Ambarella, Inc.

Thank you for standing by, and welcome to Ambarella's third quarter fiscal year 2024 earnings conference call. At this time, all participants are in a listen-only mode. After the speaker presentation, there will be a question and answer session. You will be limited to one question and one follow-up. To ask a question during the session, you will need to press star 11 on your telephone. To remove yourself from the question queue, you may press star 1-1 again. I would now like to hand the call over to DP Corporate Development, Louis Gearharty. Please, go ahead.

Thank you, Lateef. Good afternoon, everyone, and thank you for joining our third quarter fiscal year 2024 financial results conference call. On the call with me today is Dr. Fermi Wong, President and CEO, Brian White, CFO, and John Young, VP of Finance. The primary purpose of today's call is to provide you with information regarding the results for our third quarter fiscal year 2024. The discussion today and the responses to your questions will contain forward-looking statements regarding our projected financial results, financial prospects, market growth, and demand for our solutions, among other things. These statements are based on currently available information and subject to risks, uncertainties, and assumptions. Should any of these risks or uncertainties materialize, or should our assumptions prove to be incorrect, our actual results could differ materially from these forward-looking statements. We're under no obligation to update these statements. These risks, uncertainties, and assumptions, as well as other information on potential risk factors that could affect our financial results, are more fully described in the documents we file with the SEC. Access to our third quarter fiscal 2024 results press release, transcripts, historical results, SEC filings, and a replay of today's call can be found on the investor relations page of our website. The content of today's call, as well as the materials posted on our website, are Ambarella's property and cannot be reproduced or transcribed without our prior written consent. Fermi will now provide a business update for the quarter. Brian will review the financial results and outlook, and then we'll be available for your questions. Fermi?

