Hello, everyone, and welcome to Ambarella’s First Quarter Fiscal Year 2026 Earnings Call. — Operator Instructions — Please note, this event is being recorded. Now it’s my pleasure to turn the call over to the Vice President, Corporate Development, Louis Gerhardy. The floor is yours.

Thank you, Carmen. Good afternoon and thank you for joining our first quarter fiscal year 2026 financial results conference call. On the call with me today is Dr. Fermi Wang, President and CEO; and John Young, CFO. The primary purpose of today’s call is to provide you with information regarding the results for our first quarter fiscal year 2026. The discussion today and the responses to your questions will contain forward-looking statements regarding our projected financial results, financial prospects, market growth and demand for our solutions, among other things. These statements are based on currently available information and subject to risks, uncertainties and assumptions. Should any of these risks or uncertainties materialize or should our assumptions prove to be incorrect, our actual results could differ materially from these forward-looking statements. We are under no obligation to update these statements. These risks, uncertainties and assumptions as well as other information on potential risk factors that could affect our financial results are more fully described in the documents we file with the SEC. Before starting the call, I’d like to summarize our investor events scheduled for our second 1 fiscal quarter. On June 3, we’ll be participating in Bank of America’s Technology Conference in San Francisco; June 17th, we’ll host Redburn Atlantic’s West Coast Bus Tour in Santa Clara; June 18th, we’ll host Trivariate Research and NH Investment Securities Bus Tour in Santa Clara; and June 23 to 25, we’ll be visiting Baltimore, Boston and New York City on a non-deal roadshow. Access to our first quarter fiscal year 2026 results press release, transcripts, historical results and SEC filings as well as a replay of today’s call can be found on the Investor Relations page of our website. The content of today’s call as well as the materials posted on our website are Ambarella’s property and cannot be reproduced or transcribed without our prior written consent. Fermi will now provide a business update for the quarter. John will review the financial results and outlook, then we’ll be available for your questions. Fermi?

