SES AI Aktie

Thank you. Hello, everyone. Thank you for joining us and welcome to the SES AI second quarter earnings release and call. After today's prepared remarks, we will host a question and answer session. If you would like to ask a question, please press star one to raise your hand. To withdraw your question, press star one again. I will now hand the conference over to Kyle Pilkington, Chief Legal Officer. Kyle, please go ahead.

everyone and welcome to our conference call covering our second quarter 2026 results. Joining me today are Chi Chau-Hoo, founder and chief executive officer, and Ray Liu, chief financial officer. We issued our shareholder letter just after 4 p.m. today, which provides a business update as well as our financial results. find a press release with a link to our shareholder letter and today's conference call webcast in the investor relations section of our website at scs.ai. Before we get started, this is a reminder that the discussion today may contain forward-looking information or forward-looking statements within the meaning of applicable securities legislation. These statements are based on our predictions and expectations as of today. Such statements involve certain risks, assumptions, and uncertainties, which may cause our actual or future results in performance. materially different from those expressed or implied in these statements. The risks and uncertainties that could cause our results to differ materially from our current expectations include, but are not limited to, those detailed in our latest earnings release and in our SEC filings.

On this call, we will discuss non-GAAP financial measures as a supplement to our GAAP results. These non-GAAP measures are not prepared in accordance with generally accepted accounting principles but are intended to illustrate alternative measures of the company's operating performance that may be useful. non-GAAP measures should not be considered an isolation or as a substitute for any GAAP measure, and our definitions may differ from those used by other companies reporting similarly titled measures. Reconciliations of the non-GAAP financial measures to the most directly comparable GAAP measures can be found in our latest earnings release. With that, I'll pass it over to Chi-Chu.

Thank you, Kyle. SCS is solving two of the most difficult challenges in energy storage, accelerating product development using AI for materials and building a robust supply chain to manufacture these products. We originally focused on EV and pivoted more than a year ago to ESS and drones applications. In Q2, we began seeing significant commercial milestones, and we are very excited about the path we're on. Our Q2 revenue grew by more than 40% compared to Q2 last year. And our gross margin improved from 18% to more than 22% due to our differentiated technology and robust supply chain. And we are reaffirming our 2026 revenue guideline of $30 to $35 million. On accelerating product development using AFO materials, we release Molecule Universe MU3.0, our first agentic workflow platform that works for sodium chemistry as well as lithium chemistry and can be integrated with autonomous labs deployed fully secured and on-premise.

We shipped our first search in a box order to one of the world's largest battery manufacturers, and some of the materials discovered by Molecular Universe have completed testing and entered pilot commercial deployment. On building a robust supply chain to manufacture these products, for ESS, this is our largest revenue generating unit. We're making great progress, especially in the US market. We were selected by Solark as a certified battery partner, and we brought on Paul Deamer, of FlexPower to our board to help guide our ESS strategy. And we continue to hire a stellar team with background in leading AI data center total solution providers to execute and deliver our exciting ESS growth. For drones and unmanned systems, we are recruiting a team with a proven track record of selling to defense and commercial drones. We expect to start producing 1 million NDAA compliant cells per year in about one month at our Korea plant.

And based on the strong customer demand we are seeing, at taking orders well into 2028. I'll dive into each topic separately. On ESS, while most competitors sell either pure hardware that don't have intelligent software or pure software that are not trained on real world data. Our edge box enabled ESS systems are trained on the specific cells that we use in our hardware systems, allowing for one to one matching accurate state of health and safety management. This prediction accuracy not only helps preventing fire and other incidents. This is tremendous saving for our customers across Resolute. commercial, industrial, and data centers. One of the leading US-based FCC authorized inverter producers, SOLARC, certified our subsidiary, UZ's low voltage residential batteries for their hybrid inverter systems.

We believe this certification with SOLARC will greatly accelerate the growth of UZ's revenue in the US. especially given the recent FCC restrictions around foreign produced inverters and other electronics. We were also honored to bring on Paul Deamer to our board of directors. Paul served as the CTO of FlexPower, where he ran critical and embedded power group that was responsible for delivering power solutions to data centers and other industrial systems. Paul also ran new EV product architecture at Bort Warner. Paul's transition from EV to data centers is very similar to that of SES. On drones and unmanned systems, we expect to complete the scale up of our Korea-based NDAA compliant cell production from 200,000 cells a year to 1 million cells a year in about one month. And we expect to start producing at 1 million cells a year full speed starting this Q4.

We have already. hosted many of the largest American and allied drone makers for line audits, with many more in the queue later this year. We expect revenue contributions from NDAA-compliant sales produced in our career line to start in a meaningful way in Q4 this year and really start to take off first half next year. Even at 1 million NDAA compliant pouch sales, which we believe is one of the largest NDAA compliant pouch manufacturing capacities in the world, and combined with our best in class energy density and performance, we are looking at securing orders well into 2028. And we're also seeking additional FDA-compliant manufacturing capacities for both pouch and cylindrical cells to address the strong demand for these products. These cells will be for drones, but also broader unmanned and mobility applications. We recently announced a framework agreement with Daroni, where we'll be responsible for designing and developing the complete battery pack for their H1X Evito. It's a really cool two-seater. With Molecular Universe, we released MU 3.0.

This is the most powerful and complete end-to-end workflow automation in energy storage. We sold a search-in-the-box module to one of the largest battery makers in the world, and we are trialing full MU 3.0 workflow integrated with autonomous labs. with many more. We do have competitors for AI for materials, but none offers solutions as complete, accurate, and most importantly, secure as ours. And many of our customers switched to EMU after trying our competitors offerings. Many of our competitors try to offer building blocks in a cloud-based toolkit, but product development is more than a toolkit. And very few enterprise customers would allow their proprietary data to leave their premises or be used to train external models. It requires a fully secured on-premise integration of domain expertise, experimental data, and computation chemistry simulation full stack.

Some of the materials discovered by Molecular Universe have completed testing and entered revenue generating early stage commercial pilot development. We expect to release MU 4.0 later this year. It will feature ability to generate new molecules based on desired properties, and it will be integrated with autonomous labs So users can generate or discover new molecules, synthesize them, validate them in full devices, and provide actual experimental data back to train their own foundation models, all fully secured on-premise. This flywheel connects similar technologies simulation with experimental validation can organize and generate high quality data and train models fully secured and on-premise. And without humans in the loop, it can run much faster than humans ever can. I do think a lot of investors are underestimating molecular universe, especially purely through the lens of near-term monetization. But I believe in the next three to five years, molecular universe will power majority of product development, definitely in energy storage and expanding to complex fluids and eventually other material applications.

The SCS team is solving two of the most difficult challenges in energy storage, accelerating product development using AI for materials and building a robust supply chain to manufacture these products. We have a healthy cash runway, highly differentiated capability across products and manufacturing, and one of the most dedicated teams. I'm incredibly proud to work with our team on these critical challenges, even when the market may perhaps underestimate us. And I'd like to thank the team for their hard work.

Now here's Ray for the financial updates. Thank you, Chi-Chao. I'll walk through our second quarter 2026 financial results. Second quarter revenue was $5.1 million compared to $6.7 million in the first quarter of 2026 and $3.5 million in the second quarter of 2025. Notably, this quarter validated our commercial momentum. For the first time, we saw revenue contribution across all product lines. ESS, drone battery cells, materials, and molecular universe. Our gap gross margin was 22.6% in the quarter, an improvement from 18.1% in the first quarter.

The improvement was particularly driven by the ESS business, where we saw a higher mix of international sales and continued pricing discipline. Turning to operating expenses, our GAAP operating expenses for the second quarter were $20.3 million compared to $19.1 million in the first quarter. The slight sequential increase was primarily due to a bad debt provision related to a legacy EV service contract. Year over year, however, operating expenses were down 26% and we remain confident in our ability to sustain the expense reduction of more than 20% year over year. Our gap net loss for the second quarter was 17.8 million, or $0.05 loss per share, compared to a gap net loss of 12.1 million, or $0.04 loss per share in the first quarter. I want to remind everyone that our GAAP net loss can be impacted by non-cash mark-to-market movement in the fair value of our sponsor earn-out liabilities. which are required to be measured each reporting period under GAAP. In the first quarter, we recorded a $4.2 million non-cash gain related to these liabilities.

That impact was significant in the second quarter. Excluding change in sponsor earn out liabilities, stock based compensation, depreciation and amortization, and including interest income. Our non-GAAP net loss for the second quarter was $13.1 million, or $0.04 loss per share, compared to a non-GAAP net loss of $11.1 million, or $0.03 loss per share in the first quarter. The sequential widening in non-GAAP net loss was primarily due to lower revenue in the second quarter and the bad debt provision that I mentioned earlier. Looking ahead, we expect our net loss to narrow in the second half of the year, driven by a pickup in revenue and continued reductions in operating expenses as our cost reduction program takes full effect. Adjusted EBITDA for the second quarter was a loss of $14.6 million compared to a loss of $12.8 million in the first quarter. A detailed reconciliation of GAAP net loss to adjusted EBITDA and non-GAAP net loss is included in the financial tables at the end of the shareholder letter. turning to capital allocation we ended the second quarter with cash cash equivalents and short-term investments of approximately 163 million Our CapEx light business model remains a core financial discipline and We're confident our current liquidity provides a runway to fund operations and execute on our 2026 growth initiatives.

We're actively looking for inorganic growth opportunities, including M&A that compliment our strategy. while maintaining financial discipline. We believe the second quarter demonstrates continued execution against the plan we laid out. broadening revenue contribution across all our products, continued gross margin improvement, and disciplined cost management. We are reaffirming our full year 2026 revenue guidance of 30 to 35 million. As we look to the second half of the year, our priorities are clear. Continue to scale energy storage systems and edge box distribution. Convert our drone qualification pipeline into commercial orders as the Chengdu ramp completes and close our supply agreement for materials discovered through molecular universe. With that, I will hand over to the operator.

We will now begin the question and answer session. To ask a question, please press star one to raise your hand. To withdraw your question, please press star one again. We ask that you do pick up your handset when asking a question to allow for optimum sound quality. If you are muted locally, please remember to unmute your device. Please stand by now while we compile the Q&A roster. Your first question comes from the line of Winnie Dong with Deutsche Bank.

Your line is open. Please go ahead.

Hi, thanks so much. I wanted to touch on the back half of this year, if you can remind a mix of revenue that's going to be driven by across the different business lines. And then you can also talk about the backlog at the energy sentiment. It seems like you've got some important certifications that might be good for the US side. what kind of demand management might we be looking at for 2027? And then I have a follow up. Thanks.

I'll be in terms of the split. First half is. it's basically majority ESS, almost more than 70% ESS. But in the second half, we do expect drones and materials to pick up. So still more than half from ESS. And then in terms of that certification with SOLARC, so basically if you're not certified, then you're competing with like 20 other companies. But once you are certified and there's only like less than five, there's like Enphase, there's Tesla, there's SOLARC, then really you're competing with like three. So later this year, we expect the hockey stick to start to pick up and then a lot more next year.

