Ambarella, Inc. (NASDAQ:AMBA) Q3 2025 Earnings Call Transcript

Ambarella, Inc. (NASDAQ:AMBA) Q3 2025 Earnings Call Transcript November 26, 2024

Ambarella, Inc. beats earnings expectations. Reported EPS is $0.11, expectations were $0.03.

Operator: Good day, and thank you for standing by. Welcome to Ambarella’s Q3 Fiscal Year 2025 Earnings Conference Call. At this time, all participants are in a listen-only mode. After the speakers presentation there will be a question-and-answer session. [Operator Instructions] Please be advised that today’s conference is being recorded. I would now like to hand the conference over to your speaker today, Louis Gerhardy, VP, Corporate Development.

Louis Gerhardy: Thank you, Daniel, and good afternoon. Thank you for joining our third quarter fiscal year 2025 financial results conference call. On the call with me today is Dr. Fermi Wang, President and CEO; and John Young, CFO. The primary purpose of today’s call is to provide you with information regarding the results for our third quarter of fiscal year 2025. The discussion today, the responses to your questions will contain forward-looking statements regarding our projected financial results, financial prospects, market growth and demand for our solutions, among other things. These statements are based on currently available information and subject to risks, uncertainties and assumptions. Should any of these risks or uncertainties materialize or should our assumptions prove to be incorrect, our actual results could differ materially from these forward-looking statements.

We’re under no obligation to update these statements. These risks, uncertainties and assumptions as well as other information on potential risk factors that could affect our financial results are more fully described in the documents we filed with the SEC. Access to our third quarter fiscal 2025 results press release, transcripts, historical results, SEC filings and a replay of today’s call can be found on the Investor Relations page of our website. The content of today’s call as well as the materials posted on our website, our Ambarella’s property and cannot be reproduced or transcribed without our prior written consent. Before we start the call, we want to inform you of our plans to participate in the following investor events during the fourth quarter.

On December 3, we will be at the UBS Global Technology and AI Conference; December 4 at the Wells Fargo TMT Summit; on December 5, we’ll be hosting BNP’s bus tour; NASDAQ’s London Conference on December 10 and 11; Nomura’s CES Conference on January 6; and Needham Growth Conference on January 14. And of course, we hope to see you during the multiple sell-side analysts hosted tours at our CES exhibition January 7 to January 10 in Las Vegas. Fermi will now provide an update for the quarter. John will review the financial results and outlook, then we’ll be available for your questions. Fermi?

Fermi Wang: Thank you, Louis, and good afternoon. Thank you all for joining our call today. Our third quarter revenue was above the high end of our guidance range increasing about 30% sequentially in both our auto and IoT business. Company-specific factors more than offset the overall weakness in the market, with our strength originating from our customers’ new product ramps, especially those incorporating our new higher-priced AI influence processors such as CV5. We again achieved a record level of AI revenue, which in turn contributed to a higher blended average selling price. We are now forecasting fiscal 2025 revenue to increase by 22% to 24% year-over-year versus our prior estimate for revenue growth in the mid- to high teens.

Last quarter, we described our new product momentum as a series of waves and the next year in fiscal 2026, we expect the first wave from CV5 to continue and be augmented with the commitment of the second wave CV7. We expect the first and second new product waves to enable us to grow revenue again in fiscal 2026 with both auto and IoT expected to grow despite the weakness in the overall market. Our CV3 AD family of SoCs for Level 2+ and high level of autonomy represents the third wave with revenue expected to commence in calendar year 2026 or our fiscal 2027. During the third quarter, we received the first silicon of our CV3-AD655 AI SoC, which targets advanced Level 2+ applications, including mass-market passenger vehicles, and we are now delivering engineering samples to customers.

As you know, the global automotive industry is under significant pressure. So we are proud to forecast our automotive business is expected to grow this year and the next. I would like to remind you our automotive business is comprised of two different businesses. Our existing ADAS business and our central domain controller business, also known as the CV3 platform. Our existing automotive business, mostly ADAS with majority of now AI SoCs will represent about $80 million this year with an estimated five year compounded annual revenue growth rate in the mid-teens. Our CV3 platform targets a much larger but still emerging revenue opportunity. Level 2+ and high level of autonomy. This new opportunity has the potential to significantly accelerate our five year automotive revenue CAGR beyond the mid-teens CAGR I mentioned for our existing auto business.