Thank you, Louis, and good afternoon. Thank you for joining us for our call today. We had an excellent start to the year with the first quarter revenue of $85.9 million, to the upper half of our guidance due to the continued strength in our edge AI business. Both our 5-nanometer CV5 and CV7 product family as well as our 10-nanometer CV2 product families contributed to the revenue growth, and our average selling price continued to increase as we capture more value per design win. Edge AI revenue, which we define as a product that integrates one of our proprietary deep learning AI accelerators, was more than 75% of our Q1 revenue. And this represents the fourth consecutive quarter of record AI revenue. This achievement demonstrates the execution of our edge AI strategy in the face of a volatile market. During the first quarter, IoT applications increased to mid-single digits sequentially and now represent about 3/4 of our total revenue with our automotive business declined low single digits sequentially, 2 although automotive revenue was up more than 20% on a year-over-year basis. In our February 26 earning call, we provided a fiscal 2026 revenue growth estimate in the mid- to high teens or approximately $327 million to $339 million with some conservative built in to our second half outlook due to the geopolitical uncertainty. Although the geopolitical uncertainty remains high, we are increasing our fiscal 2026 revenue growth estimate to the range of 19% to 25% or approximately $348 million at the midpoint. While we continue to expect that we will not face a material direct impact from the current tariffs, given the uncertainty of an indirect impact, our larger-than-normal range of guidance reflects our conservatism. We are confident in the long-term drivers of our edge AI strategy, and the business remains fully intact. Multiple factors are driving our optimism for the edge AI market, in- cluding my recent discussions with customers, representative customer engagements, I will discuss later in the call as well as evolution, evolution of our edge AI serviceable available market. Our SAM is comprised of more than 20 different automotive and IoT edge AI applications with a 5-year compounded annual revenue growth rate in the high teens, reaching almost $13 billion in fiscal 2031. In the past, a vast majority of our revenue and our SAM opportunity originated from the edge endpoint market, all the terminal device in the network. But our new analysis indicate SAM expansion in the edge infrastructure or the layers of – the layer of a network where data from multiple endpoints is aggregated and the incremental advanced AI features and services such as multimodal vision language model – vision language and reasoning models can be supported. We are already addressing the edge infrastructure market with the N1 family, and we are now developing a new AI SoC product family to enhance our edge AI infrastructure roadmap by leveraging the silicon architecture and software investment in low-power and scalable third-generation AI 3 accelerator, we are able to efficiently extend our reach. Over the next few quarters, we will be describing some of the edge infrastructure applications we are targeting in more detail. In April, I attended the ISC West security show and met with key customers and partners. Ambarella demonstrated its leadership in GenAI at the edge with 18 product demonstrations, including the latest GenAI and the vision AI capabilities. And I was very pleased with the high level of customer interest and activities around our advanced AI products. The demonstrations highlight Ambarella’s ability to enable scalable, high-performance reasoning and vision AI applications, leveraging our third-generation AI accelerator, which now supports most of the leading GenAI models from 500 million to 34 billion parameters. We demonstrated Deepseek GenAI models running on our products at 3 different price performance points, including CV75, CV72 and N1-655 processors. These demos, including advanced multi-stream video analysis, exemplify our – how we are pushing the bound- aries of real-time AI-powered security analytics by running state-of-the-art vision language models for both endpoint and on-prime infrastructure. I will now talk about some of our customer product introductions during the quarter. During the quarter, a leading enterprise security camera company introduced two new products based on our CV72 offering very high resolution and advanced AI analytics. A third product, a wearable device supporting multiple modalities, was also introduced to the market. And this first of client product is based on our A6 LLM video processor. As I described earlier, we are seeing increased traction in the edge AI market beyond our core enterprise and whole security business. This quarter, IoT edge examples that demanded – demanding our AI technology, including 360-degree portable video cameras, cyclist cameras, industrial automation and enterprise video conferencing. 4 In the portable video camera market, market leader Insta360 introduced its flagship model, the 360° X5 camera based on Ambarella’s 5 nanometer CV5. The X5 offers 8K video with advanced AI-based image processing. Also in the IoT market, Garmin announced its Varia Vue headlight camera for cyclists based on our H32 video processors. In the enterprise IoT industrial automation market, Huawei announced its R5000 series of machine vision code readers based on our third-generation AI accelerator. The 5-nanometer CV75 enables the system to read up to 90 codes per second. Also in the enterprise IoT, Norway-based Huddly introduced a new category, multi-camera video conferencing products, as an integrated system in Europe exhibition. Huddly’s new C1 video bar is part of a collaboration with technology giant, Lenovo. And the advanced system is based on our 5-nanometer CV72 with 20x the AI performance of previous generation systems. In our automotive safety and ADAS business, this quarter, we are disclosing wins in China, Japan, Korea and the U.S. A leading Japanese OEM will utilize our CV25 SoC in a data log application with a CV25 supporting both secure viewing and data analytics. The project is supported by a major Tier 1 in Japan with production scheduled to begin this year. Also in Japan, another leading Japanese OEM is utilizing CV25 for a multi-camera system providing in-cabin recording and viewing functions with production scheduled for our current fiscal year. During the second quarter, Zeekr introduced its 007 GT electric vehicle featuring an interactive intelligent B-pillar system with two cameras. Our CV28 enable access control based on face ID as well as incoming monitoring. Thinkware, a leading South Korea provider of smart car information technologies, has entered production with Harmony, a dual camera recorder based on CV25 supporting ADAS features such as forward collision warning, then departure warning and security monitoring. 5 And in the commercial fleet telematics market, U.S.-based RoadEazy introduced its RZ1 featuring a dual-view camera based on our CV25. The RZ1 capture clear road and cabin footage to improve fleet safety and accountability with integrated edge AI, identifying risks like distracted driving, phone usuage, or tailgating. As you can see from this representative engagement, security remains an important growth market for us. But we are seeing opportunities in numerous other edge AI applications with customers in both the auto and IoT market evaluating and adopting our AI SoCs. As many of you know, roughly 5 years ago, edge AI originated in an enterprise security camera market, and we were quick to lead the market. Today, we continue to lead the security edge AI market, and we are successfully leveraging our AI portfolio and the market know-how into new application verticals. In fact, security is less than half of our total revenue today, and today’s announcement are just a subset of new edge AI application we see emerging. Our investment in technology and products is driving today’s revenue growth and our future revenue growth opportunities. Our edge AI products address the megatrends of safety and security, but also automation which enables end-user productivity to be improved and enables entirely new revenue streams across many markets. While it is still early, AI is finding its way to the edge is not just a data center or hyperscaler opportunity anymore. Ambarella is the leader in edge AI with more than 32 million edge AI processors shipped on a cumulative basis. We are the established edge AI technology provider who is uniquely focused on – focused and positioned for the rapidly evolving edge AI market. We continue the pace rapid innovation. Our product portfolio and road map are highly differentiated and offer the flexibility and scalability to target increasingly diverse applications. Both enterprise and consumer-driven markets and across edge endpoints as well as edge in6 frastructure. As I wrap up today, I want to reiterate the important points we shared today. One, we delivered strong Q1 results with similar strengths projected into Q2. Two, we increased our fiscal 2026 guidance while maintaining a conservative second half steps. Three, our high-value, high-ASP products are seeing strong momentum. Fourth, we have a strong SAM outlook with the new edge AI markets in development. Five, we are an established edge AI market leader who’s innovating at a rapid pace. Of course, the geopolitical uncertainty can be a distraction, but to deal with it, I feel it is important to remain agile and be prepared for short-term surprises and to focus on what we can control while most importantly continued investment in innovation and market development that is most critical for our success. Financially, while we have generated positive free cash flow for 16 consecutive years, our goal is to develop the technology products and the customers that result in positive earning leverage and growth in our free cash flow. With that, John will now discuss the Q1 results and the Q2 outlook in more detail.