Thank you, Luis, and good afternoon, everyone. In a challenging market, our fiscal Q3 revenue was slightly above the midpoint of our guidance. Our business appears to be in the process of stabilizing as our customers seem to be making progress with their inventory management efforts. We expect our customers to emerge from the cynical downturn at different times throughout the next year. And considering all of the dynamics at this time, we continue to anticipate we will return to growth in fiscal year 2025. We remain determined and focus on our strategic R&D priorities for the introduction of a portfolio of AI SOCs and software targeting more sophisticated AI inference workloads. Our confidence in the cycle of growth opportunity for inference AI processors and software remain high. And our long-term serviceable market opportunity, CAGR estimate, has not changed. Our CV2 family of SOCs is expected to lead us out of a cynical downturn and represent a larger proportion of our total revenue in fiscal year 2025. As a reminder, this product family established umbrella in the AI inference market, and these SOCs are expected to approach 60% of our total revenue in fiscal 2024. In a single chip, the CV2 products typically provide both the video processing and the AI-influenced processing for one or more cameras. The average selling price of this product is close to $20, with a range below $10 to about $50. In particular, I want to point out that CV5, the flagship of the CV2 family and the umbrella first 5 nanometer SOC, is forecast to generate meaningful revenue next year, with more than 10 customers in production. Our CV3 platform is expected to generate productive revenue from the automotive market in calendar year 2026 and beyond. The CV3 platform includes a family of SOC and software, and we began to sample the CV3-80 SOC a year ago. And today I'm pleased to announce during Q3, we received the first silicon for CV3-80 685 SOC, the flagship of our CV3 portfolio. And we are in the midst of a successful bring up of this extremely sophisticated 10 billion plus transistor SOC. We expect to deliver engineering sample to customer in Q1 next year, and we are on track through the previously announced award from Continental to the first OEM start production in calendar year 2027. CV3 SOCs are based on our third generation AI inference technology and serve significantly more challenging AI inference workload, such as partial or complete autonomy in vehicles. In addition to providing in a single SOC the perception processing for multiple camera inputs, CV3 SOC for the first time enables centralized processing of raw radar data as well as deep low-level fusion with data from other sensing modalities. The CV3 family SOC ranging price from $50 to more than $400 per SOC. There are two additional elements of our CV3 platform, both of which are software. This includes Oculi adaptive AI radar perception software and our complete autonomous driving software stack, both optimized to run on CV3. Since November 23, we have been providing demo rides to OEMs and Tier 1s based on CV3 SOCs and our autonomous driving stack with centralized radar processing, low-level fusion, and the oculi radar perception software. And we will continue to demo this at the CES in January. Now I would like to provide an update on our GEN-AI plan. During the last quarter with the implementation of new LLM software building blocks on CV3-AD, we have now successfully demonstrated inference running LLM2-13B model with the SOC using LPDDR5 memory instead of high bandwidth memory and operating at a fraction of the power consumption of incumbent solutions. Some of our existing customers are evaluating how they will implement GenAI and large language models at the age of their networks. With our successful LLM demos and additional analysis, we have concluded the powerful and highly efficient AI inference processor embedded in our CV3 SOCs is well suited for this age market. We will have more updates on our LLM strategy at the CES 2024. Once a year, we update our automotive funnel, which is meant to be a reference for our automotive customer engagements over the next six years. We completed the funnel update in November, and it increased 4% from $2.3 billion to $2.4 billion. The funnel is comprised of one business of about $800 million and a pipeline of about $1.6 billion. With our automotive business expected to generate about $80 million in revenue this year, the $2.4 billion six-year funnel is an indication of the strong growth we anticipate from the automotive market. There was a significant number of upward and downward revision in the funnel calculation this year, including forecast changes and project delays from both Tier 1s and OEMs, projects won or lost in the pipeline, and the addition of new projects. CV3 represents a large portion of the current funnel, even though CV3 revenue is not expected to commence until calendar year 2026. I will now discuss represented customer engagement in this quarter. In automotive, we are continuing to win new designs in China, the world's largest automotive market. The open architecture of CV flow AISOCs enable leading Chinese tier 1s and ADAT software providers to develop highly differentiated, fully-featured solutions. During the quarter, GAC introduced four new passenger cars models incorporating Embraer's CV22AQ automotive SOCs in intelligent ADAT systems. This included GAC CHAMCY E8 and ES9 and ION HYPER HD and ION S MAX. Also, Hi-CAN introduced its V09 minivan with the ADAS system, also based on our CV22AQ. And the ACAIC introduced its Maxxis Dodge R7 minivan, including a camera monitor system replacing left and the right mirrors, and based on our CV22FS functional safety SOC. In the automotive aftermarket, Asian market leaders think we'll introduce its QXD DVR based on our CV25 AI SOC and the F200 Pro two-channel DVR based on our A12A SOC. And the European market leader Nexbase introduced its IQ smart dash cam product line using our CV22 automotive SOC, featuring 4K resolution and advanced AI-based threat detection. In the enterprise security camera market, A number of leading manufacturers introduce new cameras based on our flagship 5 nanometer CV5 and the CV52 families. Motorola introduces new H6A product line featuring AI-based video analytics and up to 8 megapixel resolution based on our CV52 AI processors. And Japanese market leader iPro announces new X series featuring total nine model spinning box indoor. and outdoor dome models up to 8 megapixel based on our CV-52 SOCs. And the Korean market leader Hanwha introduced its C-34404 multi-directional outdoor camera based on our CV-5 SOCs. Also during the quarter, Wakata introduced the CP-52 E-Pen tail a zoom camera based on our CV22 SoC, offering camera operators dynamic coverage of wide areas and featuring 5 megapixel resolution and 28 optical zones. In Germany, IDS introduced its first umbrella-based camera, the NXT Manable for Live Imaging Analysis based on our CV22 AI SoC. In the smart home segment, Signify, the Philips lighting spin-off known for its Hue lighting products, announces the Philips Hue Secure Cameras. Based on our CVE AI SOCs, the cameras can work with the Hue lights to control lighting and the sound alarms to deter intruders. Also during the quarter, Acroby, a leading home automation company, announces a smart doorbell camera based on our CVE 28 AI SOCs. In November, Insta360 introduced its ACE Pro and the ACE Action Camera based on our CV5 and the CV52 AI SoCs. The ACE Pro is the first 8K consumer action camera, and both cameras include advanced AI features, such as neural network-based noise reduction, 14 megapixel computational photography, voice and gesture control, and automatic video editing. Collectively, you can see overall design win activity is healthy across 13 different representative customers for 17 different projects, nine in auto and eight in IoT. Furthermore, 16 of the 17 representative engagements utilize our higher value AI SoCs. You can also see we are successfully implementing more sophisticated AI workload in our SOCs was highlighted with the identification of four different CV5 customer engagements. Looking forward, we believe our newer products, such as the CV5, CV72, and the CV3, are well positioned to support increasingly sophisticated AI inference workloads. And this new product ramp, and as we also capture more software value, we believe we can capture more value per designers. Now I would like to give an update on changes to our management team we previously announced in October 18, 2023. Brian, our CFO, will retire from his role as a CFO at the conclusion of the current fiscal year ending January 31, 2024. I would like to thank his contributions to the company, including helping navigate the challenging inventory correction the semiconductor industry is currently experiencing. as well as continuing to strengthen our financial team and position us for success as we continue our transition into larger markets. I'm also pleased to announce that John Young, currently our VP of Finance, will assume the role of CFO on February 1, 2024. John has been with Umbrella for more than six years, serving a variety of finance and accounting roles, and we are confident that he will continue to be a valuable contributor as we seek to drive our transformation forward and grow shareholder value. Now I will hand it over to Brian to discuss the Q3 results and outlook in more detail.

