Ladies and gentlemen, thank you for standing by, and welcome to Ambarella's Q3 fiscal year 2022 earnings conference call. At this time, all participant lines are in a listen-only mode. After the speaker's presentation, there will be a question-and-answer session. To ask a question during the session, you will need to press star one on your telephone. Please be advised today's conference is being recorded. If you should require operator assistance, press star then zero. It's now my pleasure to hand today's conference over to Louis Gerhardy, corporate development. Please go ahead, sir.
Thank you, Holly. Good afternoon, and thank you for joining our Q3 fiscal year 2022 financial results conference call for the three months ending October 31, 2021. With me on the call today is Dr. Fermi Wang, President and CEO, and John Young, VP of Finance. The primary purpose of today's call is to provide you with information regarding the results for the Q3 of our fiscal 2022. 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 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, including the annual report on Form 10-K, filed on March 31, 2021 for fiscal year 2021 ending January 31, 2021, and the Form 10-Qs filed on June 8 and September 8 of 2021 for the first and Q2s, respectively, of fiscal year 2022. Access to our Q3 fiscal 2022 results press release, historical results, SEC filings, and a transcript of our prepared remarks and a replay of today's call can be found on the investor relations portion of our website. Fermi will begin the call with a business update. John will review the financials, and then Fermi, John, and I will be available for your questions. With that, I'll turn it over to Dr. Fermi Wang.
Thank you, Louis, and good afternoon, everyone. Before proceeding with our Q3 fiscal year 2022 call, I want to note on November 9, we filed an 8-K stating that our CFO, Casey Eichler, will be taking a leave of absence for health reasons. We have received many of your kind thoughts and prayers, which we have shared with Casey and his family, and we wish him a speedy recovery. Meanwhile, John Young, VP Finance, has been named Interim Principal Financial Officer and the Principal Accounting Officer during Casey's absence. On October 26, our strategy transformation into a deep learning AIoT processing company took another leap forward when we announced the acquisition of Oculii, a provider of advanced algorithms for image radar systems.
Oculii enable us to not only capture incremental perception business, but most importantly, feed our long-term strategy to provide a more comprehensive AIoT processing SoCs to our customers. I am also pleased to report our market and the financial momentum was strong in Q3, with revenue slightly above the high end of our range of guidance, up 64% year-over-year. Driven by a richer mix of CV SoCs, our average selling price continued to rise, contributing to strong but positive operating leverage, with non-GAAP operating margins reaching 25% of revenue versus 5% a year ago. Supply dynamics remain a key challenge. On one hand, we started to see an expected recovery in supply from Samsung's wafer fab in Texas. However, the shortages of other companies' components have become a more significant and a gating factor to our results and outlook.
Despite these constraints, we are on track to achieve record revenue this year, with the CV representing more than 25% of our revenue. During Q3, the number of unique CV customers in production grew sequentially and quadrupled versus a year ago. As of today, on a cumulative basis, we have shipped more than 5 million CV SoCs, including more than 1 million into the automotive market. Our automotive revenue funnel is expanding. A year ago, we announced a six-year discounted automotive revenue funnel of about $600 million. Using the same methodology, this figure has approximately tripled. We will provide more details on January 4 at our Capital Markets Day. I will now provide you with some examples of progress in our target markets. During the quarter, we were excited to see Rivian, the American electric vehicle maker, start customer delivery of its R1T truck.
The R1T's Driver+ features 11 cameras, 5 radars, and 12 ultra-ultrasonic sensors to deliver true hands-free driving assistance along with a full set of safety features. The R1T's Driver+ system utilize multiple CV28AQ CVflow automotive SoCs for its AI vision processing. Additionally, the R1T also use Ambarella's CV22AQ CVflow automotive SoC for its surround view camera processing and the Gear Guard security system. The Rivian design highlights the use Ambarella's AI vision SoCs in centralized automotive computing applications. These applications represent a major new opportunity for Ambarella moving forward. During the quarter, Chinese truck maker Shaanxi enter into mass production with an ADAS system providing lane detection and forward collision warning features. The system is based on Ambarella's CV22AX automotive AI SoC and is supplied by Tier 1 Bolo.