Thank you, Fermi. I’ll now review the financial highlights for the first quarter of fiscal year 2026, ending April 30, 2025. I will also provide a financial outlook for our second quarter of fiscal year 2026, ending July 31, 2025. I will 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 adjusted for the impact of taxes. For fiscal Q1, revenue was $85.9 million, above the midpoint of our prior guidance range, up 2.2% from the prior quarter and up 57.6% year-over-year. Sequentially, automotive 7 revenue declined in the low single digits and IoT increased in the mid-single digits. NonGAAP gross margin for fiscal Q1 was 62%, slightly above the midpoint of our prior guidance range due to a favorable product mix. Non-GAAP operating expense in Q1 was $51.8 million, slightly above the midpoint of our prior guidance range of $50 million to $53 million, due in part to higher engineering costs on new and existing chip development projects. Q1 net interest and other income was $2.2 million. Comparing to our prior guidance of $1.8 million, the increase was primarily from higher other income. Q1 non-GAAP tax provision was approximately $600,000. We reported a non-GAAP net profit of $3 million or $0.07 of earnings per diluted share in Q1. Now I will turn to our balance sheet and cash flow. Fiscal Q1 cash and marketable securities reached $259.4 million, increasing $9.1 million from the prior quarter and $56 million from the same quarter a year ago. Increased cash and marketable securities benefited primarily from working capital improvements associated with increased revenue during the quarter. Receivables days sales outstanding decreased from 33 days in the prior quarter to 31 days, while days of inventory increased 1 day to 98 days. Compared to the prior quarter, our inventory dollars increased 14% to support our customers’ strong demand outlook for our products. Operating cash inflow was $14.8 million for the quarter. Capital expenditures for tangible and intangible assets were $4.6 million for the quarter. Free cash flow was $10.2 million for the quarter. During the second quarter of fiscal year 2026, Ambarella’s Board of Directors approved an extension of the current share repurchase program for an additional 12 months ending June 30, 2026. During the first quarter, we purchased 24,152 shares of our stock for a total consideration of approximately $1 million. As of today, there’s approximately $48 million available under our repurchase authorization. We had one logistics company representing 10% or more of our revenue. WT Microelectronics, a fulfillment partner in Taiwan that 8 ships to multiple customers in Asia, came in at 63.1% revenue for the quarter. I’ll now discuss the outlook for the second quarter of fiscal year 2026. Demand for our edge AI inference processors remain strong. We anticipate fiscal Q2 revenue in the range of $86 million to $94 million or $90 million at the midpoint. We expect single-digit sequential revenue growth in IoT applications with auto revenue expected to be slightly up versus the prior quarter. For fiscal 2026, we anticipate a revenue growth range of 19% to 25%. We expect fiscal Q2 non-GAAP gross margin to be in the range of 60.5% to 62%. We expect non-GAAP operating expenses in the second quarter to be in the range of $52.5 million to $55.5 million with the increase compared to Q1 driven by new product development costs, including a new AI SoC addressing the emerging IoT edge infrastructure opportunities described earlier by Fermi. We also anticipate a weaker U.S. dollar to have a moderately unfavorable impact on our operating expenses in the second quarter. We estimate net interest and other income to be approximately $1.8 million, our non-GAAP tax expense to be approximately $800,000 and our diluted share count to be approximately 42.6 million shares. Thank you for joining our call today. And with that, I will turn the call over to the operator for questions.