I think the exact number, we don't have the exact number yet, but we're quite excited to be certified by SOAR. Because now we go from competing with 20 other players to now competing with just one.

like Enphase and Tesla. OK, thank you. That's helpful. And then I was wondering, maybe just like on the capital allocation side, it seems that you're just sort of like sticking with that capex like approach. You do have, you know, 160 plus million cash runway. What kind of CAPEX might you be looking to do to spend it on? And you also alluded to M&A. What are some of the pipelines that you're looking at? What can we anticipate? I think if I look back to the capacity you have for your drone sales out of Korea, it seems like there is demand and order into 2028. So would you need capacity expansion there and the use of CAPEX for that? Thanks.

We definitely do. I mean, if we just had 1 million cells out of Korea, then literally we're going to supply like at most three customers, three drone customers that need NDA. Like most. So without additional capacity and based on this pipeline, we are looking at 2028. Now if we can double the capacity in Korea to 2 million or triple that or find additional NDA compliant capacity in Southeast Asia, so if we quadruple or even more that capacity, and all that pipeline that we were going to supply in 2028, now we can supply in 2027. So in terms of line, Not so much this year because most of that has already been spent. And then again, most of the career cap-hacks, we built that back in, you know, in the early during the JDA with GM. So now it's not rebuilding new cat packs, it is more modifying that. So not as much this year.

And then later this year or next year, if we're to invest in additional cat packs, either through direct investment or through investing in or companies that own those cat packs. I think that's TBD, but again, for drones, we're not talking about gigawatt hours scale. We're talking about, for example, 20 megawatt hours, 50 megawatt hours, like way smaller than gigawatt hours. So I think the space is a lot more efficient than the EV capex.

Thank you. And then what about the M&A portion? Any pipelines that you might be looking at? Anything that we should sort of anticipate?.

We are exploring. I think there are some companies that have interesting manufacturing capacities. in Korea in the drone space, in the ESS space. Companies that make pop shells for drones or prismatic LFP, we're very interested in working with them. both cell companies as well as PAC companies. We are still evaluating some of those capacities.

Yes, we need internally with we set up some just want to add internal guardrails. I'm doing MNA. So anything we want to do is going to fit into our strategy. And also it's going to be business and revenue accretion to our business. So we're just looking to, as the Chi Chau was mentioning, expand with the cash we have on hand. Great. Thank you so much. I'll pass along.

Your next question comes from the line of Dave Storms with Stonegate. Your line is open. Please go ahead.

Hello, and thank you for taking my questions. Just wanted to maybe start with your comments around the expected gross margin improvement. Could you help us understand maybe what might be driving this? Is this going to be continued on pricing? Is this, you know, volume increases, mixes, you're moving into more drones in the back half? Just any further color here would be great.

Yes, I can add some color. I think the improvement, as I mentioned in the announcement, is primarily from our ESS business that's mostly driven by kind of increased international sales. We typically, especially in the North America, particularly see a higher margin on the sales in North America. Additionally, we have maintained a price discipline on the ESS business. So that's contributed. I think a back half of the year Like I said, as the drone revenue picks up, we'll see margin increase because a drone sale typically carries a higher margin compared to the ESS business. So we'll see margin improvements in the second half.

Yes, I think one of the key drivers to the gross margin is, especially in ESS, is the price of the sales. AGO, SALES WERE REALLY EXPENSIVE AND WE DID NOT HAVE EDGE BOX. SO WE WERE REQUIRED TO BUY SALES PRODUCED BY ONE VENDOR OUT OF ONE OF THEIR LINES. SO WE HAVE VERY LITTLE PRICING And since we have edge box, now we're able to source sales from multiple vendors and multiple lines because we're able to monitor and then balance between the sales. So that has reduced the price of sales.

That's great commentary. I really appreciate that. Maybe, Sushant, if you want to, you know, you mentioned that you are exploring beyond drones, you know, the broader unmanned and mobility applications, you called out a new partnership there. I think you could spend a little bit time talking about maybe some of the technical challenges between drones versus the broader unmanned ecosystem, if there are any. Maybe how that market looks in terms of margins or demand or TAM compared to your current market.

anything else there would be very helpful. So a lot of it is just timing and then go to market investment and also timing. And then we focus on drones because that market we have seen that uh it's consolidating around around one up to three uh major form factors and then so the cells we we build um we definitely first and and foremost selfie to drones. Now there are other applications that recently have surfaced marine applications, boats, submarines, applications that also care about weight, NDA compliance, as well as some of the cargo planes, manned eVTOLs as well as unmanned eVTOLs. So all of the form factors that we have matured for drone applications, we're also selling those to the other adjacent markets.

That's great. Thank you very much. Your next call comes from the line of Craig Irwin with Roth Capital. Your line is open. Please go ahead.

Good evening, and thank you for taking my questions. So I wanted to ask about molecular universe. So, Shashal, you're clearly excited about this as a longer-term opportunity and the potential both revenue and IP out of the library that you've developed. Can you maybe talk about customer engagement? How actively are you marketing this to new customers at the moment? Do you have strong leads? for additional new customers at the moment? If you could maybe give us a little color on the breadth of the customer interest. And how long do you think it will take as far as customers that are already, you know, looking for leads out of your database? to make those commercial and have those revenue generating opportunities.

So I would say we have about three customers, three battery companies that we are deeply engaged with. And then when I say deeply, I mean there's at least three to five teams within each company that are evaluating different parts of the molecular universe. So the entry level basically ones that by modules, for example, you mentioned search in a box, some people by predicting a box, some people by design in a box, some by ask in a box, some by formulate in a box. So that's the entry level. companies buy these modules. Second level companies will buy the entire workflow, ask, search, predict, design, manufacture, basically this entire workflow and also with Starseeker. So it's a agentic workflow. And then the third tier companies that buy what we call called MLABS. So MLABS is the integration of the entire agentic starseeker with autonomous labs.

So we will actually combine the starseeker with corresponding autonomous labs. So for example, AscensSearch is integrated with a new molecule synthesis, ALAB. Formulae is integrated with Electrolyte ALAB. Design and predict are integrated with Electrolyte Optimization ALAB. Also part of predict is integrated with Cell Testing ALAB. Before we had pure software platforms installed on-premise and instead of having this pure software in a box delivered on-premise, now the software is delivered together with a lab. So a customer could just provide us 1,000 to 1,500 square meters of space, and then we will lay out all the other a lapse and the corresponding software.

So so the the on the customer side, Instead of needing 30 people, 50 people to run this lab, now you need just one PM, one project manager with this entire software and this entire A-Lab together, M-Lab, and then you can have a complete battery So there are three major customers that we are outlining I would say at the third tier discussion. So hopefully we'll announce some revenues at the N lab level. And then beyond these three, there are also other battery companies that we are in the module level and the workflow level.

Okay, excellent. I understood. My second question is about the cells coming online in Korea. So you're going from 200,000 cells a year in capacity to a million cells a year in capacity. And you know, when I look at your current capacity of 200,000 cells, you know, that's quite small versus the needs of a lot of the defense suppliers, the drone producers and other companies that use lithium ion batteries in military equipment. but million cells, million cells a year sort of gets you in the game. Have you received any sort of soft commitments from these NDAA-constrained customers, the defense market customers, as far as probable orders or potential orders on that million sales a year, what's your confidence level that you'll see the offtake there? And actually, I guess another question is, is it possible we see revenue from the incremental capacity at the end of this current quarter within the month of September?.

Yes, so total our pipeline for drones for just NDA, not counting the non-NDA, we're talking about 50 plus, just all NDA. And then the large customers, and we define large customers, are the ones that need a pipeline. sell 200,000 to 300,000 NDAA compliant sales a year. So we have maybe five large customers. So just THE TOP FIVE LARGE CUSTOMERS, WE'VE ALREADY EXCEEDED 1 MILLION. WE'RE AT LIKE 1.5 MILLION, RIGHT? AND THEN 1 MILLION IS THE CAPACITY, IF YOU TAKE INTO ACCOUNT SOME QUALITIES, SOME DOWNTIES, THIS AND THAT, PROBABLY WE DELIVER 700, 800 K. SO, SO, Yes, we are way under capacity purely in terms of NDA compliance. Now, we're doing two things. One is we are evaluating other applications. capacities in Korea, either through our own investment or through country manufacturing, so that we are hoping to double, maybe triple the NDA compliant capacity in Korea.

And we're also looking at additional capacities in Southeast Asia. This is one. the customers do have a range, a spectrum of NDA compliance? We have ones that want 100% NDA compliance, means cathode, anode, electrolyte separator, pouch, and entire assembly. Everything must be done in Korea. completely NDA compliant. We also have customers that are okay with 40% NDA compliant, 50% NDA compliant. In some cases, Korea's own Article 1, Article 2 compliance where you can have capital from Korea. from Korea, but then sell simply down in China just for one year. So, so, um, we reserve the 1 million capacity in Korea purely for for those that want 100% NDA compliant. And then we do have additional capacity for those that want maybe 40%, 50%, 60% NDA compliant.

So we do have alternative capacities for those that don't want 100% NDA compliance, but some portion of NDA compliance as well as we're looking for additional capacities in Korea and Southeast Asia.

Thank you for that. And my last question is, you know, you're increasing capacity or increasing your capacity commitment there five-fold. I do realize that you're using contract manufacturing, so probably you don't see as big a potential improvement in cost. Is there an improvement in cost that you can expect on this capacity increase? And the five customers that are very large in the drone market that you mentioned, have they already started sampling cells, given that you're going to be using nearly identical production, identical cell recipe, to produce a commercial product?.

products? So I just want to confirm that 1,100,000 NDA compliant cells out of Korea. Those are entirely our own production. And then we have several other million less than 100% NDA compliant. Those are through contract manufacturers. And then in terms of where we are with the customers, they've done sample testing, multiple rounds. They've done a line audit, one to two rounds. And then they're waiting for that 1 million sale capacity you to be operational in September and then visit again for additional quality audits. And in terms of pricing, yes, going from 200K to 1 million would significantly reduce the price.

Excellent. Well, thank you for that, and thank you for the clarification. Congratulations on your progress.

Well, thank you. Your next question comes from the line of Mark Shooter with William Blair. Your line is open. Please go ahead.

Hey, Chico. Chico, am I coming through?.

Yes, we're here. Great, thank you. On the ESS business.

I'm interested in if you can update us and give us a little bit of color on exactly what niche of this this segment you're looking to play in ideally. And can you update us on if you are still trying to sell product that is the full solution including the battery and the DMS or Edgebox or is there an opportunity to just sell the Edgebox to potentially a utility or an integrator?.