In the aftermarket automotive dash camera market, European market leader Nextbase announced a pilot program with Uber for its 232GW dashcam. The dual-camera design is based on Ambarella's A12 SoC and includes a cabin view camera that use infrared night vision technology and a wide angle 140-degree angle lens to provide an extra level of security for Uber drivers and the riders. In the dash camera market, Korean market leader Thinkware introduced its QXD7000 and QXD5500 HD designs based on Ambarella's H22 and A12 automotive SoCs. The QXD7000 includes a radar support mode that helps to pre-record a video 10 seconds before impact, while both models offer ADAS features, including forward collision warning, forward vehicle start alert, and a lane departure warning.
In the professional IP security camera market, we continue to enjoy strong revenue growth based on the worldwide market expansion and our increased market share outside of China. In Europe, market leader Axis, a unit of Canon, announced its first camera to use Ambarella CVflow SoC. The new M series cameras offer great image quality even in challenging light conditions and take advantage of analytics based on deep learning on the edge. Verkada, based in U.S., is simplifying the management of video security at scale through its hybrid cloud approach and its latest CD62 4K dome security camera features intelligent edge-based video analytics powered by Ambarella's CV22 SoC. Also, industrial giant Johnson Controls introduced its Illustra Pro Gen 4 mini dome indoor and outdoor PTZ model with resolutions up to eight megapixels and based on our CV22 SoCs.
With built-in AI-based object classification, the camera enables events to be narrowed to different classes, including person, car, bus, and motorcycle. In Korea, market leader Hanwha Techwin introduced multiple camera models spanning dome, bullet, panoramic, and the license plate recognition designs, and leveraging Ambarella's CV2 and the CV22 CVflow SoCs. Also in Korea, IDIS introduced four new camera based on our CV3, H3L SoCs, replacing HiSilicon designs. While CPRO introduced its first 4K AI camera based on our CV22 SoC. In November, Chinese electronics giant Xiaomi introduced a new line of multifunctional smart door locks with facial recognition using Ambarella's AI vision SoCs. The smart door locks fuse structural light and RGB camera technology and AI-based facial recognition software for hands-free access, while also incorporating other unlocking method, including fingerprint recognition and the near-field communication.
From these customer engagements, you can see the breadth of AI applications emerging, ranging from L2+ ADAS in EVs with a leader like Rivian to enterprise-class security cameras with an industry giant like Johnson Controls, and the door locks with sensor fusion for the smart home and the enterprise market with a major company like Xiaomi. This transaction can be traced back to the premium growth R&D investment we have and will continue to make into camera and the radar perception technologies, automotive functional safety such as ASIL B or D, and next generation AI processor on 3-nanometer manufacturing process node, all optimized specifically for IoT endpoint applications. In fact, in the last three years, Ambarella invested 59% of revenue into R&D on a GAAP basis, a level far beyond a maintenance level you see at most companies.
Of course, it is not just about the quantity of R&D investment, but the quality. As you know, we consider our visual AI R&D to be the best out there. Evidence supporting this continue to mount as we successfully leverage our well-established video processing leadership into the computer vision AI IoT market. This large growth R&D investment has been focused on processing data collected from the lens of camera with the algorithm-first approach. Now we have augmented this investment with acquisition Oculii, who use a similar algorithm-first approach for the high definition imaging radar market. Radar perception is an incremental market for Ambarella. However, we are most excited about the synergy with Oculii, which we expect to be derived from two principal areas. First, we expect our combined camera and the radar R&D investment to enable breakthrough levels of perception at economical price point.
As Ambarella is now one of the very few semiconductor companies to have advanced camera and radar technology development under one roof. Second, through our CV SoC roadmap, we are able to provide the processing power not just for the perception processing, but we target incremental processing for the fusion, planning, and control layers in a variety of AIoT markets. All of this is expected to drive more value, both higher selling price and improved market shares for Ambarella and its shareholders. In summary, during Q3, we demonstrate progress on the strategic front with the acquisition of Oculii and our CV portfolio ramp continues.
Our financial results are providing an early look at how we intend to capitalize on an entirely new phase of the digital AI transformation, a phase where deep learning is not just executed in servers, but in our SoC platform design for the unique needs of the AIoT endpoint markets. I want to thank all our stakeholders, employees, suppliers, partners, customers and shareholders for your continuous support. With that, John will now provide our prepared financial comments.