Thanks, Fermi. Before I begin, I'd like to thank Fermi and the Board for the opportunity I've had to be part of the special team here at Gambarella. The timing of my upcoming transition is facilitated by the progress we've made in developing the people, processes, and tools to support Ambarella at the next levels of growth and profitability. I have complete confidence in John's ability to take over as CFO and the team he has to support him. I look forward to their ongoing success. I'll now review the financial highlights for our third fiscal quarter and provide a financial outlook for our fourth quarter ending January 31st, 2024. I'll be discussing non-GAAP results and ask that you refer to today's press release for detailed reconciliation of GAAP to non-GAAP results. For non-GAAP reporting, we have eliminated stock-based compensation expense along with acquisition-related and restructuring costs adjusted for the impact of taxes. For fiscal Q3, revenue was $50.6 million, slightly above the midpoint of our prior guidance range. down 19% to the prior quarter, and down 39% year over year. As expected, the sequential decline in revenue was driven primarily by our IOT and market. Non-GAAP gross margin for fiscal Q3 was 62.6%, in line with our prior guidance range. Non-GAAP operating expense was 44.1 million, down 1.9 million from the prior quarter, and below our prior guidance range of 46 to 49 million, driven by continued expense management and the timing of spending between quarters. We remain on track to our total product development milestones. Q3 net interest and other income was 1.5 million, and our non-GAAP tax provision was approximately 650,000. We reported a non-GAAP net loss of 11 million, or a 28 cent loss per diluted share. Non-GAAP BPS was 11 cents better than the implied midpoint of our prior guidance range, driven primarily by lower operating expenses. Now I'll turn to our balance sheet and cash flow. Cisco Q3 cash and marketable securities increased 5.8 million to 222.3 million. DSO improved from 45 days in the prior quarter to 42 days, while days of inventory improved from 147 to 145 days, down $4 million from the prior quarter. Cash from operations was 7.9 million, and capital expenditures for tangible and intangible assets was 2.4 million. Free cash flow, defined as cash from operations less capex, was 11% of revenue for the quarter, and 6% on a trailing 12-month basis. We had three logistics and ODM companies represent 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 54% of revenue. Shikoni and ODM, who manufactures for multiple IoT customers, was 12% of revenue. And Hakuto, a logistics partner who primarily supplies multiple automotive customers in Japan, was also 12% of revenue. Now I'll discuss the outlook for the fourth quarter fiscal year 2024. The near-term revenue outlook appears to be stabilizing. While customers continue to manage their inventory levels and some pockets of end-demand softness persist, ordering patterns and customer feedback suggest that our revenue is leveling out and likely to resume growth in our next fiscal year. For fiscal Q4, we estimate that our total revenue will be in the range of $50 to $53 million, with IOT up slightly and automotive about flat. At this time, we anticipate that sequential revenue growth will continue in our fiscal Q1 in the low to mid-single-digit percentage range. We expect fiscal Q4 non-GAAP gross margin to be in the range of 62 to 63%. We expect non-GAAP OPEX in the fourth quarter to be in the range of 45 to 48 million. With the increase compared to Q3 driven by CES marketing activities, R&D tied to new product development, and less favorable foreign exchange impacts. We estimate net interest income to be approximately 1.5 million, our non-GAAP tax expense to be approximately 600,000, in our diluted share count to be approximately 40.4 million shares. Ambarella will be participating in a fireside chat and hosting one-on-one group meetings on December 5th and 6th at NASDAQ's and Morgan Stanley's London Conference. We also will be participating in OMERO's CES one-on-one conference in Las Vegas on Monday, January 8th. We expect to host more than 20 investor groups during our CES 2024 exhibition from January 9th to January 12th. Please contact sell-side analysts to make reservations. We will also present and host one-on-one and group meetings at the Needham Conference on January 17th in New York. Please contact us for more details. Thank you for joining our call today. And with that, I'll turn the call over to the operator for questions.