Yes, so I would divide them into three categories. One is just residential and then there's new certification. So in residential, most companies buy batteries and inverters in one pack, right? Like people don't really just buy a battery or just buy an inverter. You buy those types to connect it together. So recently when some of those foreign producing verges got blocked, that did also impact their co-op sale, co-marketing relationship for the battery suppliers because those two are sold a lot of times together. And then this partnership with Solark is one of the three along with Enphase and Tesla that have U.S. produced inverters. So this will help the residential And then yes, we do include the edge box there.

And then second is more larger scale data centers. This is quite new for us. This is not an area that UZ was in. UZ was familiar with a residential, not so much in AIDC. So here, we are trying to provide a total solution. So the entire UPS, sometimes it's the best, together with the edge blocks. This is also why we brought on Paul to help us navigate the go-to-market strategy here.

And then third one, some of the battery inverter distributors are open to us installing the Edgebox. and we are testing with them. And then that may also pick up as a revenue source.

Thank you. I'm talking about that second piece of the hyperscalers. Is there anything available or that you're ready to update us on that and give us a little bit more detail on the level of engagement?.

Yes, I think we can discuss and share more in Q3 or Q4. I think now a lot of those are very early.

Okay, understood. Thought I would try. Also, in the shareholder letter, you mentioned that most competitors just sell hardware without software. The competitors I'm thinking of have very complex and sophisticated software programs attached. So who do you consider your key competitors in this space?.

So I think the, yes, I think some of the competitors that you're thinking of, they do, and the large ones, they do. But I think for the majority, they either use like a third party software or they use an internal software that's not specifically trained on those cells. So for example, say the PAC system is using a 314 MPower LP prismatic produce Q1 this year out of a line from the US say CATL or EVE. So our edge box will be trained on that batch of cells, like literally the exact batch of cells that go inside the pack. And then if we switch to even a different vendor that makes the same 314 MPower LP Prismatic, we will retrain the software. So, the software gets retrained on the specific cell, the specific vendor, the specific chemistry that we actually put inside the pack. Okay, great. Thank you. I appreciate the call.

There appear to be no further questions at this time. This concludes today's call. Thank you for attending. You may now disconnect.

This live transcript is auto-generated without human intervention or review.

[Call has ended.]

Hello, and thank you for standing by. My name is Tiffany, and I will be your conference operator today. At this time, I would like to welcome everyone to the SES AI First Quarter 2026 Earnings Call. [Operator Instructions] I would now like to turn the call over to Kyle Pilkington, Chief Law Officer; Kyle, please go ahead.

Hello, everyone, and welcome to our conference call of the first quarter 2026 Results. Joining me today are Qichao Hu, Founder and Chief Executive Officer; and Jing Nealis. Chief Financial Officer. We issued our shareholder letter just after 4:00 p.m. today, which provides a business update as well as our financial results. You'll find a press release with a link to our shareholder letter in today's conference call webcast in the Investor Relations section of our website at ses.ai. Before we get started, this is a reminder that the discussion today may contain forward-looking information or forward-looking statements within the meaning of applicable securities legislation.

These statements are based on our predictions and expectations as of today. Such statements involve certain risks, assumptions and uncertainties, which may cause our actual or future results and performance to be materially different from those expressed or implied in these statements. The risks and uncertainties that could cause our results to differ materially from our current expectations include, are not limited to those detailed in our latest earnings release and in our SEC filings. On this call, we will discuss non-GAAP financial measures of a supplement to our GAAP results. These non-GAAP measures are not prepared in accordance with generally accepted accounting principles but intended to illustrate alternative measures of the company's operating performance that may be useful.

These non-GAAP measures should not be considered in isolation or as a substitute for any GAAP measure, and our definitions may differ from those used by other companies for similarly titled measures. Reconciliations of the non-GAAP financial measures to the most directly comparable GAAP measures can be found in our latest earnings release.

With that, I'll pass it over to Qichao Hu.

Thanks, Kyle. Thanks, everyone, for joining that. We had a strong start for 2026. The first quarter revenue came in at $6.7 million a 47% increase over the fourth quarter and well above published consensus estimates. We are reaffirming our full year 2026 revenue guidance of $30 million to $35 million with contributions expected from all 3 of our revenue-generating business units. We are executing on plan and we like the momentum we have heading into the rest of the year. Before I get into the business update, I want to take a moment to acknowledge Jing Nealis who is on this call with us today. As we announced today, Jing will be transitioning from her role as Chief Financial Officer effective April 27.

On behalf of the entire team and our Board, I want to thank her for her contributions and wish her well. We have appointed Ray Li as our new CFO effective April 27. Ray is a seasoned finance executive with over 20 years of experience in FP&A, strategic finance and SEC reported at companies, including Ayden and MetLife Investment Management. He's a CFA charterholder and CPA, and we are confident he will be an excellent partner as we scale the business. More details on this transition are in the separate press release we issued today. Now let me walk through each of our business units.

Starting with energy storage systems. ESS remains our largest near-term revenue driver and was responsible for the majority of our first quarter revenue through use Energy. We continue to see growing demand for our commercial and industrial energy storage solutions, and our global footprint is expanding. Earlier this month, we provided a business update that highlighted our strong start to the year. Today, I want to add some additional context on the commercial traction we are seeing. We have now entered the North American market through our multiyear distribution agreement with ATG Power, a leading North American distributor of renewable energy and energy storage solutions that has been operating in the clean energy sector since 2001.

This contract valued at approximately $20 million over 3 years gives us immediate access to ATTE Power's established distribution network across residential, commercial and industrial customer segments. This new contract goes on UC Energy's existing customer base in Australia, the Middle East and Europe and reflects our strategy to grow the ESS business, both geographically and through the on-premise integration of our momentary universe predict capabilities into the hardware offering and Engebo. Energy Storage Systems are financial assets for our customers. The value depends on delivering consistent long-term performance.

Our ability to provide both the hardware and an intelligent operating system that predicts battery health and reduces maintenance cost is a key differentiator. Turning to drones. We made progress in our drone cell business during the first quarter that I want to walk through I am pleased to report that we have completed the conversion of our manufacturing line at our Chengdu South Korea facility from EV pulp cells to drone format cord cells. This facility, which produced the world's first 100 empower lithium metal call back in 2021 has been NDA compliant since 2021. Our plans are for the converted line to gradually ramp up to an annual capacity of over 1 million drone cells and incorporates our AI for manufacturing capabilities to ensure quality and cost-effectiveness.

Early this month, we began shipping NDA compliance cells producing Chengdu factory to prospective defense and commercial drone customers for evaluation and qualification testing. Customer interest has been strong, and we are encouraged by the engagement we are seeing. The U.S. defense drone market, in particular, continues to be where we see the most consequential near-term opportunity and our NDA compliant manufacturing capability in Korea positions us well relative to competitors who lack NDA compliance supply chains. We continue to explore additional NDA compliant manufacturing capacities in Southeast Asia and expect to have more to update on this front later this year.

On materials, our pipeline continues to build through the Molefe Universe platform, both SES and our customers have been discovering new electrolyte materials for applications beyond our current cell production. We now have approximately half a dozen customers who have progressed through second phase testing of materials discovered through the platform. And the overall number of customers in our pipeline has increased. The progression of existing customers through the testing pipeline represents positive momentum. We remain on track with the Raizen joint venture to leverage their 150,000 ton annual global capacity to produce these materials at commercial scale as demand materializes.

And on the Moleciuniverse, we recently introduced version 2.5 of the platform, which represents our fifth major iteration since we launched in 2024, version 2.5 delivers upgraded capabilities across our 6 AI-powered workflows as search formulate design, predict and manufacture, along with expanded enterprise on-premise deployment options and covering both lithium and sodium chemistries. During the quarter, a major global battery manufacturer committed to a multiyear subscription of our molecular universe search in the box product, which we view as a validation of the platform's value to the world's leading battery companies.

While the direct on-premise revenue from the molecular universe continues to build and is expected to make a modest direct contribution in 2026. Its biggest impact remains the IT and competitive advantages it drives across our ESS, drone, and materials businesses. We will continue to explore how best to demonstrate and unlock the molecular universe value over the course of the year. As we look to the remainder of 2026, our priorities remain clear: execute on the ESS opportunity through using energy and our growing distribution network, advance our drone cell business towards commercial scale customer engagement, deliver on the materials pipeline and continue developing the molecular universe as both a revenue stream and a competitive advantage. I want to thank the team for their continued execution and thank all of you for your continued interest in SES AI.

And now here's Jing for financial updates.

Thank you, Qichao. I will walk through our financial results for the first quarter of 2026. Given that our current 3 business unit structure took shape in the fourth quarter of 2025, with the integration of us [indiscernible] Energy and the launch of our drone sales and materials initiatives. We will present our first quarter results on a sequential basis compared to the fourth quarter of 2025, which we believe provides the most meaningful view of our operating projections. Revenue for the fourth quarter of 2026 was $6.7 million, representing a 47% increase over the $4.6 million in the fourth quarter of 2025. As a reminder, the fourth quarter of 2025 was impacted by approximately $1.5 million of revenue that was pushed into the first quarter, which benefited Q1 results.

Our revenue growth reflects the continued growth from UC Energy's ESS product revenue and early contributions from our drone sales and MU subscription revenue. We are reaffirming our full year 2026 revenue guidance of $30 million to $35 million. Our Q1 gross margin on a GAAP basis was 18.1% compared to 11.3% in the fourth quarter of 2025. On a non-GAAP basis, which excludes stock-based compensation as well as depreciation and amortization allocated to cost of revenue.

Our Q1 non-GAAP gross margin was 18.3% and compared to 11.7% in the fourth quarter of 2025. The sequential improvement from Q4 2025 reflects margin improvements from the UG ESS business and higher margin from sample phone sales and new subscription revenue. Turning to operating expenses. Our GAAP operating expenses for the first quarter of 2026 were $19.1 million compared to $18.2 million for the fourth quarter of 2025. On a non-GAAP basis, which excludes stock-based compensation as well as depreciation and amortization, first quarter operating expenses were $14.3 million compared to $13.5 million for the fourth quarter of 2025.

Our GAAP net loss for the first quarter was $12.1 million, a $0.04 loss per share, compared to a GAAP net loss of $17 million or $0.05 loss per share in the fourth quarter of 2025. I want to remind everyone that our GAAP net loss in any given quarter can be meaningfully impacted by noncash mark-to-market movements in the fair value of our sponsor earnout liabilities, which are required to be remeasured each reporting period under GAAP.

In Q1 2026, we recorded a $4.2 million non-cash gain related to these liabilities. These noncash gains or losses are not reflective of our underlying operating performance. And we believe, excluding them, provides a clearer picture of the progress we are making in the business. Excluding stock-based compensation, depreciation, and amortization, change in fair value of sponsor earnout liabilities and including interest income, our non-GAAP net loss for the first quarter was $11.1 million or $0.03 loss per share, compared to a non-GAAP net loss of $11.8 million or $0.04 loss per share in the fourth quarter of 2025.