Thank you, Fermi, and good afternoon, everyone. I will review the financials for the Q3 of fiscal year 2022, ending October 31, and provide a financial outlook for our Q4 of fiscal year 2022, ending January 31, 2022. 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 and acquisition-related costs adjusted for the impact of taxes. Our revenue of $92.2 million was slightly above the high end of our guidance range, representing a sequential increase of 16% from Q2, and a 64% increase from the year ago quarter. The previously anticipated wafer supply recovery from Samsung's Austin, Texas fab helped sequential growth in the quarter.
Automotive and AI and IoT camera revenue combined increased more than 10% sequentially, and other revenue experienced seasonal growth. Non-GAAP gross margin for Q3 was 63.1%, slightly above the 62.8% in the preceding quarter, and slightly above the high end of our guidance range. The pricing environment for our products is stable, our mix remains favorable, and we are managing higher supply chain costs. Non-GAAP operating expense for the Q3 was $35.6 million compared to $36.4 million for the previous quarter. Our Q3 operating expense was below the low end of our guidance, primarily due to the timing of non-recurring R&D expenses. Other income of $408,000 was higher than expected, primarily due to the realization of gains on investments as they were sold to finance the acquisition of Oculii.
The non-GAAP net income for Q3 was $22.2 million or $0.57 per diluted share, compared with non-GAAP net income of $13.1 million or $0.35 per diluted share in the Q2. Total headcount at the end of the Q3 before the Oculii transaction closed was 824, up 8% from a year ago, with about 82% of employees dedicated to engineering. Of these engineers, about 69% are developing software and algorithms. Total accounts receivable at the end of Q3 were $44.8 million or 45 days of sales outstanding, versus $38.3 million or 44 days of sales outstanding at the end of the prior quarter. Net inventory at the end of the Q3 was $47 million compared to $42.1 million at the end of the previous quarter.
Days of inventory increased to 118 days in Q3 from 115 days in Q2. In Q3, our operating cash flow was a + $8.3 million versus a + $14.4 million in the prior quarter. Cash and marketable securities were $457.8 million, up from $449.2 million at the end of the Q2. We had two 10%+ revenue customers in Q3. WT Microelectronics, a fulfillment partner in Taiwan who ships to multiple customers in Asia, represented 63% of revenue. Chicony, a Taiwanese ODM who manufactures for multiple customers, was at 13%. Dahua and Hikvision combined declined sequentially and represented a mid-single digit portion of our total revenue.
In the professional security camera market share shifts between some of our customers are becoming noticeable. I will now turn to our guidance for the Q4 of fiscal year 2022. We completed the acquisition of Oculii on November 5, and our Q4 guidance incorporates Oculii's results from operations from that point, or about 12 of the 13 weeks in the quarter. As reported by many of our customers, underlying demand remains solid, but the supply side dynamics remain challenging. In Q3, we began to experience the expected improvement in wafer supply following the shortfall from the Texas freeze. However, shortages of other components at our customers became more acute in Q3 and continues into Q4. As a result, our guidance contemplates that some of our orders may continue to be rescheduled to ship at a later date.
Based on these factors and our best judgment at the current time, we expect total revenue for the Q4 ending January 31, 2022, to be in the range of $88.5 million-$91.5 million. Revenue from the automotive market is expected to increase sequentially, with non-auto IoT camera business expected to decline sequentially. In Q4, we expect the revenue from Dahua and Hikvision combined to decline to the low single-digit % of our total revenue. We estimate Q4 non-GAAP gross margin to be between 63% and 64%. We are experiencing some higher costs to manage the supply chain, but a healthy customer and product mix, together with a relatively stable pricing environment, are likely to enable our gross margin to continue to temporarily remain above the high end of our long-term model of 59%-62%.
Non-GAAP operating expense in Q4 is projected to be between $39 million and $41 million due to higher SoC development expense, increased hiring, and the inclusion of Oculii's operations. The Q4 non-GAAP tax rate should be modeled in the 3%-6% range. We estimate our diluted share count for Q4 to be approximately 39.5 million shares. Ambarella will be participating in Wells Fargo's TMT Summit tomorrow, December first, Imperial Capital Security Investors Conference on December fifteenth, Needham's Growth Conference on January twelfth, and Baird's Vehicle, Technology and Mobility Conference in the last week of January. Our Capital Markets Day will be held on Tuesday, January fourth, from 1 P.M. to 4 P.M. at our CES hotel location in Las Vegas, and we will be offering sell side analyst-hosted tours of our demos on January fifth, sixth, and seventh.