thank you as a reminder to ask a question you will need to press star 1 1 on your telephone to remove yourself from the question queue you may press star 1 1 again you will be limited to one question and one follow-up please stand by while we compile the q a roster our first question Our first question comes from the line of Joe Moore of Morgan Stanley.

Great, thank you. What if you could talk about the adjustments, Jeremy, that you discussed to the funnel? You had said that you saw some business kind of coming out. I know you had a number of businesses there that were probability weighted. What's happening to that when that happens? Is that because the programs are just moving more slowly? Is it competitive? Is it supply chain? Can you just kind of give us an indication of why there would be those adjustments?

Yeah, thank you, Joe. I think, let me address that question. In fact, in the script, we talk about four different factors. The first factor is that forecast change from Tier 1 and OEMs. For example, for some of the projects that we have won, and the forecast of those projects would reduce, particularly in the last several months, we start seeing that trend. So that's the first factor. The second factor is we start seeing project delays. For example, for some of the projects that we have won, the production date is pushing out. And also for some projects that were in the RFQ phase, the decision of the award has also got to push out. So I think that's the second factor. But also the third factor is really that we continue to have a new project or RFQ that we're bidding on, and we have different multiplier, like you said, different confidence level for different projects, and that's a new project we bid on. The last factor is really the project that we won and lost. at the end. So if you ask me, I think the first and second factor really negatively impact our funnel number. But the third and fourth, the net effect in the third and fourth probably help us to continue to add numbers into the funnel. So that's the reason we have this minor increase on the funnel numbers. Great. That's helpful. Thank you.

And there's still an awful lot of growth over multiple years implied here. How does that profile look versus a year ago? I know you talked about a lot of success in China. Do you see more of a China skew to your numbers as you start to ramp the advanced CV features?

Right. So I think that, first of all, this funnel, first of all, on the one portion, this is the first time we start seeing a CV3 contribution there. In the pipeline, definitely CV3 continue to getting higher and higher percentage of the a final number. One thing I can say that in addition to the first CV3 design wing we announced two times ago with County, we add extra non-Chinese commercial vehicle design wins with CV3 and in this pipeline. So we definitely continue to make progress on that. And the trigger for those design wins is after we demo the working 685, CV3-685 chip, and also demo this software stack we talk about, our AI software stack that can integrate all from the Fusion to the radar as well as all the past planning. So we successfully demo those things to OEM tier ones. Those two factors help us to secure the design wins we just talked about.

Thank you. Our next question comes from the line of Torrey Swanberg of Stifel.

Yes, good afternoon. This is Jeremy calling for Torrey. I guess first just wanted to follow up a little bit more on the adjustments made to the funnel. Can you talk about how maybe, you know, on a broad level, how maybe, you know, did the overall market in terms of maybe SAR impacted and how much that might have impacted the slight increase in the funnel? And then I've got to follow up after that.

I think the global market definitely impacts two things. One is the forecast of the run rate of the Project 41. But also, I think you can read a lot of articles talking about for the level 2 plus or level 3 card, a lot of the new projects the decision got pushed out because people are having different thoughts about their software stack, what's the function features, and particularly how to compete with China with their aggressive product planning. So I think all of those contribute to the first and second factor I just mentioned.