Adjusted EBITDA for the first quarter of 2026 loss of $12.8 million compared to a loss of $13.8 million in the fourth quarter of 2025. We believe this continued progress reflects the positive operating leverage beginning to emerge in our business as revenue scale, combined with our sustained focus on financial discipline and cost management across the organization. We remain on track to deliver approximately 15% reduction in full year operating expenses that we guided on our last call. A detailed reconciliation of GAAP net loss to adjusted EBITDA and non-GAAP net loss per share is included in the financial tables at the end of the shareholder letter.

We utilized approximately $20 million in cash for operations during the first quarter, consistent with our operating plan. We exited the first quarter with a strong liquidity position of approximately $178 million. Our CapEx-light business model remains a core financial discipline, and we are confident -- our current liquidity provides a strong runway to fund operations and execute on our 2026 growth initiatives. On a housekeeping note, we expect to file a new S-3 shelf [indiscernible] statement concurrent with our 10-Q as our current shelf expires on April 28. This is a routine administrative filing to maintain our financial flexibility.

We believe -- the first quarter demonstrates steady execution against the plan we laid out. Revenue is on plan, costs are coming down, and our multi-revenue stream platform is taking shape. We are well-capitalized, financially disciplined and positioned to deliver on our full year outlook. Lastly, on a personal note, this is my last earnings call with SES. I am grateful for the opportunity to have helped build SES's financial foundation during the past 5 [indiscernible] years of the company. SES is well positioned to capitalize on the momentum it has built, and I look forward to seeing the growth story unfold. Thank you to Chica, my colleagues, our Board and our shareholders for the trust and support along the way. Thank you.

With that, I will hand the call back to the operator.

[Operator Instructions] Your first question comes from the line of Derek Soderberg with Cantor Fitzgerald.

So just on the evaluation and qualification tests, can you talk about the typical time line? How long might it take to transition those into firm purchase orders? .

Derek, are you referring to drones qualification or electrolyte, which one?

Drones.

Drones, qualification typically 1 to 2 quarters and then we've started those later. So most of the qualifications actually have been completed, and now it's just making those in our Korea facility and have the customers come in and do the supply chain audit, making sure all the cattle powder, the powder, the processing actually take place in Korea?

And then on the on-premise solution, I think you said you're going to have some contribution this year. Is there any chance you can quantify that at all for us? .

Probably in the next quarter, and then this last quarter, we did have one, one of the largest battery companies that actually signed up to the moly universe to search in a box. So only 1 of the 6 features. And then we have a few more in the pipeline that are interested in formulating in the box, predict in the box and also other features of the tool. .

Got it. And then 1 final 1 for me. On the drones, again, what's sort of the split between defense and commercial interest? Can you maybe break that out for us at all?

It's mostly defense, even though almost all the customers come to us for sale is to use like the same drones could be used for defense colleagues commercial in biology to customers that come in. So we focus a lot on customers that want NDA compliance and then only the customers that actually want to get defense contracts would really push for NBA compliance. We don't have a specific breakdown between defense and nondefense, but because also the customers don't tell us that. But we know it's actually predominantly defense.

Your next question comes from the line of Winnie Dong with Deutsche Bank.

My first question is on the multiyear distribution agreement with ATG Power. I was wondering if you can help us understand the relationship if this is like a wholesale relationship and of the $20 million order over 3 years, like what kind of shipment cadence we should be thinking about?

Similar to what I just mentioned, it's a wholesale distribution, and then they help us bundle the products with solar and then distribute that to their customers.

Got it. So essentially, once you ship it to them, you will be able to book revenue. That's how the set of it.

In terms of revenue recognition, the timing, Jing, is that correct? .

Yes, yes. So it's based on shipments, yes, once we ship it, based on the inco term, we will be able to recognize the product revenue, that's correct. .

And then on UC, you've achieved close to $7 million, I think some were spilled over from 4Q, what is like the typical seasonality of this business? And I understand that maybe it can be a little difficult. Since you're spreading across all different regions. But like holistically, is there a seasonality that we should be looking at for this business?

I think overall, the energy storage business globally have some sort of seasonality depending on the region. And Q2, Q3 usually are higher than Q4, but it also depends on the local incentives available like Australia, everybody is trying to secure something to be installed before the incentives go away, and in Europe, there are a lot of incentives going on before it goes away. So there are certainly season based on the region. However, because the sales to many regions globally is not tied to a particular place.

So, I think for this year, at least, we see growth quarter-over-quarter. With some seasonality, but I wouldn't put a lot of emphasis on that. But Q2, Q3 are probably higher.

And then maybe just a follow-up. Within the $30 million to $35 million, what is baked in, in terms of like contribution from materials and some of the other efforts that you guys have in place?

What's the breakdown?

Yes.

I think we expect this year to come predominantly from USS and then Russ spoke between drones and materials. .

Your next question comes from the line of Dave Storms with Stonegate.

Wanted to start maybe with ESS and your pension of the hardware offering Edge box, you could maybe a little time to speak in how that plays into the sales cycle, maybe what some of the benefits of it are.

Last part of your question again, the sales cycle and then the part after that.

Yes, just maybe some of the benefits of ad and edge box to your offerings and how maybe helping the sales cycle.

Yes. So the hardware is pretty competitive and basically, you purchase sales and you integrate those into a container. And then the industry, the accuracy, the error is typically 7% or even as high as 10%. So not so accurate. And then as a result of that, for example, if your project only needs 10 kilowatts, you will buy 14-kilowatt hours to basically allow for the error. So having this edge box the edge box does 2 things. One is we can very accurately tell the state of charge, the state of health, safety, energy, power, basically, what we call [indiscernible] this picks up them.

And it can give a really accurate estimation of that. So is the are being 7%, 10%, now we're talking about 3% or even less. And then the other benefit it's itself on the cloud, which have customers don't like. It's totally secure. It's in a box that we actually put on-premise. So you also have data security. So the main benefit of that is now that instead of buying more capacity to allow for the inaccurate estimation, you can buy less. So the customers can save cost.

And for some of the customers that want to participate in virtual potline, basically, SES trading and then sell electricity back to the grid. And because we have a more accurate estimation than your peers, you can bid in a more competitive price. And also you can -- when you make the decision of whether not to participate and the trade-off versus sacrificing the battery health, you can have a more accurate estimation of that trade-off?

And then maybe just turning to materials. It was mentioned that there's several companies completing their second phase. Maybe just thoughts around timing through this next step, this third phase as they advance towards commercial scale supply discussions?

So typically, it's a 2 to 3 rounds of testing each around about 1 quarter. So we talked about 6 to 9 months of testing. And then towards the end of the last round of testing, then the customer will go through what's called commercial qualification basically the check for the plant and also check for all the toxicity, the per chemical permits needed for any special materials inside the formulation and then making sure it's complying to all the necessary local environmental toxicity chemical regulation.

And then overall, the testing 6 to 9 months and then another quarter for the commercial qualification. But again, we started a lot of this last year. So now we are -- with all of these customers, we are towards end of second round of qualification.

And maybe just 1 more quick modeling 1 for me. You reiterated 15% expense reduction throughout the year. Should we expect that to kind of go on a linear glide path throughout the year? Or maybe just any thoughts around the cadence of those expense reductions.

Jing, do you want to take that?

Yes, I'll take that. So -- we are taking lot of actions to further reduce our operating expenses starting from Q1. So you should be able to see the full quarter impact starting from Q3, there will be a little bit of a reduction in Q2, but not full quarter. But starting Q3, the full quarter impact should be coming in. So than Q4, maybe slightly lower than Q3.

[Operator Instructions] Your next question comes from the line of Sean Milligan with Needham.

In terms of the 1 million units that you're targeting, for the drone cell business. Like can you talk to what that potentially represents from a revenue standpoint? And then the second question is you've mentioned that you've been testing sells or qualifying selves with potential customers there. Is there any context you can give us to the pipeline and maybe kind of sizing of initial orders that you would expect to see?

Sure. So the $1 million is still not the full capacity that Korea factory could go up to much higher. Like all that investment we made for EV and then turned out, we excellently built 1 of the largest drone pole manufacturing factories outside of China. So we have a lot of customers that 1 NDA compliant cells come to us. And the market price for NDA compliance cells obviously, depending on the specific cell format ranges between $25 to $35 as the market price. So 1 million units it's about $25 million to $35 million. That's just $1 million, and then we could again go to much higher if needed.

And then -- in terms of the qualification process, again, we did -- we started most of the testing last year. So now we're doing -- so the performance and the product has been have been completed. And now a lot of that is actually supply chain audit and cost schedule.

Okay. Is there any way to talk about the pipeline, like the number -- so if you look at the revenue guidance this year, I think you said some of that comes from the drone business, but it obviously could be a much bigger piece of business. I'm just trying to understand the pipeline looks by number of customers that you're testing with. Any kind of stats that can help us kind of gain some sense of potential momentum.

So we have a pipeline of a few dozen customers -- and again, we focus on customers that want NDA compliant cells. And then really -- so we actually had some shipment recently. So we expect revenue in Q2 for the NDA compliant sales and then started to pick up Q3 and in Q4 and then really next year 2027 is going to be a full year where we actually have the ability to deliver a full year of these NDA compliance cells.

There appear to be no further questions at this time. Ladies and gentlemen, this concludes the SES AI First Quarter 2026 Earnings Call. Thank you all for joining. You may now disconnect.

Good afternoon. Thank you for attending today's SES AI Fourth Quarter and Full Year 2025 Earnings Results Call. My name is Tamia, and I will be your moderator for today's call. [Operator Instructions]

I would now like to pass the conference over to your host, Kyle Pilkington, Chief Legal Officer.

Hello, everyone, and welcome to our conference call [ during ] our fourth quarter and full year 2025 results. Joining me today are Qichao Hu, Founder and Chief Executive Officer; and Jing Nealis, Chief Financial Officer. We issued our shareholder letter just after 4:00 p.m. today, which provides a business update as well as our financial results. You'll find a press release with a link to our shareholder letter and today's conference call webcast in the Investor Relations section of our website at ses.ai.

Before we get started, this is a reminder that the discussion today may contain forward-looking information or forward-looking statements within the meaning of applicable securities legislation. These statements are based on our predictions and expectations as of today. Such statements involve certain risks, assumptions and uncertainties, which may cause our actual or future results and performance to be materially different from those expressed or implied in these statements.

Risks and uncertainties that could cause our results to differ materially from our current expectations include, but are not limited to, those detailed in our latest earnings release and in our SEC filings. On this call, we are introducing non-GAAP financial measures as a supplement to our GAAP results. These non-GAAP measures are not prepared in accordance with generally accepted accounting principles, but are intended to illustrate alternative measures of the company's operating performance that may be useful. These non-GAAP measures should not be considered in isolation or as a substitute for any GAAP measure, and our definitions may differ from those used by other companies reporting similarly titled measures. Reconciliations of the non-GAAP financial measures to most directly comparable GAAP measures can be found in our latest earnings release.

With that, I'll pass it over to Qichao.