Please contact us if you would like to participate. I will now turn the call over to the operator for questions.
Thank you. As a reminder, ladies and gentlemen, if you would like to ask a question, please press star then one on your telephone keypad. Once again, star one to queue for a question. Our first question is going to come from the line of Gary Mobley with Wells Fargo Securities.
Hey, guys. Thanks for accommodating me and fitting me in here early in the queue, and wishing Casey a speedy recovery. I want to start out by asking about the puts and takes as embedded in your December quarter guidance. To what extent does seasonality play into it? To what extent do extraneous supply chain factors factor in here? If not for the extraneous factors that seemingly are constraining your Q4 guide, given the backlog that you have, how many quarter-over-quarter gains or sequential improvements in revenue can you see just based on your backlog support, assuming all of the supply chain issues that you've got dealing with now, you know, work in your favor?
Right. This is Fermi. I think the seasonality is still there, but however, it's reduced from the past. If you look at the historical number in the last five years, our Q3- Q4 on average reduced by 11%, but our guidance much smoother than that, just reflecting that the mix of the product is changing and definitely helping to reduce the magnitude of the seasonality. However, the other gating factor is really that the supply chain issue of other component vendors for our customers, which we start seeing become bigger problems. It's hard for us to size up the degree, the impact to our revenue, but we do believe there is an impact to our revenue forecast. However, all of the factors have been considered when we provide this $88.5-$91.5 revenue guidance.
Okay. Appreciate that. Hopefully I'm not catching you flat-footed with this question, but you know, per your Q4 revenue guide, you're expecting roughly 49%-50% revenue growth in fiscal year 2022. Given the increasing mix of CV revenue, which I believe is gonna be more than 25% of that fiscal year 2022 revenue, can you give us a sense of the unit growth versus the ASP tailwind for the fiscal year?
Right. I really think that the majority of the growth is coming from the ASP, and we continue to see a strong ASP growth. We said double the ASP in the past, and we'll continue to see that trend in our near future. Unit number also grows. We talk about CV total unit, CV units is 5 million accumulatively since the beginning. I think that's a significant number, and also they're including 1 million automotive silicon. I think both sides that we see growth, but I think the ASP is a much bigger factor at this point.
Appreciate that color, Fermi. Thank you.
Our next question is going to come from the line of Matt Ramsay with Cowen.
Hi, this is Josh Buchalter on behalf of Matt. Thanks for taking my question and, congrats on a strong quarter. I was hoping we could dig into diversity and potential timeline of the funnel that you mentioned tripled. Are there any particularly large wins contributing? I realize it's early, but is there any radar processing or Oculii-related revenue embedded within that funnel? Thank you.
Yeah. Hi, Josh. This is Louis. You know, we'll go into a lot more detail on the funnel at our Capital Markets Day in January fourth. Just to give you know, some of the high level parameters, continues to be a 6-year funnel. It uses the same methodology that we employed a year ago. There is no radar in this funnel yet. You know, looking at the two key elements that add up to the $1.8 billion, there would be the one component which increased from $400 million a year ago to about $700 million in this current funnel. The pipeline portion of it is now about $1.1 billion.
You know, the other thing that we can add to this funnel description and characterizing the current funnel is that a vast majority of it is CV SoCs and, in particular, of the increase from a year ago, you know, a vast majority would be computer vision SoCs. Hopefully that gives you enough color, and again, we'll go into more details at the capital markets day.
No, thank you for that, Louis, and I appreciate that you got the Capital Markets Day coming up. I was also hoping you could provide some more details about the Rivian win. It seems like you're doing more sensor fusion and centralized processing than some of your other announced wins. Is that, and is this the type of win that you're looking to compete for on a growing basis going forward? Thanks, and congrats again.
Yeah, thank you. Absolutely. The Rivian side, we talk about we have three types of silicons in there. The first type is CV28AQ. We have multiple CV28AQ there in the Rivian R1T car, and the function of the CV28AQ is doing video perception. That's definitely the key components that we want to shoot for in the future. You can see that we're definitely providing the majority of the perception in the R1T car. The second type of silicon is CV22AQ, which provides the surround view as well as some security camera functions that R1T has. The third type is called B8. It's really a SerDes, serial to serial, chip for the MIPI interface. Those are three things that we provided to the R1T.
Thanks, guys. Investment is to Casey.