— Operator Instructions — It comes from the line of Christopher Rolland with Susquehanna.

Congrats on the quarter. To get to your full year guide, I just want to make sure I get the moving parts right. It seems like we are taking up our numbers in the first half. But just 9 looking at the sequential growth profile, it looks, at least versus prior that the back half, the sequentials are reduced versus our prior. And so I was just wondering has the growth profile actually changed? Is this related to the tariff kind of pull-in that you commented on last quarter? So are we taking from the – are we adding to the first half but taking from the second? Just what are the kind of moving parts in the growth profile for the year?

Well, first of all, I don’t think we are having concerns, at least our current annual guidance doesn’t have any concerns about second half strength. What we – if you look at, we extend the guidance range, if you look at the high end of the guidance range, we have regular seasonality and showing a strong second half growth. So it’s really about the – there’s uncertainty about – with the current geopolitical situation, and we want to build in some uncertainty in there. So I think we have still high confidence about our second half growth. And with our visibility in Q3 and we are building up visibility in Q4, I think that – I don’t believe we’re giving any signal that we have a weaker second half.

Chris, going into this – this is Louis. Going into this call, I think the consensus was about 51% in the first half, 49% in the second half. And at the midpoint, I do think those percents changed much, but the dollar figures, I think in every quarter would probably be going up a bit. So it’s another way to think about it. But I’d point out also that if you’re in the upper half of our guidance range, you’d probably end up with seasonality pretty close to normal. So it’s really your call. We’re just saying it’s an uncertain environment. It could happen and play out a lot of different ways.

Fair enough. I know you don’t guide a few quarters ahead, but would you expect October to be up seasonally? I know January is typically down. But is there any reason to think 10 that October should be up overall?

I think we can help you with the shape, but as Fermi said, not the absolute numbers. And I think it’s reasonable to think that, that would be a positive sequential number. And it’s probably reasonable to expect Q4 to be down sequentially. That’s what we can answer at this stage to shape, but not much more precision than that.

Our next question is from Tore Svanberg with Stifel.

Congratulations on the results. Fermi, you talked about edge infrastructure, and I’m sure this is something that you’re going to continue to elaborate on. But could you just explain a little bit what you mean by that? Obviously, we’re not talking about big AI clusters here. So yes, if you could just add some color on what exactly you mean by introducing new products for edge infrastructure.

Right. So I think if you look at how the device is being distributed on the top is really data center and cloud. On the bottom is really the edge endpoints, which is where we are having – serving our customer. But now become – it’s become clear with so many different advanced AI models happening and you just cannot upgrade the end points faster enough, right, of course, that our customers continue to want to replace the endpoints with new products or new cameras that can run efficient advanced AI models. But to upgrade the existing installed base, you can imagine that there is a – you want to integrate multiple endpoints that already in the installed base and using a server or a AI box that can integrate all of those endpoints, video input and run this model on that box right? So that would be the easiest way to upgrade the installed base. And I think that’s become a 11 trend, it becomes obvious. And among other things, this is just one early trend that we are seeing, and we believe that has momentum on that. And in the future, there will be many other on-prem servers, edge servers that can use our solution too.