Thanks, Kyle. Thanks, everyone, for joining today. We had in [ value ] in 2025 with full year-end revenue of $21 million compared to a little over $2 million for 2024. Jing will walk through our financials and the outlook shortly. This tremendous growth was due to the final contributions from our services agreement with Honda and Hyundai as we completed our DV development work with [ them ].

We also had 3.5 months of revenue from the acquisition of [ US ] Energy for our energy storage ESS business. While we are pleased to report full year revenue in the range of our previously issued guidance, as the milestone we reached [ and ] the year-over-year growth we are expecting from full year contributions in our three revenue-generating business units and for recognizing potential value in the molecular universe that has us really excited.

I'm very proud that we made more progress in the past year than the previous 10-plus years combined. This development of the molecular universe has been up for us. SES continues to be well positioned to solve the issues of battery development and safety requirements with the molecular universe, our own in-house [ Arasience ] company we have been able to help customers overcome standard time line in the adoption of new technology. We also have [ a ] how energy transition needs are requiring more integration of AI software and hardware along with precise battery health monitoring.

As we described before, SES has three revenue-generating business units. ESS, [ loans ] and materials. ESS, which is the largest market for [ Tres ] is bigger than EVs and bigger than drones. At Battery World 2024, we announced our entry into the ESS market and through our acquisition of UZ Energy, we are now serving customers across the globe, from Australia to Europe to the Middle East, and now we are entering the North American market. The [ U.S ]. business is our largest near-term revenue driver. UZ Energy leader in commercial and industrial ESS and has sold almost [ 1 ] gigawatt hours of hardware to customers ranging from [ residential ] C&I to great.

We are now able to collect the large amount of historical ESS [ LFT ] graphite data and then for all future using products, we plan to incorporate our predict feature from moat universe into a small box to achieve near 0 sales charge [ SoC ], degradation, health, safety and other SLX algorithms and automated nondisruptive retail operation. which helps improve UE's ability to predict battery health and reduce maintenance costs for customers. Energy Source systems are financial assets. The value to our customers depends on delivery consistent and long-term performance.

Historically, us supplied only the hardware to customers, mostly [ LT ] and grass lithium-ion cells. Now since we acquired UZ, as has the opportunity to provide an operating system to the hardware and sell customers a complete package to meet their USS needs. We are seeing some early traction with us sales efforts as they were able to sign a multiyear $20 million contract with a major distributor recently at the Intersolar Conference Loans for the next business unit, I want to highlight.

The drones [ base ] requires high image density and high power density batteries to achieve longer flight time and greater payload. This is where our lithium metal on high silicon carbon lithium ion batteries really shine. The U.S. defense drones market in particular is what we see the most consequential near-term opportunity, and we'll be voting most of our attention and investment. This worth spending a moment on why the drone market is a natural fit for deploying our lithium metal [ anode ] and proprietary electrolyte.

One, and drone battery needs double the end density of conventional lithium ion. Ultimately, you need at least 400 watts per kg to be state-of-the-art and a realistic road map to your 500 molest win. In other words, range and payloads matter. Drone batteries need a high sea rate and power for the drone to maneuver and accelerate, we saw for this.

Third, drone batteries need to be manufactured at scale. We've demonstrated this in our PVB sample development. Fourth, drone batteries need abundant and inexpensive material and the military demand to the supply chain to be National Defense Authorization Act for NDAA compliant. We recently announced we expect to convert our EV B sample line in [ Trubiouth ] Korea facility to manufacture NDA compliance cells for drones. This is the same facility that we've develop and build the world's first 100 power largest within metal cell back in 2021. And this facility has been NDA-compliance since 2021.

To meet the drone demand, we plan to convert our lines from EV [ pulp ] 100 cells to 10 pulp cells. In this line, we're also planning to deploy our AI for safety and AI for manufacturing to ensure quality and cost effectiveness. In addition to our Korea facility, we're also exploring even larger NDA compliance and more versatile cell form factor no country capacities in Southeast Asia. We'll have more update on our NDA compliant manufacturing capacity nations later this year. Our third revenue-generating business unit is materials.

Both FCS and our [[indiscernible] have also been discovering new [indiscernible] for other applications that we don't build sales for currently. Last fall, we announced a JV with [ Hisun ] to leverage their 150,000 ton annual global capacity to produce these materials at a commercial scale to supply to other battery manufacturers for consumer electronics and assets. At this time, we are anticipating that the [ Ryzen ] JV will only produce materials for molecule universe discoveries. Through the molecular universe, we discovered six breakthroughs that are currently being tested by over 40 customers.

These breakthroughs, which will be the basis of the revenue we expect from this business in 2026 in one, [ beta-synthase ] in stores for EV applications, two better cycle and power density for drones. Three, better low-temperature cycle life and power density for heavy-duty trucking four, [indiscernible] and longer life for consumer electronics. This beaches and low temperature performance for ESS and EVs and better cyclic in storage for consumer electronics. We also have a pipeline of new breakthroughs that are being tested by customers, which we expect will provide further potential revenue for this business.

[ Last ] not least is what we have been referring to as our own AI science company. That is, of course, molecular universe. I want to be clear on how we view the role in this company. While SaaS revenue continues to build momentum and is expected to make a small contribution in 2026, this business contribution is the inherent value of this business on its own and the IP that drives competitive advantage in the [ U.S ]. drone and materials business. Molecular universe has the potential to become a modern day inside later with battery being [ value ] 1, provides extremely valuable scientific data and intuition to [ AF ] science models.

Over the course of the year, we will continue to explore how we can best demonstrate for unlock and use value. In terms of demonstrating that value, I will point to our recent investor presentation. In that presentation, we noted there are several AI for science companies that are either pre-revenue or have less revenue than the MU that have already passed valuations exceeding $1 billion through private capital raises. These are the closest comps to the molecular universe, so with [ TCI ] on how well these transactions have one.

So we're really excited about the long-term value of molecular universe as a platform, not just for batteries, but all as well as the near-term revenue growth from drones, materials and ESS operating systems. Our priorities for 2021 and beyond are one, leverage the new business unit leadership and structure to execute on the ESS and drone sales opportunities ahead of us.

We've [ brought ] on industry veterans to lead these efforts as well as hardware, software integration sale. Second, actually on the conversion of our NDA compliant line in Korea from EV sales to drone cells and line up additional capacity in Southeast Asia that is also NDA compliant. First, continue the growth of us Energy's existing hardware business in Australia, Middle East and Europe and began expansion into the U.S. Fourth, delivery on existing novel electrolytes discover biomolecular universe in the materials business and expand our pipeline.

Fifth, leverage [ used ] material discovery capabilities to accelerate new product development and continue to focus on our CapEx-light business model in U.S. sales and materials to offset the projected R&D spend in the [ Blake ] universe.

Before I turn it over to Jing, I want to express my gratitude for our teams who are working super hard to make all of this happen. And thanks to all of you for being on this journey with us.

And now, here's Jing for financial updates.

Thank you. I will discuss our financial performance for the fourth quarter and full year of 2025 and provide context on how we're deploying our capital to support SES AI's long-term growth and the strategies Qichao outlined earlier. Revenue for the fourth quarter of 2025 was $4.6 million, representing a $2.6 million or 124% increase year-over-year. Full year revenue came in at $21 million, in line with our guidance, but impacted primarily by logistics constraints that delayed shipments at the end of the year, resulting in approximately $1.5 million of revenue being pushed out to the first quarter of 2026.

As Qichao noted earlier, revenue for full year 2025 was within our previously issued guidance range of $20 million to $25 million and was up nearly tenfold from the prior year. A year in which we first achieved revenue generation. Our Q4 gross margin on a GAAP basis was 11.3%, driven by the higher mix of ESS product sales in the quarter, which carries a lower margin profile relative to our service revenue.

On a non-GAAP basis, which excludes stock-based compensation as well as depreciation and amortization allocated to cost of revenue, our Q4 non-GAAP gross margin was 11.7%. For full year 2025, our GAAP and non-GAAP gross margin was 53.8% and 55.7%, respectively. As we have noted previously, we expect gross margin to vary from quarter-to-quarter as our revenue mix across products, SaaS and [ thesis ] evolve. We expect the gross margins on our product revenue to improve as we scale volume and optimize the cost structure through our CapEx-light business model and JV partnerships.

Turning to operating expenses. Our GAAP operating expenses for the fourth quarter of 2025 were $18.2 million, compared to $30.4 million for the same period prior year, a 40% decrease year-over-year. On a non-GAAP basis, which excludes stock-based compensation as well as depreciation and amortization. Fourth quarter operating expenses were $13.5 million compared to $24.2 million for the same period prior year, a 44% decrease. For full year 2025, our GAAP operating expenses were $93.9 million, compared to $110.5 million in 2024, a 15% decrease.

On a non-GAAP basis, full year operating expenses were $73 million versus $82.3 million in 2024, an 11% decrease. The year-over-year improvement in operating expenses on both GAAP and non-GAAP basis reflects the progress we have made in optimizing our cost structure while continuing to invest strategically in Molecular Universe platform and our commercial growth initiatives.

Adjusted EBITDA for the fourth quarter of 2025 was a loss of $13.8 million compared to a loss of $23.2 million in the fourth quarter of 2024, representing a 40% improvement. For the full year 2025, adjusted EBITDA was a loss of $52.6 million compared to a loss of $81.5 million in the full year 2024 in 23% improvement year-over-year. We believe this growth reflects the positive operating leverage beginning to emerge in our business as revenue scales, as well as our sustained focus on financial discipline and cost management across the organization.

Our GAAP net loss for the fourth quarter was $17 million or $0.05 loss per share. Excluding stock-based compensation, depreciation and amortization, changes in fair value of sponsor earnout liabilities and including interest income, our non-GAAP net loss for the fourth quarter was $11.8 million or $0.04 loss per share. This is an improvement over 2024 fourth quarter's GAAP net loss of $34.5 million or $0.11 loss per share and non-GAAP net loss of $19.9 million or $0.06 loss per share.

For the full year 2025, our GAAP net loss was $73 million, or $0.22 loss per share compared to a GAAP net loss of $100.2 million or $0.31 loss per share in 2024. On a non-GAAP basis, full year net loss was $53.2 million or $0.16 loss per share compared to a net loss of $66.4 million or $0.21 per share in 2024. The year-over-year improvement on both GAAP and non-GAAP basis reflects the progress we are making in scaling revenue and managing our cost structure as we advance customer engagement, develop the molecular universe platform and position the distance for growth in 2026.

A detailed reconciliation of GAAP net loss to adjusted EBITDA and non-GAAP net loss per share is included in the financial tables at the end of the shareholder letter. I want to highlight that our GAAP net loss in any given quarter can be meaningfully impacted by noncash mark-to-market movement in the fair value of our sponsor earned liabilities, which are required to be remeasured each reporting period under GAAP.

These noncash gains or losses are not reflective of our underlying operating performance. And we believe, excluding them, provides a clearer picture of the progress we're making in the business. This is one of the reasons where we're introducing adjusted EBITDA beginning this quarter. We utilized $10.4 million in cash for operations during the fourth quarter and $58.4 million for the full year 2025.

