ringing in my ears now. I've been hearing it all last year and recently also, we think there are 3 main thrusts, 3 reasons really driving this. First is just the cost, simple cost. So 2-nano, 4-nano SoC development, it's economics. It's so expensive. We see people estimating hundreds of millions of dollars to develop such SoCs. And in 2-nano, some people are estimating $1 billion investment. So mistake or surprise while you're doing -- while you're investing $1 billion, failure is just prohibitive.

And second reason that I think is happening is there's scarcity of design talent. So pool of engineers who can actually develop these advanced node complex SoCs are just not enough worldwide, and it's -- everybody wants to design such products, but it's just not easy to find people to do it. And the third reason, I call this Apple, Apple invested for decades to get to where they are in billions of dollars maybe tens of billions of dollars. But system OEMs need this differentiation in their products. And the heart of this differentiation starts from silicon. Silicon is differentiated silicon is what you really need. And this comes from Steve Jobs at Apple in 2000. I think he started investing heavily into this.

So those are the 3 reasons. Behind this -- in this environment, in Ambarella, we have unique system design expertise, years of years in this business. We have stable and mature SDK and vision and multimodal AI IP that's proven in the field.

So I think Fermi was referring to this 2 nano product, semi-customer SoC. We just taped out this product working jointly with our key customer and the spec and all aspects of design was jointly defined. And it will be used by our key customer, of course. And because there's just so many -- so much interest and demand we are seriously exploring expanding this custom and semi-custom SoC product opportunities. And this is something that we can look forward to in the near future. Thank you.

So Louis can come up, we are going to Q&A.

Yes. Thank you, Chan, for your presentation, and thank you to Samsung for 17 years of support. Look forward to the future together. We're going to -- we've used up our 90 minutes. We fit a lot into that. I hope you agree with us, but we'll jump right into Q&A now. And I think Casey -- Sure. I think, Casey, if you're here, if you arm yourself with a microphone. And if anyone has a question, please raise your hand in the room. Sure, go ahead. I do have a few online questions as well. So there's a question over there. Yes, Kevin? Well, let me bring you the microphone first. Yes. There you go.

Yes, thanks for the presentation and a very impressive history. And last topic caught my attention to custom SoC. ASICs were very popular for a while in the '80s and '90s and then the cost, I guess, even FPGAs or cost of doing the new design got to be too high. But now ASICs are back again because you can't get accomplished any other way, and the cost doesn't matter to some of these companies. where would your product fit in? Is it because you can't do it otherwise? Or is it a low-cost solution?

It's definitely not for low cost. It's really for a differentiated IP. All the customers who come to us who ask for this kind of service or cooperation is because they value our IPs, perception systems, particularly AI accelerators and also our 2-nanometer expertise. All of that is the reason they came to us. If they don't appreciate any of the IP, we won't take that business anyway. And also, if they want to come to us for 2 nano because of the cost, I think they come to a roundhouse. I just -- Ambarella's famous for the 60% gross margin business, and we're going probably trying to maintain our corporate gross margin on that.

Congratulations. Maybe a follow-up on Kevin's question about the semi customer custom ASIC business. I guess, just walk us through how that might impact financials how much NRE, the NRE sort of rev rec is revenue? Is it contra R&D? And then do you sort of once you get into production, sell the chip and what kind of margins? I mean, some of the bigger ASIC vendors might talk about margins below your target of 59% to 62%?

So thank you for that question. In fact, that's one of the reasons we kind of hesitant at the beginning, but I think with a lot of enough interest, we think we should need to start talking about this. Maybe let's use the first semi custom shows an example is an NRE payment that already being -- majority are being done because we take all the chip. And also that the ASP has been a green that is closer to our gross margin. I don't think there will be a significant impact. But for the future projects that you can see there for the company if they come to us, they might have the expectation. We definitely want to figure out a business model that can benefit both of us.

So I would say that for the first product, the first semi-custom chip that we design for our customer, the impact to our gross margins a little. But moving forward, we need to figure out that business model, and we'll continue to report to you when we make more progress on this.

Let me do an online question and then we'll go to you, Corey. Several different ones. Let me paraphrase it. One, I think this question, it is for Muneyb and it was about the comment scaling through the channels. And if you could talk about the expectation for kind of the sequence of events or milestones that investment community should expect as we move that direction.

Yes. I think we have some clear milestones in this year to start, We started with the ISV community. And there's a sequence too, because what ISVs like Mel CX and Cogniac kind of bring is a lot of application catalogs. So people start getting visualization of type of use cases. So as more and more ISVs open up, we'll have a lot -- a catalog of different applications and use cases. then you start getting folks from -- so we have targets of by the end of the year, we'll have dozens of then you start onboarding channels and distribution and resellers that happens in a few, and you don't want to do too many. But these are value added, not what we already do. There's additional on that side.

And then have system integrators, a handful of them also come in. So by end of the year, we should have some progress in building out the priority is set of applications, more and more, the better. And the faster we're seeing, so it will be accelerated. Second, system integrators who can kind of bring this together, putting system hardware software and taking it to market. and then having distribution and channel resellers. By end of the year, we should have some pretty significant kind of opportunity sign up. The revenue flows will start coming in the future years, but onboarding them, as you know, it takes time.

I did miss part of that question, which congratulations on having your First 2 ISVs. I'm here at the show. Let's see. I think, Tore, you have a question. Let me just get in one more online. And that, I think, is for a combination of firm in Alberta. And the question is the question is about one of your competitors made an acquisition today a company that's experienced in the autonomous driving area, in the robotics market. And so can you talk more about your robotic strategy and how it will play out for you? And also, given that you have the SoC and Alberto talked about the software being applicable for that market, why not -- the question is why not just do the robot yourself?

Well, maybe let me ask the -- answer the question -- the last question first. They are just 11,000 different type of robots. There's no reason, no way we can address that. That we can do autonomous driving software because that's 1 software can serve all the OEMs, but I don't think that a robotic software stack, there's no application that we can set for that. But however, I missed that acquisition news that you mentioned?

Yes. Mobile Eye acquired a robot company.

Oh, I see. I'm sorry, I missed that. It's a busy day for me. But I really think that this is a really new opportunity for all of us. we talk about aerial drones, and I can expect that there are huge consolidation on the drone company in the next year or so and tell applied to all the robotic companies. So I won't be surprised to continue to see more and more people.

But for Ambarella, our strategy is very simple. First of all, as a semiconductor company, we need to focus on volume. And we focus on auto driving cars and also aerial drones first because the volume can support our R&D investment. But more importantly, the second point that Alberto made in his presentation that all our investment on time driving can be a help of the robotic software stack, that's definitely something we leverage on and we're going to continue to use that software stack doesn't -- not only just for the demo purpose, but we are going to open up the software just like we opened up for our OEM to license our software stack.

We're going to open up our software stack for our customer -- our robotic customer who want to license our software. It doesn't matter what perception, the decision-making VRMs or LLM, whatever we have been putting to our system will be opened up for our customer to leverage on. So for our robotic customers, they can focus on their expertise, and we will provide other pieces of IPs that they think they can come for us. But from the technology point of view, maybe Alberto may want to have a few words.

Just perfect, Fermi.

Sorry, that's not intentional.

Yes, Tore, please go ahead.

Tore from Stifel. Fermi, could you just double-click on the timing of the strategic change to go after more of a channel strategy? I mean, is it because physical AI is an inflection? Is it because of all the IP you've developed on the auto side that cannot be leveraged to new markets? Is it the maturity of Cooper? I mean why now, I guess, is the big question for me.

This idea started, I would say, at this more than a year ago. The reason for that is when we start engaging with the robotic pain, physical application, it becomes so clear, there are many, many opportunities. Most of them data company trying to come to this market, they -- none of them has a mature solution, right? So for us, in the past 20 years, we focus on key customers in the large areas Motorola, the large companies, that has been our go-to-market strategy.

But when we start engaging on this physical application, it became very clear very quickly that our traditional go-to-market strategy doesn't work. And from there, even before we engage Muneyb, we helped tremendously have joined to put this whole thing together, but we, at the time, even before we hired Muneyb, because we've become clear that this has to be a strategy for Ambarella. And really it's because we want to be the leader of Edge AI and the way Edge AI is developing is really go through this thousands of possible customers, and we need to find a way to support them. And I think Ambarella has the technology, silicon software and everything, but that go-to-market strategy has to change fundamentally change so that we can engage the customer quickly.

You're going to continue to hear us, particularly MuneybMuneyb and myself and Louis start talking about how this give you updates on this progress. how we engage with ISVs, how they help our go-to-market strategy, and that has to be a fundamental change for this conference today.

We're already about 12 minutes behind. So we'll take maybe another question or 2 in the room before we take the break for refreshments in the back, yes. As a reminder, we'll have refreshments and some light snacks right outside to help you pregame for your night in Las Vegas.

Just -- Fermi, do you think there's any need as you go more into like the SoC ramp for AI, Edge AI that you'll need connectivity assets? Or do you think like you can do this just on the device and you don't need to do that? And then yes, I have a follow-up.

I think your connectivity means how to scale silicon to build -- to address trillion parameters type of application. I think eventually, we will, but the connectivity connection that is required for edge influence is quite dramatically different than having a connection for the training system in the top data center. We definitely think that our -- when we approach to this market, we are focused on most of the HA applications that can be addressed by single silicon without going to multiple silicon solution.

But that having a multiple selling solutions to our customers to scale up on the Edge AI performance inference, I think it's critical. But that technology doesn't need to go expensive as data centers. So we already, in fact, Bob says with histology, he cannot talk about the details, but you can imagine that, that has been a hot topic inside Ambarella, how we want to have a next-generation a chip that can scale up for those potential influence solutions. And we will give you more updates later.

Would that be licensed? Is that organic? Or would you have to license that? .