Yes. That’s great color. And as my follow-up, I know your segment revenue is in IoT versus auto, but it sounds like non-camera IoT is really starting to proliferate here with IoT, industrial, enterprise wearables and so on and so forth. Is that business sort of approaching 10% of revenue? And will you potentially eventually split that out so that you don’t just sort of have investors focus on the security camera part of the revenue?

A couple of things there, Tore, it’s Louis. Most of our revenue today, the data is getting ingested by our AI accelerator through the lens of a camera, so that hasn’t changed. Although we have said it’s likely that, that becomes an incremental opportunity for us in the future, especially as we go into the edge infrastructure. But Fermi made a comment in his script about security is an end market for us, and that’s less than half of our revenue now. So now we’re seeing very good growth. As you know, auto is around 25% of our revenue. And then in other IoT markets, we’re starting to see solid growth there and adoption of AI in a wide variety of markets. So everything is still ingesting data through the lens of the camera, but that probably changes in the future. And security is ground zero for us because that’s where AI at the edge started. We led that market, and now we’re leveraging that expertise and applying it to a lot of additional vertical applications.

Our next question comes from Kevin Cassidy with Rosenblatt Securities.

I’ll also congratulate you on great results. Speaking to those results, would think with the 12 strong product cycles that you’re in, could there be a change in your seasonality maybe as the human interfaced devices become less relevant in your revenue?

Yes. You are asking about our CV products versus human viewing products?

Just a question with regard to whether our seasonality might be as much of an impact.

Right. So I think for this year, I think with so many uncertainty on geopolitical situation, that seasonality is definitely a question mark for us. Although we are not saying there’s no regular seasonality. We just say that we provide a much higher – a much broader range for the annual guidance to indicate there’s uncertainty on the second half. But I think there is definitely a scenario that no more seasonality can happen.

I see. And you’ve piqued our interest mentioning these new edge devices. These are all your transformer-based SoC?

Transformer-based – go ahead.

Yes. I think the question was – correct me if I’m wrong, Kevin. It was a little bit hard to hear you, on the edge infrastructure, do we expect that market to leverage our thirdgeneration AI accelerator to a high degree. And the answer is yes.

Yes. Obviously, because right now, the first – we already announced N1-655 this year for 13 that particular market. But we also understand the need for the customers who are going to build another chip for the family of the product so we can deliver a full – a complete road map for the customer. All of chips we are talking about today, still leveraging our third-generation CVflow architecture and the software to minimize our investment, but at the same time, provide a very competitive solution to – in the market. And more importantly, I think as you know, that third-generation of architecture can really do all of the advanced AI models based on transformers.

It is from the line of Joe Moore with Morgan Stanley.

As you talk about these kind of edge AI focus, I guess, is this a shift in focus for you guys? And I guess, how are you thinking about the sort of more the CV3 types of larger automotive ADAS opportunities? Are you moving resources maybe away from those things towards these other initiatives? Or are those initiatives still something that you’re enthused about?

Auto continue to be a focus. But I think with our current approach for auto is we already built a complete CV3 automotive series, as you know, that we have 685, 655 and the 635 that complete lineup for the autonomous driving software. And also, we’re going to continue to invest on our software side, both for the VisLab autonomous driving software stack and auto radar software stack. So that doesn’t change. However, with the – we finalized already CV3 family for the automotive road map. We definitely have resources that we’re going to put on the edge infrastructure. And also we talk about – we add another project, which is not in our annual plan. But we think with our revenue growth, we have a chance to build another silicon for edge infrastructure, which we are doing. So 14 we are definitely add a little bit more tape-out fee to improve our strength in this edge infrastructure business.