We remain highly focused on incremental CV3 design wins in an increasingly challenged automotive market. As you know, global vehicle production growth is slow, Level 2+ market penetration remains in the low single digits and OEM projects and the software development are delayed. In this environment, we have updated our automotive revenue funnel. As a reminder, our automotive funnel represents a probability weighted estimate of automotive revenue we could generate over the next six years, from fiscal year 2026 to fiscal year 2031. At this time, our six year funnel is approximately $2.2 billion versus $2.4 billion a year ago. with one business representing more than $800 million and a pipeline of more than $1.3 billion. Due to the challenging automotive industry dynamic described earlier, there has been significant volatility in the last years within the funnel as the customer’s annual forecast was revised.

Projects will delayed or canceled, new projects were added and the projects were either won or lost. Notably, we estimate there is about $2 billion not included in the funnel beyond year six, the terminal year of our methodology. We remain optimistic about our long-term secular trend for the Level 2+ and high levels of autonomy and the role that our CV3 platform can serve in this market. We are optimistic because our CV3 platform brings solution to some of the key challenges automotive OEMs are facing today, including power efficiency, scalability and open platform with the availability of optimized software IP modules and the centralized radar. We remain diligent in our efforts to get more CV3 business into the one column. I will now discuss representative customer activity in the quarter.

In the automotive market, we highlight new model of new models featuring a variety of advanced safety and automation features. smart automotive — Automobile, a joint venture between Mercedes-Benz and Geely, introduced its smart number five model in October. This electric SUV features L2 ADA system based on our CV2 with functional safety and is supplied by Tier 1 Aptiv. Xiaopeng also on us Xpeng, an electric vehicle pioneer in China, announced the P7+, a mid to full-size electric sedan that utilize our 12 video processors for the rearview electronic mirror. This e-mirror is pre-installed 100% of the P7+ vehicles and start of production commenced in October. Also in the middle market, the joint venture between Honda and Dongfeng launched its Lingxi L electric passenger vehicle, which includes a camera monitored system.

These features include interior displays that replace the life and the right side exterior mirrors. This system is based on our CV28. Horizon auto, a Geely brand focused on development and the sales of commercial vehicles, launched its Xingzhi H8R light truck featuring from ADAS plus driver monitor system based on our CV22AQ. Turning to our IoT businesses. we are announcing the first customer for our CV server family, which represents the beginning of the second wave of new product revenue I described earlier. In the enterprise market, Verkada introduced its next generation of camera, including new 4K dorm, Fisheye and PTZ cameras based on Ambarella’s latest CV72, the new camera feature advanced analytics, including AI-powered search. Verkada also introduced a new suite of video intercom and an indoor split mini camera all based on Ambarella CV25.

Bosch announced its new FLEXIDOME 8100i, dome camera family based on CV22. They feature deep learning-based detection of persons and vehicles even in crowded or congested things. Alarm.com introduced five mega and eight mega pixel cloud IP bullet and done camera based on our CV22. The camera includes onboard recording and advanced analytics. In Japan, i-PRO, formally Panasonic Security announced the addition of 19 new models to its AeroPTZ camera list based on our CV22. We are encouraged to see better the expected adoption of our AI SoC in other IoT markets. While our products frequently target automotive and enterprise applications our AI SoCs are designed with enough programmability to drive adoption in other IoT markets. For example, in Insta360 recently introduced its Ace Pro 2 portable video camera featuring 8K video and the 50 mega pixel photos.

Based on our 5-nanometer CV5, the camera includes gesture and voice control and AI-based highlight assistance. Insta360 also introduced its link and Link 2C AI 4K webcast based on our H22 video processors. Garmin announced its GC 245 and the 255 HD met marine cameras based on our CV28 and featuring on screen distance markers and the guidance line to add with boat dockings. Grad a leading technology company based in Singapore, known for its super app, providing diversified services introduced its KartaCams 2 to collect review images for mapmaking our CV5 supports 440-mega pixel image sensors to full 360-degree viewing and provides edge AI processing. From this partial list of customer engagement this quarter, you can see we continue to build upon our well-established position for AI computer vision at the age in both IoT as well as our traditional automotive ADAS market.

In fact, on a cumulative basis, we have now shipped more than 25 million edge AI SoCs, and this helps set the table for the introduction of our new higher-value SoC supporting more advanced edge AI network such as VRM, Clip and the Gen AI. We believe the significant and continued build-out of AI training and inference capacity in data centers for more and more advanced AI network is a leading indicator for the secular growth opportunity we see for AI influence processing at the edge. Our strategic plan is well aligned with this and the first wave of new AI product revenue is underway. We expect the second wave to commence alongside the first wave next year with a subsequent wave starting calendar 2026 or our fiscal 2027 including the CV3 and our 2-nanometer platforms.