I think it's definitely licensed. And that's definitely an area that we don't think that we have in-house capability. But also, we only need the physical connectivity, right? On the higher level there's anything we can do to help to integrate the basic or basically the link service into our system and how to integrate a service into our system most efficiently. I think from an architecture level, we know better than people outside for the inference engines.

Got it. And then maybe just sizing some of the stuff. So like to do a custom, semi-custom chip, the customer has to think that the revenue like the life -- let's say this is a 3-year chip. The lifetime revenue for this ship has to be -- I mean, it must be over $100 million be able to -- is that the right way -- like is that the right way to think about it?

We do that, we won't engage -- but all the customers we're engaging have easily can produce more than $100 million revenue per product. .

And then just lastly, I guess like I think 2 quarters ago, you announced some big edge AI wins, which was positive to see. Last quarter, we didn't really hear much of an update. But -- so I guess just -- does that indicate the momentum had changed in terms of design wins? Or is it just like lumpy and you kind of -- like just the timing of announcements is different? I guess this maybe you can talk to that a little bit.

So first of all, yes, it's lumpy because they're people are trying to figure out. So we are engaging -- the one quick update to you is we are continuing to engage multiple edge infrastructure customer talk. In fact, some of them in our space here to give some demo prototype. And that particular design win will go into production in Q2 this year and definitely that will trigger other conversation because that's brand new application and then how easy to -- the easier use of that product definitely will trigger -- give people an idea how to implement similar products in the space. So I eager to work with our first customer. And maybe first, a few customers to introduce the product, so that would definitely trigger more of the discussion in this space.

I just wanted to add -- just adding that right, so you're going from these edge infrastructure. You're going from design wins from more consumer to B2B enterprise. Those design cycles are a little longer and then more mature. So it's going to be a little bit more time. So as we see this, the end market you're going is B2B enterprise, and that's a longer design cycle, if you think there. So that's also the cadence that you're going to see differently. Unlike a consumer, we're coming next, next, next, because on hand to the consumer market. Enterprise is a little bit longer maturity and cycle, but not as long as like auto, but it will take just a little longer, right? That's it.

Had a related question today. It's not coming in online, but I saw maybe John and I saw maybe 20, 30 investors today. And this question is probably for you, Muneyb maybe Malhar that Ambarella has such a reputation of taking in data from real-time sensors, but now in some of these new applications like edge infrastructure, less time-sensitive data is needed. What has to change in the Cooper development platform or what's already changed in order to accommodate that type of workload?

Sure. I think A lot of the data at the edge, and I know as you kind of look at us and look at the computer and vision data real time is one aspect. But 80% of the data today is still time series data. They think about data that comes out of machines, PCs, in IPs, et cetera, et cetera. There's a whole kind of aspect of that that's untapped. And what you're seeing is multimodality is how do you combine this to? And what could -- what ready for already a lot of time series data is just text. And with time series and parameters, our ability to handle LLMs, our ability to handle text in a fashion is super critical to now start putting context like you put context as perception for autonomy.

Now you're taking context of vision with time series from robotic arms, from PLCs and putting a scene together. So recreating a scene in a factory, recreating a seminar. So readiness because a lot of the time series data is real time, but it's also tech space but combining that with vision and putting that context gives us a huge advantage. We're already ready and already there. This is where the use cases with ISVs and all this is important because the ISVs are building such applications. Sorry, go ahead, Malhar.

On the Cooper platform, we already announced or they're using VLM Gen as a tool. So that's basically -- it's not constrained on the time like the frame per second and all that. Instead, we have this prefill time or the time to first token that's very critical. And we can use batching for various things that we already announced and VLM Gens already does that. And the tokens per second. So in the edge infrastructure, the tokens per second or the decoding rate is less important than the time to first token where we can do batching and take advantage of our architecture. So some of the tools are already added in the in the Cooper development platform.

Thanks. Any other questions in the room? All right. Well, please join us outside for a reception. And also One other thing I want to mention that Jerome, who heads our IoT marketing and Jason Hwang, are you here? Jason? If not, he'll be outside, but he's responsible for automotive marketing and system solutions as well. So they will be available for the reception also.

Thank you.

Good afternoon, everybody. So happy to introduce today, Fermi Wang, the CEO of Ambarella, for maybe the sixth time at this conference -- several times.

Several times over the years.

Yes. So I always appreciate you being here. And a really interesting time to see you. You've had a year. This has obviously been a big year around the edge AI theme. You guys have sort of pivoted and refocused a little bit. Edge AI is now 80% of your revenue. Can you just talk about where you are in big picture, where you've come from, and where you're going?

Right. So first of all, I want to clarify one thing, which is that in my opinion, we're only building one technology, which is really edge AI technology for hardware and software. And this platform, hardware and software combination can serve many different applications, including automotive, including what we call IoT space on the IoT enterprise security, any new -- a lot of new application that we talk about drone, enterprise edge infrastructure. So in my mind, there are many market segment opportunity because of the edge AI technology will provide. So for me, we're going to continue to invest on the -- enable more and more edge AI application, particularly today is video plus AI plus low-power consumption. That's the focus of the company, and that will be the core of our revenue growth. Of course, that we are talking about moving to the edge infrastructure, maybe non-video data will become also a play in the future. But definitely today, we are focusing on any applications that can take advantage of our hardware and software platform.

You've got a really important point. I mean we wrote -- I think in 2019, our big debates report was on who will win the battle for edge AI inference and Ambarella's future prominently. So this is not something that's new. This has always been a focus. What's new is maybe the broadening out of application set beyond cars into these other markets. I guess -- can you maybe talk a little bit about automotive? Obviously, this technology has really critical capabilities that you would use in the automotive market, but it's just been slower to see these kinds of features adopted. Can you just talk about where we are in that?

So I think you are talking about really the autonomous driving Level 2 to Level 4. Maybe just before I answer that question, let me answer a little bit different question. Our Automotive business is still 21% of our revenue and growing. And the area that grow is really another interesting edge AI market that we just -- we developed starting two years ago. We call the AI telematics. The biggest customer is Samsara. And that -- in that application, the first product Samsara used is to do a camera facing -- outside camera facing inside of the driver to providing more and more AI function for ADAS and the driver monitor system. But now if you look at their promotion, they are talking about more and more edge AI functions. They want to integrate to the solution, not only include more camera, but start putting large language model into that space. So that is definitely another example. Two years ago, we didn't know even this application is part of our road map, but because Samsara has using our solution penetrated, make us realize that this technology can use -- this market can use our technology. Back to your Level 2, Level 4 question, this is definitely a tough year for autonomous driving. If you look at the -- it's not only us talking about this message, but a lot of people in this industry, both OEM, Tier 1 and also semiconductor company. The reason for that, I think there are two reasons. One is, with all the pressure from Chinese OEMs and also Tesla's FSD, people come to conclusion that -- most Western OEM come to conclusion that their product line need to be reshaped to a wide SoC to become more competitive. And second thing the software stack, the autonomous driving software stack become the obvious weakness for the auto OEMs, and they are trying to figure out what's the best solution for that. With these two reasons, we saw fewer RFQ available to the bidding. And also even with existing RFQ, they all push out trying to understand what's the right spec, and what's the right timing. So with that, I -- Ambarella's approach is trying to solve the obvious problem. We are trying to offer a software stack, potentially working software stack. But hardware, we are not downloading as a black box solution like our competitors. We are trying to sell the software stack, enable the feature functions that our customers might find that they can use -- they can license part of our software or maybe even a whole software stack that we are opening up as a licensing model as a white box solution. And we believe that by providing a scalable software solution that can scale easily from Level 2 and Level 4, in fact, we already prove that scalability to some of the OEM out there. And with that, I hope we can speed up trying to solve one of the difficult problem in the autonomous driving. But nonetheless, this year is definitely a very difficult year for any autonomous driving suppliers.

Thank you for that. And I know you've always kind of led with a software-first mentality on these types of products. I know a lot of your engineering workforce is software-based. Can you talk about the importance of that? And now as you sort of take this stack of software plus hardware, and you can apply it to a lot of different markets, can you talk about the role of software?

Yes. I think that's definitely important. One thing -- one statistic is important that although we are a semiconductor company, our engineering source -- the hardware ratio -- hardware to software engineering ratio is 1:6. I just show you that how important -- software side is important for us. But it's not only just software by itself on the -- for any silicon building a software SDK is important. And it's even important for us because every time we try to convince one of our customers switching their software platform from the NVIDIA GPU Engine to our platform, the biggest resistance is how you help them convert from the CUDA to our Cooper SDK. And that is not difficult. It's really the mindset that I spend so much time getting CUDA already, why do I need to spend time to convert to a different SDK. So the only way we can solve that problem is from the business side to convince people that there's a power advantage for them to move. But at the end, that software structure that is not only mature, but also flexible enough for people to move any CUDA software to our platform is fundamentally important for us to have a successful business because almost every customer we have today, they all use NVIDIA in the previous generation one way or another. So from that point of view, that's mature software, not only just SDK, but a compiler to move -- help people to port any train model to run on our chip and also even in the application level that we show people example how to run application on our software platform. All of those software are important for us to win design wins.

Great, thank you. And so then as you talk about those new verticals, can you talk about what they are? You've talked about edge infrastructure. Can you define kind of what that means, and what some of the examples are?