We deployed $3.3 million on the UZ acquisition and $2.9 million on CapEx and returned $1.6 million to shareholders through share repurchases during 2025. This improvement in cash utilization is consistent with the adjusted EBITDA progress I noted earlier and reflects the financial discipline we have maintained as we scale the business. We exited 2025 with a strong liquidity position of $200 million, coming in at the top end of our previously communicated expectation of ending the year between $195 million and $200 million.

Our CapEx-light business model remains the core financial discipline, and we are confident our current liquidity provides a strong runway to fund operations and execute on our 2026 growth initiatives. For full year 2026, we expect revenue to be in the range of $30 million to $35 million. representing approximately 43% to 67% growth over full year 2025 revenue. As Qichao noted earlier, our full year 2025 revenue included onetime contribution from OEM services contracts.

If we compare the growth expected from the three businesses on an apple-to-apple basis, the revenue growth rate is even higher. On the margin front, our three businesses carry different profiles and the mix may shift as we scale our ESS hardware business, which will represent the largest share of revenue in 2026 is expected to operate at around 15% gross margin. As we layer in the hardware software bundle and grow the operating system attach rate, we see a potential path to expanding margins in that business over time. Our drone sales business is earlier in this commercial ramp, but we expect gross margins north of 20% as volumes build through the year.

Our materials business, which will sell electrolyte materials, through our joint venture with Hisun. It's also a product business, and we expect it to carry a margin profile in the 10% to 20% range. On a blended basis, we expect consolidated gross margin to be around 15%, with room to improve year-over-year as we scale. Our operating expenses for full year 2026, we expect approximately 15% further reduction from the 2025 level. This reflects our continued investments in the molecular universe platform, while we are committed to financial discipline, we believe this level of investment is appropriate given the long-term value we're building and the early commercial traction we're seeing from MU driven material discoveries.

We are not anticipating meaningful growth in operating expenses beyond this level, and we'll continue to evaluate opportunities to improve operating leverage as revenue scales and to accelerate the monetization of the MU platform. On capital expenditures, we continue to operate a CapEx-light model at the core financial discipline. For 2026 we expect CapEx to remain in the single-digit million, primarily directed towards the conversion of South Korea facility from EV cells to NDAA compliant grown cells, as well as the evaluation of contract manufacturing capacities in Southeast Asia.

We entered 2026 with $200 million in liquidity and we are well funded to scale and grow, can us the financial flexibility to execute on the opportunities ahead of us. We believe 2026 will be the year in which the full architect of our multi-revenue stream platform comes together and began to deliver. We are well capitalized, financially disciplined and positioned to execute on that vision. We appreciate your continued support and confidence in SES AI. Thank you.

Now I will turn the call back to the operator.

[Operator Instructions] The first question comes from Derek Soderberg with Cantor Fitzgerald.

On the final contribution from the Honda and Hyundai development work. Just was wondering what sort of next for that program? You sort of proven manufacturability recently. Obviously, your technology is sort of game changing for the EV market, what's next for that, those relationships and Hyundai, when are you going to commercialize that for EVs?

Derek. So in terms of next step, previously, the next step was to go from B sample to C sample. And I think now, I mean there's no surprise that the EV market is slowing down and almost no automaker is investing in next-gen battery technology. And I don't mean like early-stage better technology in terms of A and B samples.

But no one is investing in mass scale production of next-gen technology, which is C sample, which is what we were trying to get to next. So we hit all the technical milestones, but the C sample is on hold. And then so in terms of next step with the OEMs, we are focusing on selling materials. Selling materials that we have developed the [ laser ] materials we are [ foreseeing ] on that. And in terms of the full-blown with the metal C sample, then we'll see when the market returns.

But the technology is there, and this is why we've been forcing on converting the lines for drones production and also applying the the AI for safety, the battery analytics software that we developed for EV for ESS market. So yes, in terms of next steps for the OEMs, we're focusing on material supplies and then also converting the line for drone applications and using the the safety analytics software for ESS.

Got it. And then just quickly, what was the onetime service amount? Can you quantify that impact to fiscal '25?

Jing, do you want to take that?

Yes. So for 2025, the service revenue was $13.6 million. Those are primarily driven by the Honda and Hyundai service agreement. So that's the onetime service agreement we were talking about.

Got it. So for '26, you don't expect any of that to sort of recur just given what Qichao just explained or guidance is really just ESS, drones and materials. And I was wondering if you could sort of break that down for us or help us try to understand by segment. ESS, drones, materials. Can you help us kind of quantify how each of those contribute to the guidance range? And then any help sort of modeling kind of the first half or second half? What portion of revenue in the first half, second half, anything like that would be helpful.

I can take that. So yes, yes, for 2026 guidance, given is the first year we're giving guidance, including all three business units. So we -- first of all, we wanted to be more conservative to start the year. As far as breaking down the revenue sources, ESS going to still be a large portion of our revenue. So of that 30 to 35 guidance, the range, I would say, probably around 65% will come from ESS at least.

And then the remaining portion will be drones and materials. And then for those two, we're expecting it to be more second half of the year low [ in ] given we're still in the ramping and business development stage. So those two are going to be more towards the second half of the year. So -- but percentage wise, around 65% from ESS and the remaining for those two.

The following comes from Winnie Dong with Deutsche Bank.

Just curious, if we look out to maybe like the next 2 to 3 years, if we just look at the different business areas, yes, there's [ strong ] materials. And also, you have [ to ] molecular universe as well. how would you characterize like the growth profile of each of these areas in the next 2 to 3 years? Which one has maybe the potential for the largest growth if there's a way to think about it? And then separately, for molecular universe, I know you've been talking about different tiers of revenue from larger corporations and smaller ones. Could you share currently what might be the largest bottleneck for adoption from these customers? And then I have a follow-up.

Yes. We -- on the first one, from a size of revenue and the product perspective, ESS and drones, we expect these to grow very rapidly especially on ESS after we acquired UZ. We're not just selling the hardware anymore. Now we're adding this predict feature on top. So that's really exciting because now we are turning a regular UPS battery pack into an asset that the asset owners can use for energy trading, so like supply and demand.

And on the demand side, you have conventional [ VTP ] software but on the supply side. no of the VPP software can have as an accurate and advanced monitoring of the battery health battery safety, better degradation and all the parameters than ours. So we have a really precise estimation of of the set of health of the battery, so that gives you better. It's almost like we say that the energy trade it's almost like having [ Warren ] buffer at your fingertip when it comes to energy trading if you know that level of precise health of your battery. So that we are really excited about this Edge box enhanced virtual power plant, especially with those tool that we can add and help both use this hardware. So that's -- yes.

And in drones, so it's all about supply chain. It's all about being NDA compliant and supply into U.S. and allied drones. And then also with the new drones dominance program, -- we've had this line in Korea, and it's been NDA compliance since 2021. We built this line for GM. It was 100-hour dose metal cells, we're converting that to 10 npower cells for for drones.

So we have an advantage because we have this asset. It's been in the compliance since 2021. And then that market is also growing very rapidly, especially under the new drones dominance program. So I think these are the most exciting from the size of revenue growth. And as Jing mentioned earlier, the bulk of the revenue from this year 2026, we expect will come on these two areas.

In terms of molecular universe, we are making very exciting progress and we're getting some of the largest battery companies and car companies to sign on to this this platform, and we expect to make some announcements in the coming months. And I think in terms of bottleneck, I think just is just new.

For example, [indiscernible] has been using drug discovery a lot, not so much in materials, not so much on chemicals and not so much in in batteries. So it's just new. But finally, as we said, we use that platform and then we demonstrated that that we could actually indeed use that and develop 6 new materials. I mean, previously, it was -- it would take you a few years to discover one material. And when we've discussed six materials in just over about 9 months, 9 months to a year.

And then later, we'll add more features to it, and hopefully, we can get to six new materials per month and per week. So that acceleration of rate of discovery, that's something we're quite excited about.

Got it. Thank you for the detailed response. And then I wanted to ask the question about OpEx in 2026. It seems like you're characterizing a spending level that is lowered in 2025, and that is going to likely sustain at this 2026 [ vulnerable ] quarter basis.

Can you maybe just understand the reason for that, if you're looking to grow these different areas of the business, what is it that you've done that I guess, allows you to not have to spend further to grow the business.

I can address that. So I think the 2026 reduction partially is coming from just being disciplined on spending cash on OpEx in general. We have been reducing G&A and also R&D expenses year-over-year if you go back to 2023, 2024, 2025 year-over-year, we're managing our costs very efficiently. So that's that part.

And second, the MU as an internal tool for AI for science is creating a lot of efficiencies. So -- and also as part of growth into these three businesses, we're more focused as far as spending cash on product development related R&D and then there will be growth as far as spending wise on the SG&A side like sales and marketing, but it's not linear to the revenue growth. So overall, together, including R&D and SG&A we forecast this year to be lower than last year and then sustain at least for the foreseeable future.

[Operator Instructions] The next question comes from [ Colin Rusch ] with Oppenheimer.

Can you talk a little bit about the drone market and the volume of customers you're working with and how mature those relationships are in terms of working through the design process. and potentially being able to announce a purchase award here over the next, call it, several quarters.

Yes. The Jones market is really going through a lot of pressure to change supply chain with NDA compliance requirements. And then we are focusing on some of the top customers, [ resemble ] customers that will order in the range of single-digit millions to financially more than $10 million a year. Yes. So we have many focus on those larger customers. And we actually started testing engagement with them last year.

Last year was really when everyone tried to change the supply chain. And I think if a major Jones manufacturer hasn't changed the supply chain by now, I think it's almost a bit too late. So a lot of testing engagement started last year. And then now we are are just in the final stage of converting the lines. For example, right now in Boston, we can make pilot scale less than 100,000 cells per year. In Korea, we can make about 200 to 300 cells per year. We're trying to expand that. And then in Southeast Asia, we're also looking to expand to several million cells. This all for -- this is all NDA compliant sales for drones customers.

Excellent. And then obviously, you had some really meaningful success at the molecule level, leveraging some of the AI capabilities. I'm curious about your ability to leverage some of that know-how into pack level design and even into system integration design and modeling out some of the duty cycles that that may be just to accelerate some of the adoption as you look at some of the robotics and drone opportunities?

So you're saying we can apply those [ two ] pack level and sell cell level?

Yes, at the pack level and and even the system level design beyond that pack level?

Yes, we're starting to add that feature. We've got some requests from automakers that want to do the design and then predict features at the past and system level. So we are adding those features. And then also for energy storage, right now, we are adding molecule universe predict into the systems. [ The ] we put that in the small box, we call the Edge box and then that works at the cell level and also the tend system level.

Next question comes from Mark Shooter from William Blair.