Thank you. Our next question is going to come from the line of Tore Svanberg with Stifel.
Yes. Congratulations on the solid results. First question for you, Fermi, as relates to Oculii, obviously you talked about the radar perception market being an opportunity itself. You said longer term, obviously this is much more about creating more value SoCs with more integration. Could you elaborate a little bit on the timing of that? You know, when could we potentially see, you know, combined products from the two companies in the marketplace?
Right. The product planning to integrate Oculii's software into our silicon comes in two steps. The first step is that we are working on to integrate the software into our CV2 family, and that's a project we are working on. We know how to do it, and we believe that can be a product we start selling, I'll say, later next year. The idea is that that's going to be a product we use to target the level two ADAS application as well as a non-automotive application. Of course, for all our next generation chips, we'll integrate radar technology into our silicon. You should assume that every silicon that we're going to announce in the future will have our radar technology in there.
That's very exciting. As my follow-up question, you talked about the landscape for the security camera market kind of changing a little bit. Could you elaborate on that? I mean, I think you did mention it, you know, when you talked about the two Chinese customers, but you know, it sounds like there's some more specific things there, especially as those two customers will be very low percentage of revenues going forward. If you could talk a little bit more about that'd be great.
Right. I think from the momentum point of view, outside of China, we're doing very well. We talk about almost all the major customer outside China using our CV silicon, and in fact, most of them into production already. We mentioned Axis, we mentioned Johnson Controls, and we mentioned some Korean company this time. Inside China, I think Dahua continues to use our CV chip. If you look at what Hikvision and Dahua announced in the latest quarter, I think there are two things to be noticed. One is they all talk about a slow Chinese market because the government spending is reduced, and also they have a huge amount of inventory. We think one company has, like, 150 days worth of inventory at hand.
I think those two things really contribute a weak performance of Hikvision and Dahua. More importantly, I really think that we continue to see that there's still a lot of HiSilicon inventory sitting in the channel. I think that although it's been a while, but I think we haven't seen the end of a HiSilicon chip yet.
Great. Congrats again. Thank you.
Our next question is going to come from the line of Quinn Bolton with Needham & Company.
Hey, guys. Congratulations on the nice results and best wishes to Casey for his speedy recovery. I wanted just to come back to the $1.8 billion pipeline. I know you said it doesn't include any radar, but could you give us some sense as to what you think that radar opportunity might be as you look out several years? I mean, I assume that that's probably an opportunity that could bring, you know, hundreds of millions of dollars of TAM to the company, but hoping you might be able to size that.
Yeah. This is Louis. The radar market today in the automotive space only is maybe 90 million units. The Oculii technology, let's call it the 4D imaging radar, is just beginning to penetrate the market. If you consider two parts of their business, they have a licensing business, and then they have a module business. On the licensing basis, they can realize you know, low single digit you know, license fee per unit. At the higher end of that, it could be up to $15 of license fee. On the module side, our expectations for the units in that market are much lower than what we'd anticipate on the software licensing side. The ASPs in that market, on the other hand, are quite high.
Let's say it ranges from, you know, $200- $2,000, you know, depending on the type of module and the application and of course, the volume. Those would be some of the parameters that you'd wanna consider. You know, the key question is, you know, what's the penetration rate of this 4D technology going to be into what's already a large existing market.
All right. Thanks for that, Louis. The next question I had is just you talked about some of the supply constraints of other components potentially affecting shipments or causing shipments of your devices to push into future quarters. Do you think any of the demand that's affected may be perishable, or would you expect anything that sort of slips out of the January quarter would most likely be captured in the April or July quarters?
This is Fermi. I think part of the business is perishable. For example, all of the consumer security, which is more like a seasonal holiday season sales, I think those are perishable. I think for the most automotive design, I think it's pushed out to next year. We see that the short supply really impacts all of our customer products out there, including security camera and automotive. Obviously, automotive got bigger hit than the security camera at this point. I think that we don't see any this gating factors being reduced anytime soon. We definitely continue to watch the progress, and hopefully we can give you more guidance for next year when we talk in the CMD.
Great. Thank you, Fermi.
Our next question will come from Andrew Buscaglia with Berenberg.
Hey, guys. Thanks for taking my question. I wanna follow up on
Andrew?