Great. And then moving to the AI inference accelerator business that you've successfully demonstrated, Are we right to read into it that it sounds like this is something that you're moving forward with? And can you talk about maybe your competitive positioning relative to whether it's the current incumbents, the number of private companies that are in the space, and maybe other large competitors that might start to move in? Can you just talk a little bit about that?

I think you are referring to the LLM. So I think, you know, first of all, this time, a little bit different than the last time. Last time, we kind of talked about global, a lot of different applications because we haven't spent enough time on the marketing side. This time, we're kind of doing to say, While we understand our strength is the edge server, edge devices, particularly because we are in the market and we know a lot of customers in there. And we've also done enough demos and also talking to customers to understand our strengths against our competitors, both for larger companies, of course, also startup companies. And our strength is, first of all, we have a working silicon that we can give demo at a very good performance, a very low power consumption. I think that's key. The second thing is, for us, it's not just a silicon. In fact, we also have bundled software. We're going to show a development environment that we already introduced our customer, but we are probably introduced to at the CES about how we using a software environment to bundle with our current solution that enable edge device or edge servers quickly moving to LL model to help edge application. So I think that's where we're focusing on right now. And it's not that we don't want to focus on bigger cloud or hyper A much bigger application is really that I think with our current silicon and our current resources, we would like to focus on areas that we know that we are familiar with and we want to know that we can win. So I think that's where we are sitting on with the current plan. Of course, we definitely need to build a roadmap, but however, like we said last time, Our current commitment to LLM is using our current engineering resource, which is enough for to do demoing system, developing the application, developing software for H applications. Those applications are using current resources. Had we decide to extend our roadmap or doing other applications, we probably need to talk about different engineering resources, which we haven't committed to that yet. Thank you.

Our next question comes from the line of Christopher Rowland of Susquehanna.

Hey, guys. This is Matt Myers on for Chris. So you guys talked about first quarter potentially growing mid-single digits. Maybe that's a little bit slower than we were thinking about. So I was curious, is this more end-demand driven or inventory driven and How should we think about inventory dynamics into the April quarter by auto or IOT within enterprise or consumer too within there? And then how are you thinking about the shape of end demand into the first quarter and then through next fiscal year?

I think that the data points that we're looking at in support of the growth we talked about for fiscal Q1 is based on, number one, looking at our bookings and backlog trends. which have absolutely improved. So our book-to-bill was much better in Q3 than what we've been experiencing for a number of quarters. And as we look at the backlog build into Q1, it's tracking well above the prior quarter and it's supporting the growth range that we talked about. And so when we combine the data from bookings and billings and we We compare that to the feedback that we're getting from our customers as well as the requests that we're getting relative to push-outs versus pull-in of shipments, which have become much more balanced than they have been in the past. That gives us the confidence to talk about the beginning of growth in fiscal Q1 and then growth for the full fiscal 25. But certainly we have a ways to go. And, you know, so that recovery is certainly helped by the turnaround of the inventory situation and markets also seem relatively healthy and in support of that growth. But I would say the primary driver at this point is getting to the bottom of that inventory cycle and getting through that excess material at customers.

Got it. Thanks, Brian. And then I also wanted a question on video processors. So based off of you talking about CV2 being about 60% of revenue, there's clearly a big drop-off in video processor revenue this year. And if CV2 is going to continue increasing next year, what's your outlook here particularly for video processors and Are you still investing a lot in this business or getting new wins here? And how should we just think about this particularly next year?

Right. So first of all, I want to point out, although we haven't really taped out video processing only chips, but video processing technology continues to be a very important part of our offering in our AI chip. So CV2, all the CV family chip has a big portion of the AI, sorry, the image processing or video processing technology in there. So I just want to clarify that. So this year, our CV total revenue approach is 60%. Obviously, what I just said, and also if you look at the design momentum, we have majority of that is CV. So I think the percentage will continue to go to favor of CV. But however, I think the decline of a video processor will slow down because we have a big reduction in the last two years. But we still have long tails of a video processor chip that's selling to... different market, including the enterprise security camera as well as the automotive. So I think that we should expect much less reduction on the video processor and with a very long tail.