Right. So in fact, that I would just say 12 months ago, our largest market is still enterprise security, and it's not true anymore. It's not that the enterprise security slow down. In fact, we still see very strong growth on the enterprise security, it's really that other areas of edge AI start growing and growing faster. We talked about two new markets in the earnings call two quarters ago. One is drone, one is edge infrastructure. And drone -- these two are totally different application in my opinion. But funny thing is they can use the same hardware and software structure that we are providing to our customers. And for the drone, it's really -- we are offering two type of solution. One is if you only want drone to capture video, in fact, one of the products that our customer introduced, they put a 360-degree camera under the drone. So when you fly, you are not only seeing one direction, you see everything surround you. And that really help people to navigate the drone if you're using manual navigation, that's one product. The other product is really -- a drone is another type of robot. And all everything we develop for autonomous driving car apply to autonomous driving drones. In fact, you can view that most of the drone today, we call a Level 2+ drone because you still need people to manually control it, but there's a lot of already autonomous function in there. And the drone will move to Level 3, Level 4 probably faster than the cars. From that point of view, you need a very powerful domain controller on drone to perform those functions to avoid objects, to navigate, to understand the performance. So from that point of view, I think we -- our CV3 family product that we define for the autonomous driving will eventually go to the autonomous drone. So that is a market that we think is important. Although the market is still relatively small, it's 10 million units consumer drone today, mainly dominant by the DJI, but the window opportunities opened up because DJI got banned by the United States Government. So that 1.5 million consumer drone market in the United States opened up for fight. And we are seeing multiple customers trying to fighting on that. So that's just one new opportunity from zero to meaningful to us very quickly. The other one you asked about edge infrastructure, which is even more important for me. Edge infrastructure means in the past, we sell our solution to, we call edge endpoints, cameras or any form of cameras into a different device. But edge infrastructure is really to aggregate different type of cameras and performing higher-level functions in a box that we never do before. So in fact, we announced our first product two quarters ago. And the applications -- for that particular application is very simple. It's try to aggregate multiple camera feed. And for example, in this hotel floor, say there are 20 cameras. Most of them probably is not even AI-enabled, let along the ChatGPT. So if you want to upgrade those camera to be running ChatGPT-type models, the easiest way is in your engineering room, in this floor, plug into appliance box with one of our chip and fit those 20 camera fit into that box and then you run the large language model on that box and so that all the fit can suddenly be upgraded by the ChatGPT ready. So from that point of view, that becomes the easiest way to upgrade installed base camera that for security camera alone, it's 2 billion installed base worldwide. So that -- we are talking about a huge opportunity, not alone for the hotel, but retail. Any retail store probably have 4 to 8 cameras. You can easily upgrade in the same way. So we are viewing that as opportunity. But we're still talking about video-related edge infrastructure. They're -- definitely, they are non-video-related edge infrastructure. I think all the corporations start talking about how to upgrade using training their own LLM, but they want to run the LLM on-prem servers, not instead of trying to run at AWS or other cloud services. From that point of view, you need on-prem edge servers or edge infrastructure boxes that can provide seamless performance. And why we have an advantage because all of the application we talk about power efficiencies continue to be important. The engineering room here, I bet you, is now well air conditioned. The power consumption is definitely a problem. Even the power supply to the box sometimes limited by the configuration. So from that point of view, I think that our power-efficient solution for N1-655 is suitable for that.

Maybe we could talk a little bit about that. In the past, I feel like we've sort of moved a lot of the intelligence onto the camera, where you're doing a lot of the edge AI kind of resident in the camera. It's very clear. You guys have been pretty dominant in that business. The value proposition is pretty clear of moving that intelligence into the camera. When you talk about moving into an edge-based kind of box, do you still get the same benefit of performance per watt? Are you more putting yourself in competition with GPUs and things like that? Just what's the value proposition...

Right. So I still think that the performance per watt is important, particularly for the first application I talked about. The engineering sitting here, in fact, a lot of the box is supplied by power over Ethernet. That -- so basically, your AI performance for the box is defined by how much power efficiency that you can get out of that chip. So yes, power efficiency continue to be important. But there's another driver is really the -- to the box itself, most of the GPU box require heavy air condition, water cooling system. I don't think that's widely available in a server room even in my company, I don't have a water cooling system in there. So from that point of view, if you really want to have a powerful on-prem servers, I think that power efficiency continue to be an important factor.

Great. I guess maybe if we could talk a little bit about the surveillance market, more home surveillance and things like that. I know that used to be a bigger category for you. There's a lot of price sensitivity. The cameras have to meet really low price points. But it also seems like the value proposition is really strong. And as a consumer of video cameras where you see all you can do is turn the sensitivity up and down, there's really limitations to that when you talk about doorbells and things like that. Is there going to be an application for you guys as the sort of intelligence in those devices grows again, or to what degree have you had to walk away from those opportunities?

Right. Well, if you ask me the question 12 months ago, I was pretty hesitate. But today, I am convinced there is definitely opportunity. If you look at all of the home security suppliers that are working on Ring, Nest, they all are enabling a new service by running clip type of a vision language model on the server side. So the video streaming from your home to the cloud, at the cloud, they store the video and apply this clip vision language model on that so that you can provide more service. They are charging $9.99 per month for that service. But we all know that Ring and Amazon and Google can do that because they control the cloud. But all the other major consumer security camera customers, they -- when they try to use the cloud to [ probably ] service, they are limited by the cost and also the transmission bandwidth, the storage costs and the processing costs on the cloud. In fact, I will -- in fact, when I talk to them, they are convinced that if this kind of similar service can be offered using an edge device that -- if the clip model can run on the camera. And although you pay a little higher price on the processor and the memory, but it can easily be compensated by the lower cost on the cloud as well as the transmission cost. From that point of view, I think that new service enabled by ChatGPT -- by vision language model is a clear way to upgrade that service. And I believe that our new chip can run 2 billion parameters ChatGPT model for 2-watt chip. That will definitely enable this kind of service in the future.

Yes. I mean the value of these applications really seems to be growing. Can you talk about robotics a little bit? And I guess it seems like drones is on the path there. You see -- you go around Los Angeles, you see little refrigerators driving around delivering stuff. Like it seems like there's a lot of -- before we get to the humanoid robot upstairs, there's a lot of applications for vision in these robots. Can you talk about your view on that market?

Yes. It's become clear. In fact, I have been saying this before, I view that autonomous driving car is just one schedule type of a robot. That applies to drone, too. So I think today, if you look at the biggest robotic application is autonomous driving car and the drones. And new application are popping up. And I think -- so when I look at this robotic application, I focus on what we call mobile robots. Any robot they need to move and that will take -- can take advantage of all our investment on our CV3 technology for design for the autonomous driving. So AMR or any other human role in the future, any drone need to move, and they need to understand environment, need to decide -- finding a way to maneuver over different objects and design the path they will need to move. And then finally decide what kind of function you need to do. This really sounds like autonomous driving car for me. So from that point of view, we believe we will continue to focus our robotic development on the mainstream revenue generation models first, meaning cars and drones and use that to fund -- continue to fund our investment in this direction. That's why, in fact, we definitely continue to invest on autonomous driving car because everything we invest in that direction will be heavily reused in the robotic application. But the biggest problem for me in all the new robotic application is, it's very segmented. There are a lot of developers, and they're all trying to demo and showcase their products in a prototype form. How to enable those guys is important for me because we are not talking about 1, 2 large customer anymore. We're talking about hundreds of different robotic applications, and we need to engage with them. So we do have a plan. In fact, at the CES, we're going to have a technology conference. We're going to highlight our new product and new technology, and we definitely will highlight how we want to develop a new go-to-market system that to address this robotic application.

Yes, it's interesting because you've historically had fairly concentrated customers in automotive, enterprise security markets like that.

That's right.

Okay. Makes a lot of sense. One of the questions we get a lot, particularly when you start thinking about these more consumer-centric applications is gross margin. You have a model of 59% to 62%. You've had a tendency to walk away from markets that don't -- where you don't see the value a little bit. Is that going to be the right margin structure as you think about your future business mix?

Well, in fact, all of the consumer occasion that you're talking about, looking at our drone, we're talking about a $25 chip. So that in fact, the customer drone -- the consumer drone they are selling to $1,000. So it's not cheap. So definitely, there is value and people want to buy high quality, particularly if you want to compete with DJI, the quality has to be one of the major concerns. So from that point of view, definitely, price is important, but gross margin, I think, is important. The most important thing for me in the last few years, we gradually start to realize that while we try to maintain the 59% to 62% gross margin target, we are willing to trade off a little lower gross margin back to higher revenue and therefore, higher leverage on the operating margin side. That's the thing we are trying to talk about. I think we're only willing to do with large customer. And today, in the past, we talked about automotive customer can be one of them. But today, our largest customer is on the consumer side. So it's not the consumer side market driving us to lower price. It's really that they have the volume, they have the potential higher revenue growth for us, and that's where we are willing to trade off our gross margin.

And drones in particular, I mean, DJI was once a big customer for you guys. And I know geopolitics was part of the issue there. But is it also that there's just a lot more value going into these drones now when you were doing more kind of image sort of capture, now you're doing more image analytics.

Right. So if you look at how DJI drone has been used in -- although it's a consumer drone, for the consumer video capture, but they have been reused in many different applications, right? And I've seen people using DJI drone for inspection for the many different type of other application that's not possible to use in any other technology. So drone, to my surprise, when we worked on drone 10 years ago with DJI, the host market was like 1.5 million units, and people think that will be saturated maybe too. Today, we're talking about 10 million units of consumer drone. And out of that drone market, 9.2 million is consumer or prosumer and 800,000 is commercial. So I do believe that this drone market will continue to grow because people start identifying more and more commercial application. But I think the right approach for me is we need to focus on the customer who has an ambition to be the player in the consumer side or prosumer side so that they can drive to the scale to get the best -- from commercial scale so that they can compete in that 10 million units market. And with that, I think they will have a capacity to develop a solution for commercial drones. That commercial drone is a lot more profitable. But however, if you don't have the scale, you won't be able to compete with a company like DJI, which is already in the market and dominating the market. So I think that the market -- the business model approaching this drone market is very important. I think that technology matters, quality matters, but more importantly, there is already a dominant supplier. You need to find a way to coexist with that.

And just to double-click on that, the military drone market seems like a very obvious application where you really need good computer vision, but it's also one that's specialized people that are optimized around military applications. Could that be an application for you guys as well?