Hey. I just wanna follow up on, you know, can you talk a little bit more about home security in this quarter and into Q4 and potentially if the momentum even into 2022? I know that second wave of growth was supposed to be pretty strong. I just, you know, like to hear more about some of the factors influencing that Q4 guide as it relates to that specific market.
I think for consumer security camera, we do believe that the percentage of CV revenue goes up significantly from Q3- Q4 and will continue to see next year. Particularly, we think there are a few customer will be in production in the near future. We believe the second wave of revenue will continue. The significance of that business is really that people need to justify why you don't want to use a computer vision on the server side, but instead trying to use a computer vision on the edge side. I think most of our customer realize that's a need, and it's just taking time for them to optimize the product to deliver to a customer.
I think that trend continues, and we don't see a problem that we're going to play a major, you know, a major role in that market.
Got it. Your comments on some of your customers having issues. I know you kinda touched on this in the last question a bit, but specifically for auto, do you see this influencing any of the more recent design wins you expect to see converting to revenues, you know, beginning early next year, just given what you're hearing and all the headlines in the markets, you know, suggest there's maybe some delays?
Well, I think delay is not because of short supply. It's really about how much volume can be delivered. I think that all of our customers show their desire to take the product into a market as quickly as possible. I think the number of shipments into the market will be limited by the supply situation. That's why it's hard to really size it up. From the production delivery point of view, I don't see much of delay. I think the biggest concern for all of us, including our customer, is whether they can get enough supply to meet all the demands they have.
Yeah. Okay. Thank you, Fermi.
Our next question will come from Kevin Cassidy with Rosenblatt Securities.
Yeah, thanks for taking my question, and congratulations on the great results. For me, you gave us a lot of information, a lot of good updates. One thing I just wanna know if you have an update on the CV5 progress.
Yes. Thank you for the question. We didn't talk about CV5 this time on the script. I think first of all, CV5 has been successfully sampled to many, many customer, security camera, automotive, and as well as some of the consumer products. Both hardware and software sample has happened. I believe that CV5 will be in production H2 of next year.
Okay, great. Along with that, you'd mentioned the 3- nanometer and 4 - nanometer or 4 - nanometer and 3- nanometer R&D happening now, when can we expect something like that to happen? Is that two years out, three years out?
Yeah. Well, I think we are focusing more on the 3-nanometer at this point because 3-nanometer is auto-grade, and that's essential, and that's where we focus on. The timing of we using 3-nanometer really depends on how fast the sensor can be matured. We definitely plan to kick off a test chip sometime next year to test out all of IPs and engineering development. The products really depends on how fast that we can get good yield out of 3 nanometers. During the meantime, we are very confident and comfortable with sensor 5-nanometer now, and we are going to continue to see multiple project going to a sensor 5-nanometer next year.
Okay, great. Thank you.
Our next question will come from Suji De Silva with Roth Capital.
Hi, everybody. You talked about supply chain shortages impacting your customers' components. Did you specify which end markets were being impacted, or was it broadly across the end markets?
It is across the board. In fact, every customer we talk to, they got a different type of impact. Some of them just a little, but almost all our automotive customers are talking about a different type of supply shortage. Microcontroller is probably the biggest one. On the consumer security camera, most of the consumer security camera, Wi-Fi supply is a big problem. I think that it's really across the board at different levels.
Okay. Thanks for clarifying that. I know you reaffirmed the 25% CV revenue contribution in 2022. I don't recall if in the past, you've talked about where you think that might go next year, 2023, or thereafter. Any color there would be helpful.
Yeah. We haven't addressed next year yet, but I definitely think we need to give you more guidance at the CMD time.
Okay. I look forward to that CMD. I don't know if you'd already provide some of that. I want a follow-up on Rivian. You talked about the wins and the car has 5 cameras and 10 radars. I'm just curious if that's the kind of proportion or maybe I got backwards. 10 cameras and 5 radars. The proportion of that is consistent what you're gonna see going forward. With the chips you've sold, the multiple chips, are you now causing other AI chips in the car to be de-specced or perhaps replaced by this? I'm trying to understand how new and perhaps forward-looking the design Rivian is.
First of all, yes, I think the ratio sounds right because first of all, the radar, the 5 positions for radar are really the 1 front and the 4 corners for radar. Some of them start using a radar for internal perception also. That's a unique opportunity. For cameras, I've seen customer going anywhere between 10 to 15-16 type of cameras moving forward. I think that ratio is probably stay there. However, from our point of view, you can see that we're handling quite a few cameras with only two, I would say three combined CV2 and CV22 chips. Moving forward, I think that will be unified to a domain controller type of chip.