New product success is a key factor in determining our incremental revenue growth next year. We are pleased to return to non-GAAP profitability in Q3. We are highly focused on driving revenue growth and positive operating leverage on the path to our target long-term non-GAAP operating margin of 30%. We have delivered 15 consecutive years of positive free cash flow through the year of fiscal — through the end of fiscal 2024, and we are optimistic our new products can enable us to build upon this positive record. John will now discuss the Q3 results and the Q4 outlook in more detail. John?

John Young: I’ll now review the financial highlights for the third quarter of fiscal year 2025 ending October 31, 2024. I will also provide a financial outlook for our fourth quarter of fiscal year 2025 and ending January 31, 2025. I’ll be discussing non-GAAP results and ask that you refer to today’s press release for a detailed reconciliation of GAAP to non-GAAP results. For non-GAAP reporting, we have eliminated stock-based compensation expense along with acquisition-related costs and restructuring expense adjusted for the impact of taxes. For fiscal Q3, revenue was $82.7 million, above the high end of our guidance range, up 30% from the prior quarter and up 63% year-over-year. Non-GAAP gross margin for fiscal Q3 was 62.6% at the low end of our prior guidance range due to product mix as we opportunistically drove some revenue upside from certain legacy processors at lower-than-planned margin.

Non-GAAP operating expense was $49.1 million, about $900,000 lower than the midpoint of our prior guidance range, driven by continued expense management and the timing of spending between quarters. We remain on track to our internal product development milestones. Q3 net interest and other income was $2.1 million. Q3 non-GAAP tax provision was approximately $200,000. We reported a non-GAAP net profit of $4.5 million or $0.11 of earnings per diluted share. Now I will turn to our balance sheet and cash flow. Fiscal Q3 cash and marketable securities increased $6.7 million from the prior quarter to $226.5 million. Receivables days sales outstanding increased from 33 days in the prior quarter to 38 days and days of inventory decreased from 108 days to 94 days.

Capital expenditures for tangible and intangible assets were $2.5 million in the quarter and $6.2 million for the nine months ended October 31, 2024. We generated positive operating cash flow of $6.6 million in the quarter and $8.4 million through the first three quarters of fiscal 2025. Free cash flow in the quarter was $4.1 million, with year-to-date free cash flow of $2.2 million. We had two logistics companies representing 10% or more of our revenue in Q3. WT Microelectronics, a fulfillment partner in Taiwan that ships to multiple customers in Asia, came in at 66% of revenue for the third quarter. Chicony, an ODM, who manufactures for multiple end customers was 11% of revenue for the quarter. I’ll now discuss the outlook for the fourth quarter of fiscal year 2025.

The continued strength of our customers’ new product ramps, especially those enabled by our new product wave one from our 5-nanometer CV5 caused us to increase our Q4 estimate. We are expecting normal seasonal decline in Q4 following the stronger-than-expected Q3. Fiscal Q4 revenue is expected to be in the range of $76 million to $80 million, with IoT and auto, both flat to slightly down sequentially. We expect fiscal Q4 non-GAAP gross margin to be in the range of 61.5% to 63%. We expect non-GAAP OpEx in the fourth quarter to be in the range of $49 million to $52 million with the increase compared to Q3 driven by CES marketing activities, increased head count and project-related engineering expenses. We estimate net interest income to be approximately $1.8 million, our non-GAAP tax expense to be approximately $600,000 and our diluted share count to be approximately 41.8 million fully diluted shares.

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

Q&A Session

Operator: [Operator Instructions] Our first question comes from Ross Seymore with Deutsche Bank. Your line is open.

Ross Seymore: Hi, guys. Congratulations on the strong results and guide. I guess my first question for me is really what changed? I know you talked about the different waves and when they’re coming, but the inflection point, growing maybe 5% above the midpoint of your range, but an impressive 30% sequentially. And I know you called January seasonal, but it seems like it’s even better than seasonal. So what’s the activity at your customers that’s changing? And I guess what I’m really getting at is I understand it’s the new product adoption, but is there also just kind of shipping closer to end demand. So is this the beginning of secular growth? Or is it also bolstered by a cyclical rebound?

Fermi Wang: Thanks, Ross. I think, first of all, Q3 result, definitely, there is — this first quarter, we rebound from the inventory correction. So when we give — talk about last quarter, we did talk about the inventory correction will be done in the middle of Q3. And with that — after that, it will be our growth of new products. And the Q4 is all about new products growth right now. And in fact, that if you look at our run rate of CV5. If we can look at two different angle, one is from the market angle that for both auto and IoT growing and for IoT is really growing and the IoT enterprise and as well as IoT other, these two categories, they both grow in a similar rate and driven — the major driver is that new product cycle for CV5.