Well, we don't design chip for the military grade. So however, I do believe some of our customers or design houses building a camera that with our commercial grade chip and selling to that market. But we don't have any customer is really in a military level of customers.

Okay. Great. Maybe if we go back to the automotive opportunity, I mean the technology that you've delivered is really a breakthrough, and we've seen that years ago. And you've gotten wins with some of the biggest Tier 1s that specialize in autonomy. And we just haven't seen adoption yet. I guess where do you think that stands if you look over the next 3 to 5 years? Can people look at the advances of Tesla FSD and do nothing? Do you think that there's a call to action there that we need to start implementing some of these features?

Absolutely. In fact, that one of the things we talk about this is really a bad year for the autonomous driving, but people are still trying to figure out how to compete with FSD. And now I starting hearing people that are talking about end-to-end model in the Western world, which is a good thing because without that, I don't think we can compete with FSD. But however, to run the end-to-end model, both on the hardware side and the software side is a huge commitment. We know that because if you look at the software model that we work with VisLab, the company we acquired, and we take a few years to get to a point that our software stack is true large model. But to make that combine true large model become one end-to-end model, it takes effort. But we know how to do it, we do it, but it takes years to get there. So I really think we talked about this just a few minutes ago. I think one of the biggest bottleneck for us to -- for Western -- and for us to get penetration into that market is we need to start selling our software in a way that adds value to our customer. How to get a better perception with our perception module that we can do sensor fusion between a camera and the 4D image radar and also running everything in a large end-to-end model that runs on our 685, we can demo it. When we demo this and take that software ready to be in production, I think that's where one of the solutions, we think we can help that to resolve the current situation that people are looking for software stack, and they haven't found one. But more importantly, we believe our approach is scalable. When I say scalable means I think our approach can scale from Level 2 to Level 4. Of course, you need to reduce the number of hardware, number of sensor. But if you are training that model properly, you should be able to scale your performance down in a way that you can easily using a end-to-end model to address Level 2+ and Level 4 applications.

Great. I want to follow up on that. Let me see first if we have any questions from the audience.

Just wondering what do you think the market is missing?

About Ambarella?

Yes, yes.

Well, I think 99% of the AI investment is still on the cloud. Although I think a lot of people here to listen to this presentation because you appreciate edge AI, but I think the majority of the industry still think edge AI is on a niche. Maybe that's -- if you don't think they could become big, then that's probably one of the reasons that they don't pay attention to Ambarella. But I really think personally, I think give another 10 years, I think edge AI can be as big as the cloud because there are so many applications that you're looking at today has to be implement on the edge robots, it's obvious one. There are many other applications. If the latency matters, if the privacy matters, if the private data matters, it has to be on the edge AI. So from my point of view, I truly believe that when people realize there are new applications that will require running the AI on the edge, that we should get our fair chance to be competing in the space.

Questions?

Maybe just to follow up on auto. I mean, how much of these advances are tied to EV because it feels like with internal combustion implementing a higher degree of autonomy, there's just a lot of technology challenges that need to be solved with physical actuators and things like that. It's just easier if you're redesigning the whole vehicle around EV to start implementing these features as Tesla has as Rivian has. I guess, do you agree with that? And it seems like that's a really strong positioning for you guys because a lot of the stuff that we're seeing in internal combustion is not going to translate into an EV world, they just don't -- can't meet the power budget that you can meet.

Well, if you ask me this question 12 months ago, I will agree with that. EV and autonomous driving really come hand to hand. But now with the new people start delaying the EV distribution and slowing down the time line for the EVs, we start hearing a lot of OEM customers start saying how we can implement autonomous driving on ICE cars. And in fact, we start seeing RFQ bidding on that because those cars need to have the autonomous driving to be -- stay competitive. So with the EV schedule got delayed, you really bring more attention to the ICE car and the autonomous driving. So I think although we just start hearing it, I won't be surprised to start seeing autonomous driving function being able on ICE.

I mean it's incredible to me that we've had the breakthroughs that we've had on reasoning models at the edge, and we've actually moved backwards in autonomy. It seems like we can only move forward at some point.

Well, I don't want to comment on the political environment, but that's a reality to deal with. But reasoning model, let's give you another example. We can run a reasoning model on our 2-watt chip today. We talk about this that our CV75 is a 2-watt chip, we can run a 2 billion parameter DeepSeek model on that. But the problem is what's the real application with the reasoning model for edge device. I think whoever figured that out going to be one of the biggest potential customer for me, right? We are not the one to drive application for edge AI, but we are enabling all the functions that not possible in the past, but now we are definitely thinking that with our silicon, we enable something that's impossible and hopefully, our customer can take advantage of that.

Great. Well, congratulations on all the progress, and we'll wrap it up there. Thank you very much.

Thank you.

Good afternoon, everybody. For one of the last sessions of the day, we have here with us Fermi Wang, the Chief Executive Officer of Ambarella, Inc. Fermi, thanks so much for being here.

Yes. Thank you for inviting me here.

Great. So I've got a few questions of my own, but if anybody in the audience has questions, there's going to be a QR code up on the screens besides me, and those will show up here, so I can ask those at the end of the session.

Okay. So to get started with, you guys have sort of transformed the business and IoT is now driving the majority of your revenue, and it's kind of clearly overtaken the auto end market. But on the other hand, your 2030 addressable end market estimates are more auto than IoT. So how should investors think about Ambarella today? Is it an IoT edge AI company? Or is it more auto-centric once CV3 reaches scale?

Well, the way I think about this is Ambarella is an edge AI company, which include automotive. If you look at edge AI, how it defined, my definition of edge AI is very simple, that you -- for any AI application, you don't require connected cloud. Majority of AI is running on the edge on some device. That's edge AI for me. By that definition, autonomous driving use edge AI device. So the way I look at it, Ambarella is focusing, investing on the edge AI market, including autonomous driving, autonomous drone, robots and any other system that require automation.

Okay. And how do you think about the balance of those 2 end markets? Is that -- do you have a strong view where that's going? Or do you just make your products and wait for the market to tell you?

Right. So I think that autonomous driving, obviously, the largest edge AI market today. And so that's why we still continue to think that it's a huge opportunity for us, and we're working hard to continue to secure the first major OEM design win for us. However, that other edge AI market continue to popping up in the last 2 years, and we start seeing more and more opportunity on the other edge AI device, we call the IoT side. It doesn't matter it's autonomic drone, edge inference boxes or -- that we talked last quarter or AI video telematics box company like Samsara, all those new applications that we didn't -- were not part of it 2 years ago, now all of the opportunity show up, and that continue to drive our revenue growth.

So my opinion is auto continue to be the largest opportunity until it being taken over. And I think that there are definitely opportunities in any other space that will continue to become a bigger portion of business. But if you look at only 2030, I think automotive will be probably 50% of our potential TAM or SAM in that time.

Okay. You emphasized that there's a common hardware and software platform that you develop internally for both IoT and auto. So it's not like there's a team doing research for IoT and another for auto. So -- and at the same time, you've also shipped more than 36 million SoCs, right? So really significant installed base. So how durable is that platform advantage that you started to build with an Ambarella as you've got larger competitors and new ASIC vendors that are starting to target this like edge inference opportunity?

Right. So first of all, I think that the same hardware and software platform across all our silicon portfolio is important because, for example, one silicon we build can serve automotive, can serve robotic, can serve telematics, can serve enterprise security with same software SDK, but we help our customers build application on top of that.

So from that point of view, and it's not only just one silicon, right? We have a family of chips. For example, CV2 family, we have 6 family of chips with different performance level, different price points on each chip. And the beauty for our customer is they must have multiple different product line from the high end to the low end. With the same hardware, same software, with a family of chip, they can pick any performance price point to serve the one market.

But with the same software, they can compile and go to a lower-end chip or high-end chip and build the product without really redo the -- reinvest on the R&D dollars. That becomes a really important -- really a factor that we can continue to secure to our investor and our customer because they understand the strength that we have this unique offering that with multiple chip and they only do software, hardware once and can do multiple products, I think that's definitely helping us to keep our customers with us.

Got it. And when you look across the market at the other people trying to address the same end market as you, you've got some really, really large-scale players as well that you're up against, right, some much larger than you. So does anyone have a larger platform advantage than you have?

Well, in different place, definitely, right? For example, NVIDIA, which is really the largest possible AI company out there, I think they have a very strong position on the cloud, on the data center, which we will now touch. But on the edge, in fact, there is definitely a lot of discussion about -- in fact, a lot of our robotic customers using NVIDIA chip as a first prototype or first generation technology.

And I was talking to one of our customers. They tell me why you want -- how you want to compete with NVIDIA on the edge side. I said for the robot, unless you are willing to put in the water cooling system, you don't have a solution for you. So definitely that I think CUDA and NVIDIA chip is very powerful and has performance, but it comes with some issues that I think we try to address. We try to count to edge device, which the latency matters, the power efficiency matters, cost matters, the bond cost matters and more importantly, in certain application, video quality matters. All the things are my benefit.

So I think from that point of view, we do believe that that's the reason we focus on edge AI instead of trying to try to compete data center. I think we have a better position in those in the market.

Got it. So portable video has been one of the really major IoT growth drivers. There's action cameras, body-worn cameras, aerial drones. How broad is the customer and SKU footprint in portable video today?

Well, first of all, it's beyond that 3 type products you talk about. There's a wearable camera. There are people using that web camera, they're doing the video conferencing and drones. So there are -- in fact, the portable video category can go into 6, 7 different product lines. And we have multiple customers in this -- we call the portable video and other segment.

So from that point of view, I think this continue going to be a growth area for us because some of the market is really old market, for example, webcam, but webcam was built for AI using AI to improve video quality is a brand-new offering. And we definitely think this is offering something that we can add value to. But also that drones with 360-degree camera, I really think that's a breakthrough because there's a customer going to build a drone with a 360-degree camera.