I definitely think this is the first generation. In fact, we start working with Rivian, I would say three years ago, and took three years to reach a point. We do believe that moving forward, you will see a much less number of digital signal processing chip, but it's much more powerful to handle all of cameras and radars.
Okay. Thanks for the color, Fermi.
Our next question will come from the line of Joseph Moore with Morgan Stanley.
Great. Thank you. I wonder if you could talk about your progress in L2, in passenger cars, both in the U.S. or in the West and also in China. Do you see the biggest successes there coming from replacing the incumbent with single camera forward-facing, or would you anticipate that it's things like radar or stereo vision that's the new technology shift that brings Ambarella into those sockets?
I think the biggest near-term opportunity is to replace the incumbent of a forward-facing ADAS market because our current CV2 functional safety chip is well suitable for that, and now we can integrate radar solution into a CV2 functional safety chip and provide you know a combined sensor fusion on that chip, become even more powerful solution for our customer. I really think for near term, this is the biggest opportunity for us. However, that I think, based on the Rivian, you can see that similar type of design can be used in a consumer vehicle, but I think for the high volume consumer vehicle, you would require much more integrated solution in terms of number of chips and doing a video and the radar integrating to a single domain controller. I think that's probably a little further out for us.
Okay. If I could just ask a follow-up. I mean, it seems like the case for stereo vision is pretty compelling. You guys have been talking about it for a number of years. It seems like with the type of density and frame rate that you guys can pick up, that you can do really good distance triangulation at speed. You know, what is the barriers? Is it just price? Is it just sort of time for your customers to do the development? You know, what is it that kind of gets you into those wins longer term?
Right. First of all, I think the stereo is really trying to providing a different type of a sensor modality against LIDAR in a Level 4 car, a Level 3 or Level 4 cars. In a Level 2, when you're talking about smaller number of cameras, the price really become sensitive. I think for most of the, for people who are evaluating and adopting, the stereo processing is really on the become. They want to have a multiple sensor to do providing the depth information. I think that's definitely the limits the number of applications that can that requires stereo processing. However, I still believe this is a very powerful tools for us.
In fact, we do have projects that we already take that into production, and also that we, in the near future, can talk about more design wins we have with stereo. It just take a little longer time.
Great. Thank you and congratulations on the funnel.
Thank you.
Thank you. Our next question is going to come from the line of David O'Connor with BNP Paribas.
Yeah, great. Good afternoon, gentlemen, and thanks for taking my questions. Maybe for me, a couple on my side on Oculii. Firstly, can you talk a small bit around the positioning of Oculii radar from the processing perspective versus the existing radar processing guys, where you see the microcontroller sitting behind the sensor. And also then when you integrate Oculii into the CV chip, you know, what type of gross margin upside are we kind of looking at there, given the software component to that? And lastly, on Oculii, do you plan to maintain the module business there initially? Just trying to understand the synergy there of having that module business with the existing chip business. Maybe one final one for John.
I'm not sure, I joined the call a bit late, so, but did you quantify how much revenue you're leaving on the table given these supply constraints? Thank you.
First of all, on the radar side, we do plan to continue with the module business because, first of all, it's a very well integrated product, easy for any of our customer who want to evaluate our technology, so that we can easily give them modules so they can understand how the Oculii algorithm software works and to the evaluation. More importantly, there are some small volume customers, they probably will adopt our module and get into mass production. We do plan to continue to our module business.
From the difference, the differentiation between Oculii and their competitors, I think it really is Oculii, for us, is algorithm company and their unique AI algorithm, what they do is they try to generate this, we call intelligent transmit waveforms, and that adapts based on the environment. After those waveforms got sent out and on the receiving side, when we're receiving those return signals, we process those information close time so that we can improve the resolution. It's really about the signal processing algorithm that Oculii delivered in the last several years that which significantly improve the resolution both on the angular or horizontal, but more importantly, also on the depth of the on the range side. I think it's very few company out there or radar company out there is focusing algorithm side.
You see a lot of companies focusing on just boosting up on the hardware side, boosting up their transmitters, number of transmitters and receivers so that they can transmit a lot more signal, which also means a lot more expensive and demand a lot of power consumption. I think Oculii definitely have unique approach to solve this resolution problem that which is important for radar.