I believe I heard you say that the auto OEM JVs are on hold. Could you clarify that a bit? And then I'm seeing that the industry is, especially [ auto ] industry is kind of move away from lithium metal. However, at the same time, I'm seeing some local competitors actually enter the public markets here with our lithium metal product. So -- is any of the engagement appetite with your OEM customers changed around lithium metals? How are you looking at this?

Yes. So I mean, we were developing pure lithium metal as well as hybrid lithium metal and silicon anode, and we met all the technical requirements. It's just in terms of OEM appetite for high energy density batteries, I would say, back in 2021, the OEM appetite for high energy density batteries was very high. But now, and that's still there, but maybe at R&D level, a sample level and demo car level, but not at C sample level, which is like mass production. And and I'm not seeing any other OEMs that's going to mass production with a [ nation ] chemistry. There's a lot of price pressure, cost pressure and most OEMs are switching to just [ LFP ] graphite.

Yes. Okay. And that makes a lot of sense. And that justifies what we're seeing, too. So that confused me a bit. But switching gears to the ESS market, which is rapidly growing, that is a fragmented market with many different levels to it. And I'm wondering what do you see as the most value add? Where would you play? What is the strategy for the UZ Energy acquisition? And what section of that energy storage market would you play in? And why are you most advantaged to that section?

Yes. Exactly. The ESS market is very fragmented. It's got a long tail. And and the benefit that we bring to the U.S. market currently does not have a stable widespread operating system, maybe except for Tesla. But in the U.S., it is at Tesla in [ the ] long tail very fragmented. So what we can provide is almost like the Android version. And then -- so for commercial industrial and for data center applications, the asset owners actually want to use their battery packs for energy trading, but they're not able to do that and are not able to differentiate if they use conventional virtual power plant software.

So what we can do is because we actually collect the data from the battery, we can have a very accurate estimation of sales charge, still [ health ], sale safety degradation, power all these different features. What that means is when you do the trading, supply and demand. Demand is set by the market by weather, by if there's any major sporting events. But the supply, that's that's set by accurate estimation of [ SOX ] and then that we can provide. So that's where this Edge box enhanced virtual power plant, really shines. So I think the value that we can provide -- we can provide this operating system to not just use these battery packs, but to multiple of the long tail, this fragmented market.

There are no questions from queue. So I'll pass back over to Kyle Pilkington.

Thank you. As within our past earnings calls, we offered investors the opportunity to submit questions in advance and we'll cover a brief selection of those questions now. The first question is, is there a binding definitive agreement with top material yet? And can you generally provide some details on current NDAA compliance status?

Yes. So that material is one of the options we are exploring. And then as I mentioned earlier, we've had this Korea facility since 2021. It's been NDA compliance in 2021. And then we are focused on converting those to produce drones, batteries. And then in addition to our own career facility, we're also looking at options in Southeast Asia and then potentially offer better pricing, better value and larger scale to customers. And again, they're all NDA compliant. And actually, we just updated our website and now we have updated drones battery product brochure that's on the website. So and then the cells are NDA compliant.

Great. I think we have time for one more. So the question is, with Wildcat and BMW and Ionics and [ Portia ] securing JDAs for AI dream materials, how is SES protecting its molecular universe data advantage to ensure it doesn't lose OEM partners to lease specialized private competitors?

So I think we announced that we're offering molecular universe to the public mid last year after these other announcements were made. And then what universe, if we look at the latest version of [ Mi ] Universe 1.5, and we have a 2.0 coming out soon, it really is a game-changing platform for battery development in the EV space. So we're seeing a lot of the OEMs and battery companies actually using this platform.

Great. That's all we have for the investor questions. So I'll pass it back to the operator to conclude the call.

This concludes today's conference call. Thank you for your participation. You may now disconnect your lines.

Hello, and welcome to the SES AI Third Quarter 2025 Earnings Release and Call. My name is Carla, and I will be coordinating your call today. [Operator Instructions] I will now hand you over to the Chief Legal Officer, Kyle Pilkington, to begin. Please go ahead when you're ready.

Hello, everyone, and welcome to our conference call covering our third quarter 2025 results. Joining me today are Qichao Hu, Founder and Chief Executive Officer; and Jing Nealis, Chief Financial Officer. We issued our shareholder letter just after 4:00 p.m. today, which provides a business update as well as our financial results.

You'll find a press release with a link to our shareholder letter and today's conference call webcast in the Investor Relations section of our website at ses.ai. Before we get started, this is a reminder that the discussion today may contain forward-looking information or forward-looking statements within the meaning of applicable securities legislation.

These statements are based on our predictions and expectations as of today. Such statements involve certain risks, assumptions and uncertainties, which may cause our actual or future results and performance to be materially different from those expressed or implied in these statements. The risks and uncertainties that could cause our results to differ materially from our current expectations include, but are not limited to, those detailed in our latest earnings release and in our SEC filings.

This afternoon, we will review our business as well as results for the quarter. With that, I'll pass it over to Qichao.

Thanks, Kyle. Thanks, everyone, for joining today. We had a record third quarter with more than $7 million in revenue. That's more than 100% growth over the second quarter. Our all-in on AI strategy is working remarkably. Today, I want to highlight some of the successes we've seen with this strategy and what it means for the future.

We reached a major milestone this quarter that we expect to have far-reaching consequences across the revenue machine we described in detail during our last call. That milestone was the release of our latest version of Molecular Universe, MU-1.0. MU-1.0 is a powerful and complete end-to-end AI for science workflow that includes 5 features, ask an Agentic LLM with access to what we believe is the world's largest database of battery relevant literature, search and formulate what we believe are the world's largest databases of battery relevant molecule and formulation level properties enabled by GPU-accelerated quantum mechanics computation and machine learning accelerated property prediction, end design and predict chemistry-specific and chemistry-agnostic machine learning models, respectively, that can accurately predict battery state of health and end of life.

Due to the popularity of the enterprise tier, we also launched 3 sub-tiers within enterprise to provide greater value to more enterprise users. In addition to cloud-based molecular universe, we expect to launch on-premise molecular universe, providing greater data security to more enterprise users. This new on-premise capability, which we will be describing in more detail in the coming months, addresses specific security and privacy needs of the world's largest battery makers that should unlock a greater share of our addressable market. We are incorporating MU1.0's Ask design and predict into our ESS products deployed by UZ around the world to collect data for on-site model training.

This unprecedented ability for safety and health prediction, combined with reduced maintenance costs, truly helps differentiate our products and attract new customers. Since we completed the acquisition of UZ Energy in mid-September, our ESS revenue has been growing and is already responsible for approximately 45% of our third quarter revenue. We're very excited about the revenue potential of deploying Molecular Universe to enhance our ESS products.

MU1.0' Ask, Search and Formulate are also helping our users identify several new electrolyte materials that we are commercializing. These include: one, improved low-temperature rate performance of lithium ion phosphate lithium-ion cells for ESS applications; two, improved cycle life for 12% silicon lithium-ion cells for EV application; and three, improved cycle life for lithium metal and 100% silicon lithium-ion for drones and UAM applications and many more.

To supply these materials discovered by Molecular Universe to our customers, we entered into a joint venture agreement with Hisun New Energy Materials, a leading electrolyte manufacturer with 150,000 tons of annual capacity to contract manufacture these materials, so we stay CapEx-light, laying the groundwork for exciting revenue growth in the coming quarters.

Another revenue opportunity we expect to grow in 2026 and beyond is in drones. A dependable supply chain of high energy density pouch cells is extremely rare and critical to the development of the American drones industry. To better address this burgeoning market, we are leveraging our Chungju, South Korea cell factory, incorporating the latest materials discovered from Molecular Universe to meet customer demands.

In terms of potential revenues from EV, we completed B-sample line site acceptance test this summer with one auto OEM. As a result, in 2026, we expect to start commercial supply of electrolyte materials and partner with them for cell production. Overall, it's hard for us to comprehend a more consequential period than what we have experienced over the past 2 quarters, particularly as it relates to delivering on the goals we outlined coming into this year.

For instance, we noted we wanted to break into the ESS market in a big way. Now we've done so with the acquisition of UZ Energy, and the acquisition has already delivered significant revenue in 2025. We launched 3 versions of Molecular Universe this year. The discoveries made by us and our customers so far have accelerated our push into materials to supply them through the Hisun JV.

We'll have more to share on our fourth quarter call about how we expect 2026 to shape up for us, but we expect success for us will look like a hardware, software integrated platform with multipronged and multi-revenue streams. As Molecular Universe, a complete AI for science workflow SaaS platform accelerates innovation across all battery chemistries, we are working with our JV partners to provide a dependable hardware supply chain for the cells developed from Molecular Universe.

Just as AI for science is completely changing other industries, Molecular Universe is now transforming all battery chemistries across all applications. We are excited about the revenue growth potential brought by Molecular Universe and we'll continue to assemble the best talent, data and compute resource needed to build AI for science for energy transition.

Lastly, I want to express my gratitude for our teams who are working super hard to make all of this happen. And thanks to all of you for being on this journey with us. And now here's Jing for financial updates.

Thank you. I will discuss our financial performance for the third quarter of 2025 and provide some context on how we are deploying our capital to support SES AI's long-term growth and then the all-in on AI strategy Qichao mentioned earlier. Revenue for the third quarter was $7.1 million, representing a $3.6 million or 102% increase from the previous quarter.

Our Q3 revenue was approximately a 55-45 split between our service revenue from our automotive OEM customers to develop AI-enhanced lithium metal and lithium-ion battery materials for EV applications and product revenue, primarily from UZ Energy's energy storage system sales. For the full year 2025, we are updating our revenue guidance to $20 million to $25 million due to UZ's contribution going forward.

Gross margin was 51% for the third quarter, which is a combination of 78% gross margin from the service revenue and 15% gross margin from the product revenue. As a reminder, with UZ Energy now part of SES, we expect gross margin variation from quarter-to-quarter as our service and product revenue mix will fluctuate. Our GAAP net loss for the third quarter was $20.9 million or negative $0.06 per share.

In Q2 2025, our GAAP net loss was $22.7 million or negative $0.07 per share. As of September 30, we had 365 million Class A and Class B shares outstanding, which were down $1.3 million from the previous quarter, mainly due to the share repurchase we executed during the third quarter. In the third quarter, we repurchased and canceled 1.3 million Class A shares for a total investment of $1.6 million or roughly $1.20 per share. We utilized $14.3 million in cash for operations in the third quarter. We exited the third quarter with a strong liquidity position of $214 million.

As mentioned, we closed the UZ acquisition in September and recognized some UZ revenue during the third quarter. We see a tremendous opportunity to grow the UZ Energy business from approximately $10 million to $15 million in projected full year 2025 revenue to a much larger growth in the coming years as we execute our go-to-market strategy that Qichao outlined to make market share gains in the $300 billion global ESS market.