And when you put on goggle and watching the drone flying around, just turning your head, you see the surrounding of the drones. That's a perception you've never seen before with any kind of drone. So from that point of view, I think this has become interesting with AI, more innovation can go into products. Therefore, we think with more AI technology, adding more AI performance in there will trigger even more interesting products in this portable video categories.

It's interesting looking back because the company's roots are very much in security video, right, in this one kind of narrow category, and it's becoming a smaller and smaller portion of your mix over time as nonsecurity things are far outgrowing it. So how do you see that mix developing over the next 3 to 5 years? And what are the trade-offs as those other categories grow more in terms of pricing or gross margin?

First of all, our enterprise security, the percentage dropped, but revenue grew. It will show you that -- in fact, our overall revenue is growing. And in fact, one thing I want to say, for example, one silicon we use in that enterprise security is called CV5. The same chip is using in the drones, using in the portable video, using in the security camera and using in the video conferencing and also using in the car. In fact, Rivian is using that chip in their car.

So that just showing you that, that can really reuse our R&D dollars for same silicon, same software with a totally different application that help us to minimize R&D and leverage that to enable more different applications. In the future, I expect the market that we're going to focus on is the application can reuse our hardware and software platform and just by using -- by adding go-to-marketing and [indiscernible] to support the new application, that's the market that we're going to focus on.

I want to go back to the aerial drones for a minute. So they're clearly a meaningful new leg. How large do you think the volume opportunity is in terms of units over time? And are the barriers from here, are they more technical? Are they competitive? Are they regulatory? What needs to happen for that category to grow?

I think it's all above. Let me talk about this. But drone today is roughly 10 million units. If you take out the toy drones, that's not accounted. All the other drone combined 10 million units, including we call prosumer drones and also commercial drones. Prosumer drone is -- I define that as people use that to capture video, right, and to produce video. This is 9.2 million units and dominant by DJI, which is well known.

And nobody really come close to compete with DJI in the last few years. Until recently, the window opportunity opens. And that opens because the U.S. government put a ban on the DJI drone in the United States. And that creates roughly, I would estimate 1.5 million units of market available for the second best. And I think that creates opportunity for us. DJI have been using their own silicon. So that is now available to us, but the market, I think, opened up for that. I think there -- we're going to see multiple drone company going there to -- go after this market. And then hopefully, with their momentum, they can go to other market and compete with DJI, but that's definitely a difficult situation.

In terms of regulation, it's very well documented, right? In U.S., there are definitely regulations about what kind of drone you can fly, what kind of rule you need to apply, all of that, I think most of the drone players will follow that rule. I don't think that's an issue. So from the point of view that I think there's definitely a window opportunity for new players to come in, and we would like to be one of them.

Shifting over to auto. So you continue to invest behind the CV3 family for L2+ to L4 ADAS feature domain controllers. So what are the gating factors that are controlling the pace of you announcing your first OEM award?

Right. I would say this year is definitely particularly difficult for Level 2+ car for the Western players. If you look at, we talked about -- in January, we talked about the VW bid, right, and we lost to Mobileye. But after that, even our competitor didn't talk about many design win opportunity in the Level 2+ or above.

I think there are 2 reasons behind it. One is, I think a lot of people still cannot come out a video -- a software solution that really can address Level 2+ efficiently. And that really delayed the project. The other thing is really that the feature and performance of Chinese OEM product, the L2+ and also Tesla's FSD proved to be really powerful products and try to compete with them. I think a lot of the Western OEMs are still trying to figure out what's the spec, what kind of time line they need to. So I think they delayed the decision because of those 2 reasons.

And so we are continuing to believe that this is a market we need to focus just for the reason that this is the largest opportunity for the edge AI market, and we need to continue to invest, and we continue to work closely with the OEM.

But there's another reason. In the last 2 years, it's become very clear that our investment on the -- for CV3 for the domain controller, both hardware and software can directly apply to robotic, particularly the mobile robots, right? In fact, if you look at what the requirement for mobile robots, you need a great perception to understand the environment, then you need to path planning to plan how you move around, then you need to a decision-making in the autonomous driving car is how you drive in the robots is how to using environment information to perform other functions. So from that point of view, from the perception and path planning, everything we did for the CV3 can be totally reused for this function.

So with autonomous driving continue that we continue to focus on. But with the robot momentum on the market, we definitely believe this is an area that we can reuse our technology. That's the reason -- another reason we continue to focus to do the technology development.

Yes. So you mentioned Mobileye as one competitor in the automotive space. There's also Qualcomm and NVIDIA. There's a lot of emerging Chinese vendors as well. So how do you like the competition in that space? And what's your right to win or your competitive differentiation against them?

Well, I think the biggest problem for us, we are the smallest company in that list of the competitors. However, I do believe that we have the best technology among them in terms of power efficiency, the performance per watt, also providing a software package as a licensing model instead of trying to bundle hardware and software together. So from this point of view, I do believe we still have opportunity to continue to gain momentum in CV3.

So competing with large corporation is difficult, but the key is today, most -- like I said, most OEMs trying to figure out how to compete with Tesla is technology, right? And I think that we believe that if people really focus on technology differentiation, we have a chance to be selected as a supplier to most OEMs.

And once you win your first one of those opportunities, can you size how big one of those programs could be over its lifetime? And what sort of slope or ramp we would see once won?

Well, let me use the VW project that we lost as an example. We were runner up in that project. Had we won that, it's easily $700 million or $800 million lifetime value to the company while we are doing $390 million targeted revenue this year. So that impact can be huge to our company in terms of momentum of the revenue growth as well as the investment in the different directions.

Yes. So in video security, competitors like Hikvision are normalizing inventory and then there's some export control considerations as well shifting around. So how is that market looking for you? And how do you feel about your long-term share opportunity relative to like the domestic Chinese?

Enterprise security, you are talking about?

Yes.

Okay. In fact, we are out of China already. In fact, 5 years ago, when Entity List was implemented by the Trump administration at that time, we were banned from selling to Hikvision, Dahua. And right afterwards, the Chinese government basically says they prefer using domestic solution in China. So we have 0 revenue from China market for our enterprise security.

However, that help us to do -- we kind of transition that our focus to focus on non-Chinese solution. In fact, if you look at the major enterprise security camera vendors outside, we are probably the biggest supplier to all of them. And so from that point of view, I still feel that enterprise security continue to be an area that will give us more growth. And the growth is not coming from -- it's not only coming from the size, the growth of the market share, but the AI ASP that we continue to drive higher and higher. Our ASP 6 years ago was $6. This quarter was $16.

And the reason is because we introduced AI chip in that 6x and gradually move up our AI ASP. And the AI ASP growth will continue because all the chips that we introduced, for example, our third-generation AI chipset starting from CV75 and CV72, the ASP is higher than our CV2 family by 30%, 40%. And then the CV3 is close to $100. So we expect our ASP will continue to grow. So I think -- to go back to enterprise security, the growth -- of course, some of that growth will come from the market share growth, but it's limited, but ASP potential growth can be a driver for that market.

That's actually one of the places I wanted to go next. So for fiscal 2026, the guide is that revenue growth will be more or less even amounts from units and ASP. Clearly, you've got some much higher-end SoCs that carry much higher pricing. So in the kind of nearest term, there's a lot of potential for ASP growth. But over the longer term, how does the unit versus ASP mix shake out?

It's really about what kind of design win we can get. Had we got the VW, the unit can grow also significantly. So I think that, that's why we need to continue to focus on bigger TAM of the market, right? So when we look at market opportunity, we definitely want to say, if there is a 2 different market, the selection has to go to the higher market TAM so that we can get more potential growth out of that.

So definitely, we want to go to the bigger market so we can grow unit number. But however, the ASP growth is really about more and more -- deliver more and more AI performance really which is required by our customer. And more AI performance means people want to implement not just higher tops, if you may, but also different type of model. Our second-generation AI chip, we focus on CNN type of AI model. Our third generation will focus on transformer or Gen AI type model.

And majority of revenue that we see today is still from our second generation. Our investment on third generation haven't generated much revenue at all for us. So I think when we move to the third generation, ASP will go higher because of performance, and that will definitely continue to help to drive our revenue growth.

And each generation is clearly being introduced at a much higher performance than the last. But if we think about over the life cycle of any given chip, what's the deflation or the kind of cost down that customers are expecting at any given level of performance? [indiscernible]

Is. Yes, this is a faith of a semiconductor industry, right?

Yes.

Customers expect you -- every year, you drive down your cost. There are 2 things we need to do. We drive down our cost of supply chain at the same time, which is getting more difficult because we are already in 5 nanometers. We're going to do 4 and 2 nanometers. With limited suppliers in this space, driving down that cost is getting difficult. So however, while you do this, you continue to drive down the cost or drive down your ASP to a certain product, it's getting more important, you overlay with new product line, which has a higher ASP and also kind of compensate the ASP drop and the revenue drop because of you give the better pricing to your customer.

And to overlay that product line is important. And you cannot just fake overlay, but most of the time, if your customer telling you, you need -- this is a new feature, new performance, new video quality, we require and willing to pay more. This is the better way to overlay a new product line, which give us a benefit of compensate that the product -- the price reduction every year.

Do you feel that you're approximately able to match the cost-down expectations of the customers with what you're experiencing on the cost side and sustain gross margins? I think there's a 60%?

In the last few -- the last 20 years, we have been successfully doing that. Then going to 2-nano will be an interesting experience for us because 2-nano really is really powerful, right? But however, it's very expensive. And we haven't -- we're definitely counting on our supplier to continue to work with on the cost structure and the price. But I think that's -- it's our -- definitely our goal, and we believe that we will continue to maintain our long-term gross margin model of 59% to 62%.

What about operating expense? So right now, it looks like OpEx is growing in the low teens kind of range. There's a lot of investment on new products and field engineering and things like that. So -- and the world is changing fast and you want to be ready when it gets to your right. So how are you balancing continuing to invest in what you see as the most interesting opportunities on one hand with showing operating leverage in the model as you're growing revenue? How do you weigh those?