In the automotive market that you see that we talk about new customers like Sensera with CV2 and Rivian with CV5, I think those are definitely helping us to have a growth on the automotive side. So I have to say that maybe Q3, we rebound from the inventory correction and assessing growth on new product line in Q4 is all secular about our growth. But I also want to point out our Q3 is at $82-point something million revenue and the Q4 guidance in the midpoint $78 million. So we’re still seeing some kind of seasonality for Q4.

Ross Seymore: Whatever you’re doing, it seems to be working. So congrats on that. I guess for my follow-up, switching over to the automotive funnel. It seems a little bit odd that the number would go down year-over-year considering you’re adding a year where the growth rate at the tail end should be larger. So I guess the big picture question is, to the extent over the last couple of years, we’ve kind of thought as your auto business was going to be the incremental driver of significant growth going forward. Now it seems like the IoT, edge AI, whatever you want to call it, seems to be the bigger driver. Can you just talk about has the growth profile of the company really switched more to the IoT side? And are you as optimistic as ever on automotive? Or is something different the appropriate interpretation we should have?

Fermi Wang: I think we’re still very optimistic about CV3 domain controller opportunity in the market. And I would say that if you look at the number, our 1 count continue to grow, but our pipeline is reduced. The reason for that is really that we — first of all, the overall market is weak, clinically automotive market. And I think that you can give a similar feedback from everybody. But I think the most important thing for me is the Level 2+ adoption rate is slower than we expected, right? We start seeing — we expect to see a lot of engagement. We continue to see a lot of engagement. But a lot of the projects are pushed out into the schedule for different reasons and some projects even got canceled. So I really think that Level 2+ adoption is not as fast as we thought.

That’s probably the reason we start seeing that our funnel is not increasing. However, I still believe that Level 2+ will become an important automotive market for everybody and that will replace the current Level 1 and Level 2 solution in the market today.

Ross Seymore: Thank you.

Operator: Thank you. Our next question comes from Tore Svanberg with Stifel. Your line is open.

Tore Svanberg: Yes. Thank you. And let me echo and congratulations on the strong results. For me, could you just give us a little bit better sense for the mix here between CV2, CV5. I mean I assume not a whole lot of CV7 revenue yet. I think on the last call, I think you talked about CV5 potentially reaching 1 million units this year. So any more color you can share with us on the mix of CV would be helpful?

Fermi Wang: Right. So maybe let me put some data together so to give you some point how to think about this. First of all, we talk about our AI revenue is roughly 70% in this quarter. And also, we talk about our CV5 easily going to ship more than 1 million units who are probably north of that by a margin. So that CV5 doing well. So — and our CV5 ASP is anywhere between $25 to $50. That gives you an idea of our CV2 and CV5 contribution. And also I want to point out, in Q3, this is the first quarter in the last three years that our video processor grow. And that’s — I think that’s really because that inventory rebound — inventory correction finished the customer rebound from there help the video processor has a growth in Q3, but we expect that you will go back to decreasing — gradually decreasing in the coming quarters.

Tore Svanberg: Great. And that was actually going to be my follow-up question. So as we think about fiscal 2026 and we think about the cyclical recovery and so on and perhaps even impact the gross margin. So video processing bouncing back, it’s probably more of a temporary phenomenon, and you’re not really expecting that to continue to drive a higher mix to fiscal 2026?

Fermi Wang: That’s correct. We haven’t given any guidance on the fiscal year 2026, but we do believe that in fiscal 2026, our both IoT and auto well growth, and we’ll provide more details about the gross margin and the OpEx in the next conference call.

Tore Svanberg: Sounds good. Congrats again from me.

Fermi Wang: Thank you.

Operator: Thank you. And our next question comes from Quinn Bolton with Needham & Company. Your line is open.

Quinn Bolton: Hi. Let me offer my congratulations again. I wanted to ask for me, just kind of coming back to the auto pipeline, maybe sort of a follow-up to Ross’ question. You talked about some pushouts and maybe even cancellations in Level 2+. Can you give us a sense what percent of that pipeline is now sort of driven by the Level 2+ opportunities and how that might have changed from the last year’s pipeline?

Fermi Wang: Right. So maybe I’ll give you a high-level description. I think in the pipeline, we have one color and also — sorry, for has one color and the pipeline color. And in one color, I would say CV3 percentage is below 50%. But in the pipeline color, CV3 domain controller is well above 50% in that pipeline. So in terms of Level 2+, maybe I’ll add a little more color on that. We believe that when we look at the current price delta between Level 2+ and Level 2 and Level 1, we think the price delta is still high, and that really keep OEM worried about introducing a brand-new product in that price point. So I think a lot of OEMs are thinking about how to optimize the price and for introducing a better Level 2+ function features.