Okay. That’s helpful. And then, I guess, I know you don’t like talking about the sort of more futuristic humanoid robots and things like that. But there’s obviously a lot of kind of upfront investment in kind of paving the way to those types of markets and you have technology that should be important. So just how do you kind of frame that? Is that an opportunity that you’re willing to invest resources into?

Yes. In fact, we are investing the resource. Let me maybe go a little bit deeper than before. The way we look at robotic operation today, we look at – we view that market very similar to autonomous driving 5 years ago. What that means is that most of our customers, instead of trying to find the most efficient solution, they are still trying to piece different pieces of the solution together to build a prototype because the size of the market for each customer is still small. So we start seeing people trying to – using one box for the video perception, the other box for radar perception and using a CPU to integrate them together. So this really reminds me 5 years ago of the first generation of Level 2 car coming out. It’s a similar architecture. And we are that – in that stage right now, and we already have solutions like CV5, CV7 to provide video perception and the radar perception for those kind of solution. But – however, we also believe, just like autonomous driving car, moving forward for the high-volume robotic application, you need a domain controller and you need the end-to-end AI software to drive this application. So we’re going to definitely using our CV3 solution to continue to drive this application. But everything we are doing for this edge AI infrastructure, you can imagine that, that also can help the robotic solution. But 15 more importantly – that’s talking about silicon and software. But really go to market, you’re going to start seeing, maybe in the next quarter, we’re just going to start telling – introducing an idea how we’re going to change our go-to-market because we realize that in the past, we focused on addressing large customers. Now with a robotic implication, the customer – the market is very segmented. Most of the customers has small volumes. So we need to find a different approach, go-to-market approach to address this need, and we will probably definitely start talking about that approach next quarter.

Joe, it’s Louis. Just to wrap some hard numbers around that. So for the co-processing, like, say, the perception that would be parts like CV5 or CV7. And then, of course, the essential brain, the domain controller Fermi was referring to would be like the N family of products.

It is from Suji Desilva with ROTH Capital.

Fermi, John, Louis, congrats on the progress here. Maybe you can help me frame this edge AI server opportunities. Is there a way to think about the size of that relative to maybe the end devices? Some ratio or some way of thinking about the content of these servers relative to the device content? Any way to frame it so we can think about how it’s going to grow in your revenue?

Right. So maybe let me help you to the number that I’m thinking about. If you look at the aggregate, the current camera space, that’s using security camera as an example. There are roughly 1.2 billion installed base camera, which need to be upgraded, either upgrade by new cameras of this AI technology or upgrade with, let’s call it edge infrastructure box. 16 And those kind of box usually integrate, I would say, 8, 16, 32 different cameras into a box. And the content for that box for us is 3 digits, in the low 3 digits. So that – I hope that gives you idea of how we look at this market opportunity.

And one thing I’d tack on there also is it – having AI in the endpoint or in the edge infrastructure is not like a mutually exclusive thing. You can you can have AI in the endpoint along with the edge infrastructure servers.

No, that’s great. Understood. And then my other question around the edge infrastructure market as you’re going into this. How does the competitive landscape maybe shift and some perspective there versus things like FPGAs, GPU, CPUs that already target that market? Do you think about the competitive landscape differently or is it similar?

So in that market, it’s a very new market. When you’re looking at the near edge and the far edge of the market. And so the SAM numbers like we’re using are fairly small. So you do have some general-purpose-type processors used in these applications, whether it’s FPGAs or, of course, GPUs. We approach this market with a much more efficient solution when you measure it in terms of like performance per watt and consider thermal impacts on the total system cost. And so kind of the same advantages that we’ve talked about in other markets, we’ll be applying to this edge market, initially, say, the near edge. And your first question, you mentioned AI servers. That’s probably going to be part of it, too. But maybe initially, you’ll hear about the progress in some of the near-edge markets first particularly those that use cameras.

Our next question comes from Quinn Bolton with Needham & Company. 17

It’s Shadi Mitwalli on for Quinn. My first question is on some of the conversations you’ve had in regards to your customers’ supply chain. I know last quarter, you mentioned customers evaluating their own supply chains, which has caused uncertainty in the back half of this year. So just curious on how these conversations have progressed.