I think getting operating leverage is the most important financial metrics for us. We continue to value our revenue growth and to determine what kind of OpEx growth every year for us. We have to make sure that we continue to show up with operating leverage for our investor and also for the health of the company.

Okay. So when you think about your kind of cash generation in the near term, how do you think about allocating between R&D, potentially M&A and returning cash to shareholders?

Right. One thing that our investors continue to give us feedback is that Fermi, you have been spending 40% of your total revenue on R&D. I think that's a price to pay to compete with NVIDIA or Qualcomm. But however, we need to be -- continue to show 2 things. One is we need to show -- continue to show the leverage on the operating profit that we talked about. But the other thing is on the cash -- the operating cash point of view, we have been positive on the operating cash for 16 years in a row. So that just show you that we are trying to balance both, right? We want to invest heavily, but we know there's some financial disciplines we need to follow. And that's what we try to do.

In fact, this quarter, we generated $30 million cash flow. And then also that our cash position is roughly $280 million, I think very healthy for our size. And we'll continue to invest R&D, but we think the discipline we talk about. But also we are looking at M&A. I think although today, I think the market valuation is really high for the private companies, but we continue to look at opportunity, particularly on 2 things. One is algorithm, one is software and that related to our market, particularly on the edge AI-related market. I think if there's anything that close to our space and the things that we cannot do internally, I will definitely continue to look at M&A as a way to improve our -- offering to our customers.

And what's the return from an acquisition like that? Is it something that helps kind of establish and increase the moat? Is it something that with more software capabilities, you can then charge more for the products, and so it just flies directly into revenue?

We have been very selective on our M&A. Most likely is that in the product offering that we need to go, but there's a piece of technology that we -- that is required for serving certain markets, but we don't have that. And we look at it as taking too much time for us to develop ourselves. That's where we're really looking at M&A at a time. So it's really about making sure that we don't have holes -- technology holes in our offering to our customers.

Got it. One more question on M&A. Clearly, the strategic value of edge AI assets is really high to many larger players. There's been a lot of conversation about consolidation as well. So how do you weigh -- for Ambarella, weighing independent and getting to own your own destiny on one hand versus your potential value to an acquirer on the other?

Right. In fact, I have been saying that we -- I believe that if there's a bigger -- much bigger platform owns Ambarella technology and willing to invest on that, I think we can grow faster than we are, right? And that -- I still believe that. But I think there are 2 assumptions. One is people willing to take over this and continue to invest on that. I think that's the way -- the kind of what I'm looking for.

Got it. Well, it's been fantastic to catch up with you today. Thanks so much for your time, Fermi.

Yes, [ Grant ]. Thank you very much. Thank you.

Good day, and thank you for standing by, and welcome to the Ambarella's Third Quarter Fiscal Year 2026 Earnings Conference Call. [Operator Instructions]. Please be advised that today's call is being recorded.

I would now like to hand the call over to your speaker, Louis Gerhardy, Vice President, Corporate Development. Please go ahead.

Thank you, Victor. Good afternoon, and thank you for joining our third quarter fiscal year 2026 Financial Results Conference Call. On the call with me today is Dr. Fermi Wang, President and CEO; and John Young, CFO. The primary purpose of today's call is to provide you with information regarding the results for our third quarter fiscal year 2026. The discussions today.

Responses to your questions will contain forward-looking statements regarding our projected financial results, financial prospects, market growth and demand for our solutions, among other things.

These statements are based on currently available information and are subject to risks, uncertainties and assumptions. Should any of these risks or uncertainties materialize or should our assumptions prove to be incorrect, our actual results could differ materially from these forward-looking statements. We're under no obligation to update these statements. These risks, uncertainties and assumptions as well as other information on potential risk factors that could affect our financial results are more fully described in the documents we filed with the SEC.

We're starting the call. We hope to see you at one of the following investor events that we have scheduled during the fourth quarter. December 2 and third, we will be at the UBS Global Technology and AI Conference in Scottsdale, December 9 and 10 at NASDAQ's London Conference January 6 from 4 to 5:30 p.m. at our CES location, we'll be hosting a technology and product briefing. January 6 to 10 and we'll be hosting more than a dozen cell-site analysts tours of our CES demonstrations again at our CES location in Las Vegas in January 17 at the Needham Conference in New York.

Access to our third quarter fiscal year 2026 results press release, transcripts, historical results, SEC filings and a replay of today's call can be found on the Investor Relations page of our website. The content of today's call as well as the materials posted on our website, our Ambarella's property and cannot be reproduced or transcribed without our prior written consent.

Fermi will now provide a business update for the quarter. John will review the financial results, and then we will be available for your questions. Fermi?

Thank you, Louis. Good afternoon, and thank you for joining our call today. Before we proceed, I want to let you know that Les Kohn, our Co-Founder and CEO, will be stepping down from the Board of Directors to become our Chief Technology Adviser.

We will continue to oversee our technology direction and development, but without management responsibilities and we reuse and commitments. Les and I have worked closely since 1994 across 4 companies. I'm grateful that Les will continue as my close partner for over 31 years and beyond. And history the best ACIs. I'm happy he will have more time to pursue actions, but I will definitely miss our daily conversations on various topics.

Turning to our fiscal third quarter, we are reporting another strong quarter with both branding and non-GAAP EPS exceeding expectations. We achieved record quarterly revenue of $108.5 million, slightly above the high end of our guidance range. HAI revenue, we defined as a product that integrates 1 of our proprietary deep learning AI accelerators was about 80% of of our total revenue, representing our sixth consecutive quarter of record HAI revenue.

We have increased our fiscal revenue guidance, which projects an all-time fiscal year total revenue record for Ambarella. The strength in our average selling price and with breadth of demand we are raising our fiscal 2026 revenue growth guidance for a range of 36% to 38% or approximately $390 million at midpoint. This compares with our prior estimate provided on August 28 for 31% to 35% year-over-year growth or approximately $379 million at midpoint.

These results are very encouraging, but I'm even more excited about the HAI. There are 3 key factors behind our enthusiasm and our strong commitment to AI. First, the breadth of application demanding AGI technology and our product is expanding. Second, the AI performance requirement for our product road map is expected to continue to rise, driving robust new product cycles. Third, our ASP has been increasing. And in the long run, we continue to see an excellent opportunity to capture more value per design.

I will elaborate on those points. First, AIH is becoming more prevalent, driving an increasing breadth of applications in both enterprise and consumer tier. Our HI business started in enterprise security followed by automotive safety, smart home, kinematics and more recently, the portable video market, which includes action camera, panorama cameras and the body own cameras.

Looking ahead, high-value shipments into the aerial drone market are expected to commence this quarter, representing just the beginning of our realization of the large robotic market opportunity. There is also strong interest from existing and new customers in our infection and road maps, and we are committed to develop this incremental opportunity.

In addition, ADAS and vehicle economy remains large market, I can leverage our technology to a very high degree. Second, we see a large opportunity to execute an age the increasingly complex AI technologies currently implemented at the core of the network or in the data center. The challenge and our opportunity is that the solutions used to network are open, not simple for the edge. Whether a performance requirement is rising in each market low-power consumption, real-time processing, privacy, security, small phone factors, thermal network brand worth efficiency and lower price points are critical.

At Ambarella, we continue to invest heavily in our proprietary [ HAI SoC ] technology and products to support this unique and increasing requirements. For example, our 10-nanometer CV2 family support CNN network and our financial meter embedding our third-generation AI processors in scaling our customer into world complex and generative AI application simultaneously.

Third, we see an excellent opportunity to continue to increase our ASP, the shift from CPU workload to high level of accurated computing or [ ANI ] is well underway. The adoption of increasingly Hong Hank's data center technology for the age is another driver.

Finally, the extension of road map in a endpoints and into the age infrastructure and auto economy is also expected in particular to boast our ASP. For example, our SoC blended ASP in Q3 was up about 20% year-over-year and as our third-generation AI SoCs and over new product become a more material portion of our revenue.

We intently further increase in the value we earn per design wins. I will now describe some of the representative customer engagements that reflect the factors I just described. In the enterprise security market, we are very pleased to share a significant milestone with our customer Sparsh, who became India's first security camera manufacturing to receive HTQC certification for its complete rent. At the heart of the collaboration is our CV28. This gives us a tremendous start to accelerate our adoption in a rapidly growing made in India market.

Ability to span out of Bosch announced their AutoZone 7100 moving PTZ camera with building a ethics. Ultra HD image based on CV72. They have also announced their dining thermal security camera that is fixed on CD22 around their CNN models to detect and classify objects accurately up to 2,000 feet.

Verkada announced their upcoming CV75-based A act Station Pro, which enable secure physical access with a facial recognition powered face unlock alongside traditional badge and the mobile access method. The company also launched a new CR 63E remote security camera and leverage the power efficiency of our CV 75. They also expect the CD72-base multisensor security camera product line, including CH53, CH63 and CY63.

Motorola has developed their additional half smart sensor on our CD25, which is an all-in-one environmental monitoring and the security device that is designed for areas where cameras are restricted to detect events like smoke, fire and audio anomalies. In the robotics smart home market, one of our customer whisker announced the liter [indiscernible] UFO, their first model with facial recognition to support true 9-vision-equiped AI power camera based on CD28.

We are seeing great momentum in our portable video market with [ Arash ] who released 2 models this quarter. The export air at just 165 grams is a new lightest compact AK360 action camera is based on CD5 and is first in the range to support AP30 per second active HDR. Arash also launched the latest version of their body camera GoUltra. Based on CB52, it captures 4Kp60 for ampersecond video and the 50 megapixel photo with improved performance in life environment.