That’s something where we think we can help because we keep telling people that, first of all, I think our bond saving for our OEM is significant compared to our competitors. More importantly, our software can be easily adapt from a high-end Level 2+ to lower Level 2+ products by really with a simple modification. And that software compatibility from the low end to high end Level 2+ will save a tremendous amount of R&D cost to our customers. So I think we are trying to address the pain point our customers when they’re trying to — the reason of today’s Level 2+ cause activate, one is the cost, the other one is really the software development. That’s why we think we have a solution that we can address the pain points of our customers.

Quinn Bolton: Perfect. And then the second question is just kind of regarding the automotive pipeline. Geographically, how diversified is that pipeline? Is it pretty concentrated in China or another geography? Or do you see pretty good geographic distribution of that pipeline? Thank you.

Fermi Wang: I think the distribution is pretty fair. In fact, a lot of people think we have a high concentration in China who is wrong. I would say that 15% of our pipeline is from China. And from that, you can see we had probably a little higher percentage in Europe. And that’s probably the — and everything else is probably well distributed.

John Young: And Quinn, I might add on top of that, our pipeline methodology is really it was started and designed to focus on L2+ opportunities. But the pace of adoption in China and the design cycles, are very quick relative to the rest of the market. We kind of pegged our six year cycle to try and be a good proxy for a Western design cycle for models and for programs. And so our six year funnel I think may not show all of the opportunity in China because those programs in China are quicker. So when Fermi says that the funnel has approximately 15%, that’s another factor to consider.

Quinn Bolton: Got it. Thank you.

Operator: Thank you. Our next question comes from Christopher Rolland with Susquehanna. Your line is open.

Christopher Rolland: Hey, thanks for the question. Yes, similar to the last one, can you talk about your current geographic mix and why this might have led to outperformance versus some of your peers, those with, I guess, EU exposure in particular? And then just tying into this as well, there’s a worry that there might be Chinese EV ship ahead in front of tariffs, both the EU and U.S. Do you think that, that’s going to come back to hunt us all in the auto segment at all? Or do you think you’re pretty clean from that perspective?

Fermi Wang: Right. So first of all, that exposure I talked about while answer Quinn’s question is about the distribution of our pipeline. It’s not about distribution of current revenue. That distribution of current revenue is quite different than the pipeline. So that if I understand your question, you’re asking whether we have any — a lot of exposure on different geographic locations particular focus on our current revenue. I would say majority of our revenue coming in the U.S., although it’s manufacturing in Asia, but the end market, a big portion is the U.S. And then there’s some European market then Japan, Korea are all very big. And in China, okay, also 15% of total revenue exposure in China. So that’s how our current revenue distribute based on the geographical locations.

And then you asked about the geopolitical situation. My gut feeling is if that situation change Ambarella is not going to be the only one. Maybe we’ll probably have some impact but not the biggest impact. But however, this is going to change if the geopolitical situation continue to get worse, which I think well, it really depends on how much more tighten the rule going to be. And the worst situation, of course, that the whole supply chain gets separated that U.S. component cannot go to China and vice versa. Then then we’re facing a total different environment and then we probably need to write off our 15% total Chinese revenue. But I will say that’s not just impact to Ambarella. It’s impact to the whole industry. For us, I don’t think there’s anything particularly only target for Ambarella in terms of geopolitical risk.

In the past, we had a lot of risk with Hikvision and Dahua that was four years ago, they’re way past us, and our current revenue exposure is 15%, half automotive, half IoT. So I would say that we don’t have a very significant geopolitical risk target on Ambarella only.

Christopher Rolland: Thank you for that Fermi. And then you mentioned something about in the prepared remarks about a legacy processor, I think that continued to sustain and maybe that weighed on margins. I wasn’t quite sure there. But maybe talk about that and then when we could get a margin lift in particular in next fiscal year and what the moving parts there would be for that gross margin lift. Thanks.

Fermi Wang: Yes. Let me point you to the direction that, for example, we did mention our video processor revenue grew in Q3, that was the first time for the last 10 years. Although when we talk about gross margin, we talk about our what we use the…

John Young: Legacy.

Fermi Wang: Legacy product. So video processor is part of that, and there are other processors in there. But however, you can see that definitely, that’s one reason we start feeling a little pressure on the gross margin side. But I will say that the gross margin is really about mix. Every time when our mix change, usually, you will see our gross margin move up and down a little bit. But that like what we have been saying for quarters, we continue to believe that our gross margin will move gradually into our long-term gross margin model, which is 59% to 62%, and it will over time to get there.