Thank you. Stand by for our next question, which comes from the line of Kevin Cassidy of Rosenblatt Securities.

Yes, thanks for taking my question. This quarter, Hokuto, the Japanese auto distributor, came up as a 12% customer for the first time in a year or more. Is there anything happening there, or is it just because the overall revenue is down, or is there some new designs happening in Japanese automobiles?

Well, I think it's really that if you compare to a previous quarter, our other business dropped quickly, and the Japan automotive, usually they just take at a fixed rate, and it's seldom to change the forecast. So it's not particularly anything happening in Japan, but it's really that outside Japan, the drop is significant.

Okay, thanks. And just as you're going forward with the consumer market, and even as consumer surveillance cameras, and with large language models being used, is there a time when the input for the consumer will be just their voice? They can program the device. Is that something you're working towards?

You know, we are working now with just a language. We're working on multi-modality. In fact, we think that the LRM for us, for our customer's application, the most important thing is processing text, also speech, as well as image. All of them together, that adds the biggest advantage to our customer, and also that's probably one unique function that we can really differentiate against other LLM suppliers.

Thank you. Our next question comes from the line of Quinn Bolson of Needham & Company.

Hey, guys. This is actually Neil Young on Bitcoin. Thanks for taking my question. I wanted to ask, so it's been about a year since you announced the partnerships with Continental and Bosch. When do you believe we'll start to hear about design wins from these partnerships?

Well, we already announced a design win with Conti. We couldn't mention OENM, but I think we already did that. In fact, I would say that our relationship with Conti continues to improve. and we are building our projects with OEMs together. In fact, we cannot announce OEM design without customer approval, but one thing I can say is that activity with County is heavy, and also we continue to believe that the engagement between County and us are getting tighter and tighter, both on the hardware design as well as the software design.

Okay, great, thanks. And just another follow-up. So there's been some press reports suggesting that Continental is considering divesting its automotive assets, including its autonomous mobility business. How would a divestiture of this business affect your guys' partnership? Thanks.

Well, yeah, we heard similar reports. I think one thing I can say is, based on our conversation with Conti, we continue to believe this division that we're working with... is continue to be strategically important for both Umbrella and also county. And like I said, and with the close development that we continue to have, we didn't see any sign of slow down on the county side. So we continue to believe this relationship is important and also the product that we're working on is continually important for both Umbrella and the county.

Thank you. Again, to ask a question, please press star 11 on your telephone. Again, that's star 11 to ask a question. Our next question comes from the line. Martin Yang of Oppenheimer and Company.

Hi. Thank you for taking my question. Just one question. Can you maybe comment on the progress for customer design engagement regarding your CV72 products in automotive market?

Yes. CV72 is automotive product is designed for Chinese market because It's not ASO, so that only Chinese, only applicable to the Chinese market. And also last time we talked about we have multiple tier one design wins. And also at that time, last time we talked about multiple, all of the tier ones are working on building a demo system to show to OEMs. In fact, one of our tier one just yesterday demo live, in a broadcast live demo driving a car based on CV72 ECU from the Shanghai airport to the downtown Shanghai autonomous driving. So I think that we continue to make huge progress and we are ready to, in fact, we already started that demo into OES and also, like I said before, I think CV72 position in a very favorable price performance for the middle end of a level 2 plus solution in China. And we continue to believe that we have a competitive solution for that particular market. And we are working closely with TO1 to talk to OEMs. Hopefully, they will announce some design wins soon.

Thank you for me. A quick follow-up on that Civitronics 2 activity. Can you maybe remind us the expected timeframe of design wing and shipments?

All the projects we're bidding on is production the end of 2025 and the revenue coming from 2026. So the design wing has to be maybe first half of next year and spending 18 months of development cycle to get it done.

Got it. Thank you.

Thank you. I would now like to turn the conference back to Dr. Fermi-Huang for closing remarks.

Thank you, and thank you, everybody, for joining us today, and I'm looking forward to talking to you soon. Thank you.

This concludes today's conference call. Thank you for participating. You may now disconnect.