In our automotive safety [indiscernible] business, I would like to share some key customer wins during the quarter. Zika Unitil has developed their in-cabin DVR system CV28 for the 9X [ Husi ] luxury model. [indiscernible] expands their global market presence they have built older driver management system for Ollie Expo models on CV28. Solera, a global leader of vehicle life cycle management announced their new ESAPI powered smart camera in over based on our CV22.

In the erosive, the A5 camera is powered by AI plus humanities with revolutionary approach in fleet analytics that combines AI-based analysis with human oversight to improve safety, efficiency and operations. From this representative customer engagement, I just described, the strength of our current product portfolio is currently represented. With 7 examples on the 10-nanometer CV2 family, and several examples from our financial meter generation. These products all available today, offers customers a wide variety of options ranging from CNN to transformer network processing. 1 to many sensor input support multiple sensing modalities, all at a wide range of price points.

Our new product road map will expand this portfolio further. In addition to our comprehensive and expanding AI SoC portfolio, another important distinguishing characteristic of our portfolio is the ABSI technology we offer to customers at the age -- for example, 5-nanometer-based product represent more than 45% of our total Q3 revenue, with products based on more advanced nodes in development.

In summary, the first 3 quarters of fiscal '26 are steps in the right direction with strong revenue growth, new product execution, profitability and with our cumulative year-to-date free cash flow almost 14.8%. We continue to forecast a large age serviceable available market of $12.9 billion by fiscal year 2031. We recognize the HAM market is still in the early innings of development and to successfully address this large sent, we remain highly committed to our R&D investment that enable us to build upon our existing leadership position.

I hope to see you January 6 at our CES 2026 product and technology briefing, which will give you a key return on our new technologies and products and meet a growth set of our management team.

With that, John will now discuss the Q3 results and the Q4 outlook.

Thanks, Fermi. I'll now review the financial highlights for the third quarter of fiscal year 2026 ending October 31, 2025. I will also provide a financial outlook for our fourth quarter of fiscal year 2020 ending January 31, 2026. I'll be discussing non-GAAP results and ask that you refer to today's press release for a detailed reconciliation of GAAP to non-GAAP results. For non-GAAP reporting, we have eliminated stock-based compensation and acquisition-related expenses, adjusted for the impact of taxes.

For fiscal Q3, revenue was $108.5 million, above the high end of [indiscernible] of $100 million to $108 million, up 13.5% from the prior quarter and up 31.2% year-over-year. Sequentially, automotive revenue increased in the low single digits and IoT increased in the mid-teens. With IoT growth led by the adoption of Edge AI in enterprise security and portable video applications. Non-GAAP gross margin for fiscal Q3 was 60.9%, slightly above the midpoint of our prior guidance range of 60% to 61.5% to product mix.

Non-GAAP operating expense in Q3 was $55.3 million, slightly below the midpoint of our prior guidance range of $54 million to $57 million. Q3 and net interest and other income was $2.1 million. Q3 non-GAAP tax provision was approximately $900,000. We reported a non-GAAP net profit of $11.9 million or $0.27 per diluted share in Q3.

Now I will turn to our balance sheet and cash flow. Fiscal Q3 cash and marketable securities reached $295.3 million, increasing $34.1 million from the prior quarter and $68.8 million from the same quarter a year ago. Increased cash and marketable securities benefited primarily from operating cash flow associated with increased revenue.

Receivables days sales outstanding decreased from 41 days in the prior quarter to 36 days and days of inventory decreased from 85 to 76 days. Operating cash inflow was $34.3 million for the quarter. Capital expenditures for tangible and intangible assets were $2.9 million for the quarter. Free cash flow was $31.4 million.

We had one logistics company representing 10% or more of our revenue. WT Microelectronics, a fulfillment partner in Taiwan, that ships to multiple customers in Asia, came in at 70.2% of revenue for the third quarter. I will now discuss the outlook for the fourth quarter of fiscal year 2026. The breadth of our Edge AI business is expanding. Together with strong unit volume and average selling prices. As a result, in Q4, we forecast revenue in the range of $97 million to $103 million or $100 million at the midpoint.

With a higher percentage of revenue coming from our high-volume customers. Sequentially, due to seasonality, we expect a mid- to high single-digit decline in both our automotive and IoT businesses. We expect fiscal Q4 non-GAAP gross margin to be in the range of 59% to 60.5%. We expect non-GAAP OpEx in the fourth quarter to be in the range of $55 million to $58 million, with the increase compared to Q3, driven primarily by employee-related and CES expenses.

We estimate net interest and other income to be approximately $2 million, our non-GAAP tax expense to be approximately $600,000 and our diluted share count to be approximately 44.5 million shares.

Thank you for joining our call today. And with that, I'll turn the call over to the operator for questions.

[Operator Instructions]. Our first question will come from the line of Tore Svanberg from Stifel.

Congrats on a record quarter. My first question, when we think about that, let's call it, 36% to 38% growth for fiscal '26 -- how much of that is unit versus ASP Because, obviously, CV now is becoming a pretty high percentage. But even within CEV,obviously, you have ASP increases. So just trying to understand how much of the growth has been driven by ASP versus units?

Right. So I think both of them contribute to our growth. I would say I don't have the exact number but if I guess it's probably half to half. I think our unit growth definitely continued to contribute from the CV side, but ASP growth is also significant like we talked about in the script. So I think that both of them contribute to our end results.

That's very helpful. And as my follow-up, you talked about the portable video market. Could you just add some more color there? I mean, it sounds like you have some new design wins. These are obviously AI-based drones. But -- just -- I know you've been in that market for a while, and obviously, that market sort of faded and now it seems to be coming back. So how should we just think about that market driving growth for Evrell going forward?.

Right. So we talk about portable there are multiple product line in there, I want to be a little bit more specific on that. In fact, we talked about action sports camera like you said, we have been cared for many years. And the new new category is panoroma camera that Arash is famous for. And also, we talk about drones that is also part of the portable video. But in addition to that, our wearable camera web camera, video conferencing product, all of them are part of the portable device because that's where our customer is focusing on.

So overall, that's an area that providing a big portion of our growth this year. And we believe that this market is going to continue to grow. And in fact, to say that I'm surprised -- a little surprised by the size of the market that growing over the year, but definitely momentum is there. Our job is trying to not only secure our market share, but hopefully, that we can grow some market share in the future.

Tore, it's Louis. Technically, we call it portable video and other. So there's a lot of things in there, as Fermi said.

One moment for our next question. Our next question will come from the line of Ross Seymore from Deutsche Bank.

A couple of questions and congrats to Les. I guess, first, you talked about the breadth of your business, especially in the Edge IoT side of things or Edge AI, IoT, whatever you guys are calling it now. Can you just talk about a little bit about the consumer versus kind of the enterprise side? I guess where I'm going is the portable side is great. but we've seen volatility around any sort of consumer applications in years past and cycles past. And I just wondered how you're managing that in this instance.

Ross, it's Louis. The split is roughly 50-50, 50% kind of enterprise CapEx-driven and 50% consumer. And then within that 50% that's consumer, you've got some kind of consumer durable things like, say, smart home cameras that get replaced every 5 or 6 years, but then you also have consumer discretionary, which I think is some of the more volatile things you were referring to. So it's pretty evenly split at the highest level between the CapEx-driven markets in the consumer, but different types of consumer spending.

And I guess one for John. How are we thinking about gross margin as we look into next year, just conceptually what the pluses and minuses would be I know you have the long-term target of the 59% to 62%. You're a little closer to the lower end of that in your fourth quarter guide. But just running through any of the puts and takes would be helpful.

Yes, Ross, thanks. So as you said, our long-term model is 59 to 62. And as we said in our Q4 guide, the composition of gross margin really depends on the contribution of like our [indiscernible] high customers. So whatever the gross margin is from quarter-to-quarter, that's -- at least in the near term, that's a primary driver.

Our next question will come from the line of Joe Moore from Morgan Stanley.

I also wanted to ask about that gross margin target. And I guess, just as you've kind of refocused the business around a lot of exciting opportunities, is there any chance to really fully participate in some of the consumer markets that you might accept lower gross margin in exchange for growth?

And then I guess you've talked a lot on this call about average selling price. Sort of what that's -- what's driving that focus is ASP versus kind of gross profit dollar per device, things like that?

Yes. So as far as the ASP goes, that is primarily a function of the technology and features that come with these more advanced technology tape-outs that we're doing and products that we're developing on our road map.

As far as the gross margin goes, like I said, 59 to 62. I think as far as consumer -- on a case-by-case basis, depending on the volume that we see, the opportunities that we see, we're not opposed to gross margins that are maybe not strictly within the 59% to 62% range. But the goal at the corporate level is to, over the long term, stay in that range.

Great. And then I guess, there's a lot of enthusiasm for drones, which is a market that you've kind of been in, in the past. Can you talk about what are the new elements of that market that probably might drive you to a higher content over time. And what -- is it sort of you think about delivery drones and industrial drones and things like that. Is that a pretty big category for you down the road?

Right. So first of all, we were big in the past, as you said. But we will we stopped in that market because of geopolitical [indiscernible] because of our technology solutions. And this time, we came back because we continue to believe a few things.

First of all, there was a dominant player, but I think that in the U.S. the market is wide open at this point for everybody to fight in that capacity. So with our video technology, particularly our panorama camera that we help our customers to build is well suited for this space. So for -- first of all, the driver for us is continue to provide the best video solution in the drone market. But more importantly, I think moving forward, is all the drone going to be autonomous in the future.

We kind of say [indiscernible] strong level true and Level 3 and Level 4 drones coming and probably going to drive faster than autonomous driving car. And we believe that in order to have a Level 3 drone that it would require a really powerful chip in addition to the video processing, and that's really played to our strengths also that our investment in autonomous driving directly applied to here.

So is from a technology point of view, the video processing plus AI is the key driver. But as you said, today, the biggest market opportunity for us is consumer capture, but moving forward, we start seeing opportunities on the commercial side, which is going to continue to drive the growth. So we are excited that, first of all, we have real technology that we think is differentiated in this market. But more importantly, the market -- the service market for us is growing fast, so that the 2 reasons that we feel excited about this market.