Christopher Rolland: Excellent. Thank from me.

Operator: Thank you. Our next question comes from Kevin Cassidy with Rosenblatt Securities. Your line is open.

Kevin Cassidy: Yes. Thanks for taking my question. Maybe a slightly different subject. But can you talk more about your 2-nanometer development when you expect tape out and the target end markets for those products?

Fermi Wang: Yes. Thank you. First of all, we only kick off 2-nanometer projects and on our engineer road map, and we have engineers working on it. We expect the first 2-nanometer chip will tape out, let me say, Q4 next year in that period of time. And of course, the first ship target is for IoT and our enterprise IoT as well as IoT other section will benefit from this. But of course, that our 2-nanometer is also very important. You should consider 2-nanometer as a family chip just what we did with 5 nano and 10 nano. And for 2-nanometer chip, it’s important for us to consider build up our new technology for — to address the need of new AI platforms and new AI workload like Gen AI and other type of transformer network, and we definitely will upgrade our architecture based on our 2-nanometer process node.

Kevin Cassidy: Okay. Great. It sounds exciting. Maybe just from John’s point of view, as the OpEx is all the spending for 2-nanometer as you go through the fiscal year 2026.

John Young: Yes. I mean, we haven’t given guidance for 2026 for OpEx yet. But as Fermi said, we’re already amortizing the costs for the 2 nano project. So I think OpEx — I would expect OpEx to increase as just absolute number next year, but the 2 nano project is already baked into the run rate.

Kevin Cassidy: Okay, great. Thanks. Congratulations.

Fermi Wang: Thank you.

Operator: Thank you. Our next question comes from David O’Connor with BNP Paribas. Your line is open.

David O’Connor: Yes. Great. Thank for taking my question. Maybe two from my side, if I may. Just first, you guys on the auto funnel, again, going from that $2.4 billion down to $2.2 billion. Can you kind of break that out? How much was kind of canceled and then how much was added back in just to give a sense of kind of the relative size of those cancellation. And also kind of on those cancellations, anything you can share geographically of where you saw most of those cancellations. And I have a follow-up.

Fermi Wang: Right. So I think the right word to describe that chart, if you look at all of the decrease and increase is volatile, right? And in fact, is volatile for reasons that, first of all, a lot of our customers pushing out the project and some of that reduced the forecast. But definitely multiple projects get canceled. The most of the consolidation happens in Europe in U.S. And obviously, there is definitely project that can because software reason of other transactions that impacted the road map of the company. So I think that definitely is clearly a big portion of the movement of our pipeline. But I would say that we add a project in there also based on the engagement of Level 2+. So that’s a plus and minus. But one color I want to make it clear that, of course, we have new design wins in there that we already announced their design wins we haven’t announced.

But they also — there’s — the downside is also mainly for a lot of our project that we won last year and the forecast continue being reduced. Some is reduced by 10%, some reduced by even larger margin. So that’s a combination that we’re dealing with in our pipeline.

David O’Connor: Thanks for me. That’s very helpful. And just one follow-on. Just on China again, you mentioned — I think John mentioned it 16% of sales today. And also you said that kind of in the pipeline, it’s 15% as well China — but then China is innovating a lot faster. So can you just go back and explain why China isn’t higher in the mix? Or where should China land in kind of a steady state in terms of the mix really? Thanks.

Fermi Wang: Well, first of all, I think John just tried to explain that even we won a design win from China because the design cycle usually two to three years instead of five to six years. So the impact — every Chinese design win doesn’t matter it’s one in pipeline, you only occupy three out of six years of that funnel. So that’s why just from that point of view, China represents smaller than the reality because all the other projects like U.S. or Europe or even Japan, Korea, usually, any design win is five to six years of pipeline. So I think that’s one reason John tried to say is because of that two to three year design cycle, you make China number smaller than it should be.

David O’Connor: Thanks guys.

Operator: Thank you. Our next question comes from Suji Desilva with ROTH Capital. Your line is open.

Suji Desilva: Hi, Fermi. Hi, John. So on the L2+ us win do you have, can you talk about the competitive landscape and your design win share across those? And I’m curious how impactful the Conti and Bosch partnerships are in helping you secure those wins?