Talking to our customer about our supply situation. So we continue to have the conversation. One of the worry last time we talked about is whether our customer building up inventories. I think that we continue to have that conversation with customers All of them told us that they are not building inventory. In fact, they are watching the situation, and none of them is really eager to build any inventory at this point. So from that point of view, I think we feel comfortable with that. However, there’s still always a geopolitical situation. Every day, as we know, things can change. So that – we cannot speak for what we don’t know in the next – in the second half. So that’s where uncertainty is.

Got it. And my follow-up on gross margin. It sounds like some of your new CV chips have been tailwinds to ASP. However, gross margin is expected to decline next quarter. So I was just curious on what is driving the decline.

Yes. From any quarter-to-quarter, it’s really a combination of customers and product mix. That is the primary driver of how that corporate gross margin rolls up. And so ordering patterns of different customers and their contribution, that’s really the, I guess, you could say the primary driver for any one quarter’s gross margin guide.

Our next question comes from Gus Richard with Northland Capital Markets. 18

A couple of questions. The video management systems that – the 32 cameras or 16 whenever are attached to, those are coming out with, obviously, AI capabilities and the camera has AI capabilities. And I was wondering if you could help me understand how that AI split happens and why you need it in both places.

Right. So the quick answer to that is with installed base, you just cannot replace all the installed base camera fast enough with advanced AI cameras. So to enable the installed base with advanced AI models, this box – this kind of boxes is required and probably easiest way to upgrade. So that’s just the first answer. The second answer is with a lot of different AI improvement every month or every quarter, I can imagine that in the future, you’re going to continue to see more and more advanced model coming up, the camera can run a portion of it. But every time this camera comes out, it’s easier to upgrade the service with a box approach. So I think the combination of those two really drive this – the upgrade cycle.

Gus, it’s Louis. Just to add some comments. John kind of touched on it earlier, but you could have CV2-based cameras in the field doing detection and classification with CNN networks. And then you could provide an incremental layer of service with one of our GenAI chips that could accommodate much larger parameter models on the infrastructure side, the point of aggregation. And so maybe that’s one example of how it would be architected.

Got it. And then just thinking about the market, at this point, China is not part of your market. And I was just wondering if you could comment on how big the not China market 19 is for security cameras and sort of what you see your market share is currently?

When we talk about our 7, 10 numbers, we don’t include China number anymore in any security market. So that’s where we are at. And in terms of market share, outside China, I would say we definitely have a majority of the market share for the security camera in the mid and high end. On the low-end side, there are plenty of Chinese and Taiwanese suppliers try to compete with the low end with a $2 to $3 chip, which we don’t compete there. So if you look at – if you separate the line with the mainstream high end to the low end, on the top, we are probably the majority leader. And on the bottom, we’re just one of the players.

— Operator Instructions — We have a question from the line of Martin Yang with Oppenheimer.

First question is on the edge AI infrastructure product, is the second chip something new, meaning that you are pulling forward the development reacting to end market demand or something you have long planned in the road map?

It’s the first case. In fact, that after we talk to so many customers and what they need, we realized that N1-655 is great for the first product but we do need to have a second chip to keep competitive. And so I think that second chip – however, the second ship is leveraging our current CV3, our third-generation CV4 architecture and software. So the development is going to be fast and also the cost will be – we think can be easily controlled. But the add value is really helping customers have a better performance per watt and higher performance in the same silicon. 20

And then in this quarter, accounts payable trends a little higher than normal, is that associated with this new chip development?

Not specifically, Martin, no. I think as we started to grow the Q2 top line guide, it’s really more a function of building the inventory for – to support the demand that we’re seeing. And so corresponding with that is the accounts payable associated with it.

And ladies and gentlemen, this concludes the Q&A session. I will pass it back for final remarks.

And thank you all for joining us today, and I’m looking forward to talk to you next time. Bye.

Thank you. And this concludes our program. Thank you for participating, and you may now disconnect. Copyright © 2025, S&P Global Market Intelligence. All rights reserved 21