Our next question will come from the line of Christopher Rolland from Susquehanna.

Congrats on the results. I guess my first question is around an update perhaps for the infrastructure opportunity and the N1 655

Yes. So first of all, we announced our first design wins last quarter. And after that, we continue to see very strong design win activity and interest from different type of customers. In fact, we do -- in the last -- in the last few months, we see customers who want using video-centric products and also customers want to use for non-video centric products. So we are using a wide range of opportunities. So -- and we are also continuing to see our chance to not only building up but also win new line in the next -- in the near future.

So we are totally committed to this market with N1655 and the new road map that we will talk about in the future.

And perhaps if there are any updates on 2 other opportunities I guess the first would be the home security market with AI feature integration? And then the second would be any kind of design activity? I know it's further out, but around humanoid robots, I think that would be interesting as well.

Right. So first of all, for the home security, I think we do have design wins with our CB 75 that we haven't announced yet, but definitely is in design. However, I think this is a market that is price sensitive. So I think the progress or the movement towards this AI type of home security camera for -- based on the camera solution, not the cloud solution, we are really focused on just HAI for this market.

So we that, I think that market is not developing as fast as we expect, but we do have design wins we hope we can talk about sometime next year. From the human rolling, I think this is a long-term market that we definitely want to participate. However, I think it will take time to get to a human rolling. I think there's multiple steps for robotic from today's situation to the human [indiscernible]. And I think, like I said, even done if you treat for robotic application, there's a Level 2 to Level 5.

I think human roll is like Live of drones for your different applications. But there's intermediates about stat we need to go through, and we definitely have design wins and also design activities in those steps that will lead us into the humanoid. I just want to be more specific. We're operating 2 type of solutions to the robotic today. One is for people only interest with a video technology. So you want to have a really awful eye that not only can see the object, but also can do quick optic detection based on CNN network.

We have that kind of solution based on our CV2 family or CB72CSemotrax solution. So that's 1 product line we are providing.

The other product line we're providing to the robot is really a brand. right? So our online 655 type of solution can be a central processor for the any type of a robotic out there. So I think we're upgrading solution will take time to develop 3 or central domain controller like to driving car, that kind of solution will be required to do a human solution in the future.

Our next question will come from line of Suji Desilva from ROTH Capital.

John, and best of luck with the next step transition here. So maybe in the Edge AI market, looking ahead to calendar 26 perhaps, which of the 2 or 3 segments would you describe as the highest kind of growth opportunity for you at drones or other areas? Any color there would be helpful.

I think some time [indiscernible] that is going to be a growth area for us. And I also believe that even for the endpoint, we continue to see multiple opportunity coming up with different type of products. For example, wearable camera, we talk about this for many years. But right now, we are excited because wearable camera is not only for placement anymore.

We start seeing that going to totally different conversion -- so that's just another example that the technology becomes ready, low power and also AI on the camera, all of that enables a new application for wearable camera. That's another really high-growth area that we're seeing, and it's not only what an follow our customers, you will see that our customers saying similar things.

So those endpoints definitely work area for us. But I also want to bring your attention to the aged infrastructure we'll talk about last quarter. I think although that's not immediately you're going to see high revenue growth. But I think long term, that will be a very important market for us. and we'll definitely cover our thoughts on each infrastructure at CES and give you more insight to our plan.

Great. Then Fermi, one specific question on drones. Do you have any visibility in your pipeline beyond consumer commercial perhaps in any government programs? Or is that going to be a separate part of the market, handling that versus you guys?

Old customer, in fact, it's not really it's all cost cost -- all the customer has a desire to serve multiple different mice segments. But most of the customers are focusing on consumer commercial -- and I don't think that our [indiscernible] usage is all focused for most of our customers yet.

Our next question comes from the line of Martin Yang from Opco.

First question on IoT, especially with growing customers like Arash, could you maybe comment on those customers' growth and this relative contribution to overall ASP and margins?

Right. So first of all, Ara is -- I think our largest customer in our top 10 list and their revenue roughly doubled from last year to this year. but they are using multiple chips and selling into multiple ODMs. It's hard for us to track exactly the revenue contribution, but we have no doubt they are the largest customer right now.

Another question on drones. So when you're referring to next year's product, are those drones using your image processing capabilities? Or do you expect them to deploy AI functions that relates to autonomous flying capabilities?

Both. I think that, like I said, there are 2 types of solutions we're offering some of them using just a video plus AI to apply CN type network for similar AI functions, but they will be definitely customers using our AI for flying to avoiding objects for to determine the buying [indiscernible] most of them.

Our next question will come from the line of Quinn Bolton from Needham & Company.

I know the folks in the business has sort of shifted to AI in the future. infrastructure. But in the past, you gave us sort of an automotive funnel. You haven't provided that. So just wondering how should we be thinking about how are you guys approaching the automotive market do you still see opportunities in Level 2+? Or are you kind of deemphasizing some of the automotive applications?

All right. Thank you for that question because we did not commit on that market. In fact, we continue to focus the market. We are engaged in multiple OEM Tier 1s at this point for auto-driving Level 2, Level 2+, some even Level 3.

So from a engineering activity and business activity point of view, we are all in on this market before. Definitely, that -- from the final discussion point of view, like I said last quarter, we will provide final discussion in the next quarter release. But the 1 notification that we'll do, we will start using a probability-weighted matrix. We are trying to go to give you just direct opportunity we're looking at. So that will be the 1 change we're going to offer, but we will definitely provide more guidance on this -- how we look at this market.

Got it. And then I guess, for John, just you mentioned that it sounds like the mix towards high-volume customers pushing the gross margin down to the lower half of your long-term range. Can you give us just beyond the January quarter? Do you think that mix continues to be pretty heavy with higher-volume customers? Or do you see this as sort of a temporary shift just for the January quarter and it normalizes beyond that?

Thanks, Quinn. Yes. At this point, we don't want to give a guide with regard beyond Q4. But I think that commentary with regard to is one to be relevant going forward, the ratio of high-volume customers to the total revenue for the quarter.

Sorry, John, you cut out there a little bit, did you say that the mix would stay pretty similar beyond January?

No. What I tried to say was that we don't want to make any guide beyond Q4, but that the commentary about Q4 with high -- the ratio of high-volume customers to the total revenue that dynamic will continue to be a factor going forward. So to the extent that the high-volume folks are a higher percentage of the revenue, that will have its impact.

[Operator Instructions]. We have a follow-up question for Tore Svanberg from Stifel.

John, just a follow-up for you. So this year, you guys demonstrated some pretty good operating leverage I'm just thinking as we look at fiscal '27 and OpEx growth, obviously, you're not giving a growth target per se, but we should assume that OpEx would grow at a slower pace than revenue growth for fiscal '27?

Thanks, Tore. Yes. We're not giving a guide at this point. But I think what we have said in the past, kind of as you articulated, is that long term, we expect to to create operating leverage by having revenue and obviously, gross profit outpaced the increase in OpEx on a non-GAAP basis.

Our next question will come from the line of Kevin Cassidy from Rosenblatt.

Congratulations to Les for a legendary career. Again, I'm interested in that. But I want to know how much of your software and development that you've been able to work on with the automobile for L2 to L4, -- can you apply -- is it a relatively easy market for you to transition into or are there other software or other issues that would happen in robotics that is in automotive.

I think, Kevin, you pointed out is really a great direction because like I continue to say, time driving is just a special kind of robots. And so is a drone. And in fact, if you look at the details of functions inside ton-driving car, Level 3 drone and also robots, ID is really a bunch of sensor fusion and you make a decision on the environment and you decide a pony you control either car to or some more robust moving around performance actions.

From that point of view, a lot of hardware have a commonality and in fact, a lot of silver on the sensor fusion side with the perception, there's a huge release on the older robotic applications. So in fact, we definitely believe that that a lot of our investment, both on hardware and software side for auto driving will directly apply to all the future long and other robotic application that we're talking about.

Our next question will come from the line of Ross Seymore from Deutsche Bank.

A couple of follow-ups. On the consumer percentage being about half of your IoT business, what was that mix last fiscal year, a year ago?

I don't have that figure for you, but I would say the dominant part of our mix was enterprise. CapEx-driven markets.

Got it. And I guess the follow-up to that, if the consumer business does sound like it has increased, does that change seasonality of your company. I know kind of the first and the fourth quarters tend to be relatively speaking, the weakest sequentials, and then the mid-2 quarters are the largest. Does that change at all either directionally or kind of magnitude just because consumer is a bigger portion than it used to be?

Yes, that's a very good point. And the answer is yes. And I would look at -- the next question is what's normal. And really, the last 3, 4 or 5 years hasn't been very normal. So I'd look at the last 10 years because those first 5 years in the 10 years, did have more consumer like you're asking about. So I'd look at the averages over the 10-year period rather than just the last 2 or 3 years, which really weren't normal.

And then maybe 1 last follow-up. How do we think about taxes, either dollars or percentages next year and the year after. I know it kind of goes between the dollars and percentages and the former might be more applicable. But just an idea of how we should think about that.

Yes. Thanks, Ross. So we tend to think about it in a -- from a dollars perspective as opposed to a rate based on the way the company is structured and where the profits are located in various jurisdictions internationally. So I would expect well, the dollars will increase, but it won't be -- they'll increase with revenue but it won't be a significant change to the story.

I think on a full year basis, if you look at the rate on a non-GAAP basis, that will give you some indication to be able to model going forward, I would say.

That's all the time we have for question-and-answer session. I would now like to turn it back over to Dr. Fermi Wang for any closing remarks.

Thank you, and thank you all for joining our call today, and I hope to see you -- some of you predicted on our January event at CES. Thank you.

Thank you for your participation in today's conference. This does conclude the program. You may now disconnect. Everyone, have a great day.