Fermi Wang: Yes. So in fact, the competitive landscape didn’t change. Outside China, in fact, the worldwide is NVIDIA, Qualcomm and Mobileye and us, right, the four company. And in China, you have to add the Horizon Robotics in there. So I think those are probably the company we’re competing against. And I think Conti and Bosch continue to play a major role because right now, with all the OEM design wins outside China, OEMs still want to work with a Tier 1. Our role to OEM is really providing silicon and sometimes providing software, but they always need a Tier 1 sitting in between. So that Conti and Bosch always play a major role because they are bidding on the design win. We are helping them to work, to bid on those design win. So working closely with the Tier 1 continue to be important for us.

Suji Desilva: Thanks for me. And then my other question is on the customers and helping kind of adoption. You said customer software readiness is one of the factors in the timing of adoption. Is there anything you can do from your end to help that equation, help them speed that cycle along with software rates? Or is that something you just have to have the customer do then the ready to adopt you?

Fermi Wang: Absolutely. That’s one thing we really focus on to do. First of all, for example, we already introduced our software stack to many people. And more importantly, we already show some of our important customers that using our software stack is design for Level 4, we can easily in less than two months adopted to the Level 2+ type of sensor configuration. So that a quick updation from Level 4 to Level 2+ really help customers to understand using our software, they can easily — and our software, of course, it can easily move from Level 2+, Level 3 and Level 4. And that is important because that really significantly reduce their software investment. That’s one. Two, we also have a business model to enable our customer by licensing whatever software module they think that we — that they can leverage.

For example, we knew that we are doing well quite well on the perception side, and we definitely believe that we are one of the few really demoing using HD map to do a perception and driving. And those kind of function features are very welcome. So any — we are helping — if any customer wants to use those function features, we are open to license and helping them to integrate to their other software stack. So I think the combination that we have our own software that is scalable from the Level 2 to Level 4 and also we are willing to license IP that can help our customers. I think those are two areas we can definitely help our customers to speed up the software development.

Suji Desilva: Okay. Thanks.

Operator: Thank you. Our next question comes from Martin Yang with OppCo. Your line is open.

Martin Yang: Thank you for taking the question. Only one question regarding the automotive pipeline change. Given where you’re seeing with European customers and U.S. customers, does it change your outlook for potential margin contribution from those automotive design wins?

Fermi Wang: Yes. In fact, we talk about how we go into production with CV3, our gross margin definitely is going to be in the low end of our long-term gross margin model. So I think that’s because we are competing with the largest possible semiconductor company out there and we’re expecting a steep competition. So that definitely will change. But however, in the short term, I continue to believe that we will maintain our current gross margin model, which is 59% to 62%. And although we are running above it for many quarters now, but we think we’re gradually moving down back to that range.

Martin Yang: Got it. A quick follow-up. So again, on broad picture company-wide margin. Do you think — where do you think — what factors would drive the margins? Do you think it’s mixed or any other factors you would point out that have a bigger influence on the long-term margin outlook?

Fermi Wang: Yes. So if you look at only short-term quarter-to-quarter, mix is only reason. Every quarter when our product mix that we sell differently that change our gross margin. That you see that in Q3 gross margin go to low end because of mix. But if you look at long term, I really think that before we hit CV3, we continue to move to — our strategy is to always try to sell value to our customers. . And when we move to 5-nanometer, people ask what you can really continue to maintain your gross margin model with higher cost of 5-nano, and we prove that we can. So I think for our current business particularly current IoT and automotive business, we are quite comfortable about our guidance on the gross margin profile. But I think CV3, you since have time to work on it to work on it, to work with our supply chain and so on.

So I will say that I think although we guided lower than our current gross margin — to the low end of our gross margin model, but I still think we have time to work on it.

Martin Yang: Got it. Thank you, Fermi.

Operator: Thank you. Our next question comes from Richard Shannon with Craig-Hallum. Your line is open.

Richard Shannon: Thanks for taking my question as well. There haven’t been a lot of discussion on Gen AI in this call. So maybe, Fermi, if you can talk about the progress here since you announced this initiative, I mean, just over a year ago. How that’s building out here. Maybe if you want to use it in the framework, you’re talking about your funnel in the automotive space, talk about it to Gen AI. I guess I’d love to hear about that.

Fermi Wang: In the last four quarters, I think it’s become very clear to me and to the company, Gen AI is going to impact not only just outside our current market. In fact, overall current market will be impacted by Gen AI in a different way. In fact, all our customers are looking type of Gen AI model going to impact their business. So it’s important for us to understand what our customer wants and engage with them as early as possible. So we engage with them start on, I would say, six quarters ago, and we show them what we can do with start with N1, I would say, five, six quarters ago, and we show them what we can do with N1 type of performance, and we’re still working with several customers on the POC and potentially products.