When we report Q4 earnings, we expect to provide a more definitive outlook on how we see the full year 2026 revenue growth shaping up from UZ's growth in ESS, SaaS subscription use, contributions from Hisun JV and potential for the start of commercial production of electrolyte and/or battery cells from automotive OEMs, drones and robotics. The potential growth of these revenue streams, which all have different margin profiles will be much larger than what we have experienced in 2025, but the growth isn't linear from quarter-to-quarter and the margin may vary quarter-to-quarter as well.

Looking ahead, we remain focused on executing our strategy while continuing to grow our top line while remaining financially disciplined. With substantial liquidity, we are well positioned for sustainable growth and long-term success. We appreciate your continued support and confidence in SES AI.

Thank you. Now I will turn the call back to the operator.

[Operator Instructions]

And our first question comes from the line of Derek Soderberg with Cantor Fitzgerald.

On the Hisun JV, can you talk about how that opportunity came about? Was the company paying for Molecular Universe access? Or was this sort of an internal project at SES? And then what type of battery will this electrolyte enable?

Derek, so very good question. And actually, the Hisun JV came as a request by some of the Molecular Universe enterprise users. So Molecular Universe since we launched this -- earlier this year has been growing really fast. And we had almost every major battery company and battery materials company in the world trialing this. And so in addition to the SaaS platform, both on the cloud and also on-premise, several of the battery companies that are using the Molecular Universe enterprise tier also ask us, okay, we found these materials, these formulations, these molecules through Molecular Universe. Why don't you just make these and then sell these to us because they currently buy electrolytes from companies.

So it's a quite mature business model. So we said, okay, yes. we're happy to sell these materials to you. These are new formulations that they cannot buy anywhere else. And so we formed this joint venture. It's a CapEx JV we control. We control 90% of the JV, and we contract manufacture with a company called Hisun to produce this formulation, and then we sell that formulation to the cell makers. And then some of the applications, so in the call, I listed 3 and then actually all 3 are being produced.

So the most popular one is a new formulation to improve low temperature performance of LFP for ESS batteries. A lot of these LFP batteries for ESS, when they're deployed in like Northern Europe, these cold places, they don't work so well when it's a low temperature. Another one is for cell phone applications. So it's a high-voltage electrolyte for LCL cells. And then another is a 12% silicon lithium-ion cells for EV applications also to improve the cycle life.

So these 3, we are requested by the user, the cell maker to actually supply these materials at commercial scale to them. And this was discovered through the Molecular Universe platform.

And I guess just to that point, I guess I wasn't imagining a JV coming out of this first in Molecular Universe. Can you just talk about how you expect the monetization of that business to sort of play out over the next year, beyond sort of JVs, do you expect Molecular Universe to grow sort of as a traditional Software-as-a-Service business where every quarter you sort of add additional seats?

Or do you expect there to be sort of like a stair step up on revenue as you sign kind of larger agreements? How do you sort of see the monetization of MU playing out sort of over the next year? How should we think about it?

Yes. So MU is a mix of SaaS platform and materials. And then for the SaaS platform, and then we laid out the different tier pricing on the website, molecular-universe.com. And so we have the individual tiers and then the number of individual tiers is growing. And then also, we have the enterprise tiers. These are the major battery companies and then the material companies and the chemical companies. And then -- and a lot of these companies prefer on-premise. So we also sell them this Molecular Universe in a box on-premise solution.

So we charge them monthly subscription. And also we sell this computer that we actually deploy on site and also service on top of that. I think the SaaS platform, we do expect to see a growing number of seats per month and also per quarter and then the material supply because a lot of these companies eventually want us to supply the materials. But the revenue coming from the Molecular Universe discovery materials is actually going to be much higher than the SaaS revenue.

And then just one final one for me, just a broad update on Molecular Universe. I think in the past, you said you've got like 2 dozen or so companies in trial testing. Can you just give us an update on maybe where that is? And I think in the past, you have mentioned there are other potential large or medium-sized battery OEMs as part of that group. Where are you at with some of those players in negotiations? And when do you expect maybe a medium or large-sized OEM to sign on to MU through a joint development?

Yes. So the number now is getting close to 40 enterprise and then -- and they have gone through MU-0.5 trial, MU-1.0 trial, which is the latest version and it's gone through the initial cloud-based trial. And now we are planning for on-premise deployment because a lot of these medium-sized, especially the larger-sized enterprises, they can do the cloud trial, but then eventually, they will need this on-premise to actually deploy this. This is why we're moving towards this Molecular Universe in the box on-premise solution.

[Operator Instructions] The next question comes from Winnie Dong with Deutsche Bank.

I just wanted to follow up on that last question that was asked. You talked about launching 3 sub-tiers within the enterprise subscription. I was wondering if you can maybe just elaborate on that? What are they sort of looking for, what are your customers looking for? And then if you can just maybe just remind us of the other subscription options that's also bringing this recurring revenue opportunity.

Yes, so the enterprise 1, 2, 3, basically, they differ in terms of size of market database, the depth of the models used and also the knowledge and the know-how the models are trained on. For example, enterprise 1, we see that as like a PhD student level and the enterprise 2 is like the postdoc level and the enterprise 3 is a senior scientist level. And then, for example, enterprise 1, when they answer a question, they'll answer question typically less than 3 minutes and the enterprise 2 postdoc level answer question in about 5 minutes. Enterprise 3 will answer question in more than 30 minutes.

But the depth and the quality and the new discoveries are much deeper. So a lot of the enterprise 1 are medium-sized companies and also some start-ups and then the larger companies. And so for enterprise 1 and 2, we offer only cloud and for enterprise 3 and even higher tier joint development, we offer a combination of cloud and also on-premise, which most of the larger companies want.

And then in addition to these tiers, we can also do the joint development tier with the larger customers. That's where -- so now the cloud version Molecular Universe is trained only on SES internal data. But then for the joint development, we will actually put our Molecular Universe in the box encrypted and then also organize the users' data, the cell makers data and then after that I train all Molecular Universe in a box. And then we're helping them do something that they've always wanted to do, but they've never had the resource and then never had the capability to do that.

And then so this quarter, you saw some very meaningful revenue contribution from UZ Energy. So I was just curious if you can just maybe take a step back and help us sort of get reengage with the background of this company, where the markets are, what you think the growth could be from this business? And then maybe more broadly speaking, as you look out to the different revenue streams, so UZ Energy, Molecular Universe, you have the relationships with OEM partners for electrolyte production. Just maybe give us a sense of which channels you're most optimistic about as you head out to growth in the next 2 to 3 years?

Yes. So UZ Energy is an energy storage company and they serve mainly the behind-the-meter commercial and industrial applications. And before we acquired UZ, we actually used to supply this database machine learning model to UZ to help improve the accuracy of their BMS, battery management system in terms of predicting battery health and then end of life.

And then since we acquired them, now we are adding more of these machine learning-based models, which is the predict feature in Molecular Universe, MU-1.0. We are adding that into their BMS. So when we sell these -- when UZ deploys these packs to customers around the world, a lot of the customers, for example, in Norway or Northern Europe, they complain that the packs, especially in winter are just not so accurate and the BMS just stop working. So by adding this predict, we can actually -- so each UZ pack is a box, is a Molecular Universe in the box.

And then we can gather data and train the predict feature of Molecular Universe locally in the box and then make predictions about these issues before the customers can actually find out about the issues. And then -- so since we've added this predict, the amount of customer calls, their complaints actually dropped. And then we plan to continue to do that.

So this application is actually really exciting. So think of each ESS pack as a Molecular Universe in a box. And then we collect the data, we do local training and we make prediction on the safety. And then I think that's a really exciting connection. So we get data and we get revenue. In terms of the overall growth of revenue of the different businesses opportunities that we mentioned, UZ this year, revenue, we expect to pull about $15 million to $20 million. And I think next year, ESS, we can at least double that.

So this is one area. Another is drones. This year, we are getting our initial contracts and also a lot of the drones customers in the U.S. really want sales made outside of China. So we are making these cells in our Chungju, South Korea facility. Actually, the amount of capacity for pouch cells for drones outside of China is like super rare. It's hard to find pouch cells capacity outside of China. We have them. So we expect drones to also grow in a big way for us next year.

And then EV and so for example, the joint development we had with Hyundai Motors and also Honda, we are using Molecular Universe to discover these new electrolytes. And then once we identify these electrolytes, then our plan is to use the Hisun JV to contract manufacture this and then supply. So I think next year, revenue-wise, total, we should be able to at least double, if not triple, what we did this year because of all these opportunities that are enabled by the Molecular Universe platform.

[Operator Instructions] And as we have no further audio questions, I will hand back over to Kyle for the other questions.

As in past quarters, we have received some written questions from investors and time permitting, we'll go through a selection of those questions, which have not yet been addressed on the call. So the first question that we have relates to liquidity and whether management has a view on where liquidity is expected to be at the end of 2025 and will kind of scaling up of Molecular universe or the UZ Energy integration impact the cash burn or CapEx plans?

I can cover that. So -- yes, I'll take that. So our liquidity balance is very strong. And given that we closed the UZ acquisition, also, we did some stock repurchase based on our program, we still expect to exit the year with somewhere between $195 million to $200 million of liquidity. And then the other thing I wanted to probably emphasize is we are laser-focused on funding growth. So our liquidity balance is sufficient to kind of grow our revenue from different streams that Qichao just mentioned, whether it's ESS drones or materials or SaaS from MU.

So the cash balance is no longer to fund runway. We have a CapEx-light business model. So the UZ acquisition is helping us to grow revenue with positive gross margin. All of our revenue sources come in on day 1 with positive gross margin. So it's not a CapEx-heavy business. So our liquidity balance will be sufficient to fund the growth going forward.

Great. The next question relates to Molecular Universe and it goes, MU-1.0 is impressive. Can you share the road map for MU for 2026? And what other features have been requested?

Yes. So for now, MU-1.0 has been mainly focused on electrolyte materials. And then we've been requested to expand that to cover electrode process optimization and also cell design and manufacturing optimization. And then another big request from some of the major battery companies, they want their own Molecular Universe, which is quite exciting because a lot of these battery companies, they want to expand. They want to build factories overseas in different continents, and they need a battery bible, so to speak, basically put all their know-how, train this model into a portable Molecular Universe. So that is going to be a really exciting project that we'll be working on with these battery companies.

Excellent. And I think we have time for one more. So the last question is, if you could provide an update on the status on 2170 and LMA pouch cells for robotics, drones and UAM. Are you seeing good prospects for sales? Or do you see a need for scale up before promoting these more heavily?

Yes. So the pouch cells, especially for drones, we're seeing a very interesting convergence and standardization of the format in the drones industry around the 10 amp-hour pouch cells. And so that's quite exciting. And then we are converting our Chungju line, which we built both the EV line as well as the UAM line. We're converting that capacity to make this pouch cell. And so, so far, the capacity that we have meets the needs, but we're also seeing fast-growing demand, especially since it's produced in Korea. So we do see a quite exciting growth in this market.

Great. With that, I'll pass it back to the operator for closing remarks.

Thank you, everyone. This concludes today's call. Thank you for joining. You may now disconnect. Have a great rest of your day.