And at the same time, when we realize that our customers — in fact, our customer — current customer told us they need to start looking at Gen AI and looking at the clip type of function or other type of large language model to help their business, we start looking at CV72 or CV75 type of product, how does that support our current customer. And frankly, we are very pleased to find out that the chip, CV72 and CV75 we define for the enterprise security business can be used to run a large language model, although it’s not large, but it’s a 3 billion parameter clip type of new network and any derivative of that. So it definitely helps our customers to start looking at how the clip can run on this kind of edge device in the camera. We’re talking about 5-watt chip running a 3 billion parameter, which is very difficult to find in the market today.

And more on top of that, in fact, a lot of our — the traditional security camera service provider, for example, those consumer or IoT home supplier like Ring, like Nest, they just announced that they’re going to use Gen AI type of service. In fact, they all announced running clip type of neural network on the server, and they charge a customer $10 a month. And we think that we can enable similar service at edge and significantly reduce the cost of enabling that service. So I think we start seeing a new trend that Gen AI is not just for large language model for open AI type of company. In fact, it will come down to apply to all the major — our current customer and maybe even other customers we can serve with. So our talk about Gen AI strategy.

First of all, we continue to use N1 and CV7 family to engage customers. We believe we’re going to see some of our customers putting a clip type of network onto our CV70 family camera sometime next year, and that will be our first revenue from CV7. And we believe 2026, we’re going to see some revenue as we expected. But more importantly, I think to address this Gen AI platform, I think our 2-nanometer process is going to play a major role because we believe to solve the Gen AI in an appropriate architecture, we need to go to the most advanced node. Of course, we need to solve not only just the processing performance, but also DRAM bandwidth. But I think we do have a plan to address both.

Richard Shannon: Okay. A lot of detail there for me. I’ll follow up with that one a little bit later. Thanks for all that. My follow-on question here is — so obviously, as you just detailed, I don’t know a prior question here about the competitive dynamics here, competitive environment, where you’re the smallest company out there and probably later to market. What kind of with all the pushouts that we’ve seen in the automotive space as exemplified in your funnel change year-on-year here, do you think this is going to end up being a net positive for you, allowing you to catch up in any manner? Maybe just kind of discuss how these — the changes in the environment are going to be beneficial for you?

Fermi Wang: Well, first of all, I think the pushback, I said the pushout is because, one, the price need to be right, right? Two, the software need to be ready. And both of them — we are — I think the market or the current — the OEMs are definitely looking for both of that and Ambarella definitely have a solution for both. Our bond costs, our lower power consumption, so we can have a much lower cost on the battery and power dissipation solution. That definitely helps. The other one is software solution that we think we have a scalable software solution that can easily scale from the Level 2 to Level 4. I haven’t seen — we definitely believe that also is helpful. So I won’t say that the delay is going to help us, but I definitely think that we can — we have a solution for the reason of the delays.

Richard Shannon: Very helpful. Thank you, Fermi.

Fermi Wang: Thank you.

Operator: Thank you. Our next question comes from Gus Richard with Northland Capital Markets. Your line is open.

Auguste Richard: Yes. Thanks for taking my question and congratulations on the strong results. Just going back to the AI consumer cameras, how much of an ASP uplift would that provide you guys as that capability rolls out?

Fermi Wang: Right. So first of all, today, our average ASP, the company-wide average ASP, I would say, is around $12, $13 and it continue going up and going up because our CV2 — for example, our CV2 family ASP is around $18, $19. Our CV5, like I said, is anywhere between $25 to $50. And the CV7 is probably somewhere between — CV7 family, let me make clear, CV75, CV72 is anywhere between low — high teens to probably $30, $40. So you can see that the trend is definitely we are building more capability into our chip for AI performance and therefore, driving up the ASP. So we continue to expect our ASP growth will be there.

Auguste Richard: Got it. And then just given the change in administration, securing the border, deporting a bunch of people. I would imagine that the demand for security cameras is going to increase. And I’m just wondering at this point, are you seeing any uptick from your enterprise customers or potentially government entities?

Fermi Wang: We definitely see that the IoT enterprise continue to grow. In fact, we always said that we believe this year, next year, IoT enterprise continue to grow in a healthy way for us. I think that might reflect what you said. But I definitely think that the current overall environment will continue to drive security camera growth.

Auguste Richard: Got it. Thanks so much.

Fermi Wang: Thank you.

Operator: Thank you. I’m showing no further questions at this time. I would now like to turn it back to Fermi Wang, President and CEO, for closing remarks.

Fermi Wang: And I would like to thank everyone to participate today and looking forward to see you on the different road shows and our CES. Thank you very much.

Operator: This concludes today’s conference call. Thank you for participating. You may now disconnect.