Yeah, David, you also had a question in there about gross margins in the Oculii business. The reason we're not prepared to give guidance on that is you're correct. The licensing side of the story, you know, would carry gross margins that are better than our long-term model. However, when you start layering in the module business, that can bring it back into our long-term model. Until we understand that mix a little bit better, I think it's appropriate to think of this blended as being in our long-term model, 59%-62%.
David, just finally on your last question, the short answer is no. We didn't quantify any Q3 loss revenue as a result of the supply chain constraints.
Understood. Very helpful. Thank you.
Our next question will come from the line of Tristan Gerra with Baird.
Hi, this is Tyler Bomba on for Tristan. Thanks for taking my questions. First, are you seeing a demand slowdown in China, and is that possibly going to extend from consumer weakness into infrastructure spending there? How do you see automotive demand trends in China currently?
I think both for China. I think in China, our focus is really on the security camera side, is really professional security camera. We don't work on the consumer security camera inside China. For the professional security camera, we talk about Hikvision, Dahua and a few more, the company that already start using our CV solution. So it's more of the professional side of business. On the automotive side, I think it's the number of units, at least if you look at the forecast for next year's, I think that the forecast for the Chinese automotive market also reduce in terms of total numbers. But however, the trend of that more people start using computer vision so they can for the advanced ADAS or AD solution are increasing.
I think that's kind of wash out in my opinion. We continue to secure design wins in a Chinese automotive business focused on DMS and ADAS. Like we announced today with Shaanxi that it's ADAS application for the commercial vehicle. I definitely think that we work hard to try to penetrate the consumer vehicle as well as the level two plus of applications.
Great. For my follow-up, in China, are the surveillance camera OEMs, the professional ones, primarily driven by government orders, or are they from the private sector?
I think both. However, majority of the, for example, Hikvision Dahua has been talked about a lot of their projects coming from government projects. That's also the reason they show a weak Q4 guidance, because I think government spending is reduced in Q4. I think that definitely is the main reason that Hikvision Dahua are doing very well in the last, I would say 10 years, is really about government-driven driving the whole security camera market in China.
Awesome. Thanks so much.
Our next question will come from the line of Martin Yang with Oppenheimer.
Hi. Good afternoon. Thanks for taking my question. My question is on your Axis giving first product with CV design wins. This, if I remember correctly, follows up with the first product for three products coming out of Bosch with CVs. Now that those two major IP camera security are adopting CV, do you see we reach a potential milestone where we see a much higher adoption of potential shipments of CV in the professional security camera market in, for instance, in the European market?
I think from the adoption point of view, if you look at it from the company point of view, the coverage is very well for us. However, if you look at the products they use CV for now, majority of them are still in the higher end side. It's really high-end product being converted to CV first. I believe that the biggest revenue opportunity for us moving forward is when our customer gradually moving downwards and putting more and more CV solution to their low-end products, then we will start seeing more unit number to our revenue mix. I think that because we provide a software compatible SDKs between our low-end to high-end CV chip.
Anybody who believe design is in for their one CV product, they can easily convert the product to a different high product at different performance and price point. I do hope that in the near future, we're gonna start seeing more product coming from all of those company using our CV solutions.
Got it. As a follow-up to that, I think, you know, comparing to Motorola or Hanwha Techwin, those two customers seems to have adopted CV in professional security camera products much earlier than the European guys. Is there any way you can tell from any, you know, product design or the way the early adopters are using CV chip, where you see can extrapolate the speed of the migrations of CV chip to the lower-end security camera products?
Yeah, I think you are right. For example, if you look at Motorola's script, the quarterly announcement, they do start expanding CV solution to many different product lines, which is faster than other company that we're seeing. I think Motorola, I would say, definitely the most aggressive one trying to adopt our CV on the high end and gradually moving down to different product line, like even for the, you know, wearable device for security guard. Those kind of application, they start using our CV chip too. I think Motorola is most aggressive. However, I do believe most of company out there will follow, because for them to continue to compete in the market, they need to have a similar offering, for the to compete with other companies.
Yeah, makes sense. Thank you, Fermi.
Thank you. Our next question is gonna come from the line of Derek Soderberg with Colliers Securities.
Hey, guys. Another question. I'm curious what you're hearing from your customers, and maybe it's customer specific. Are any of them giving you sort of a timeline for when they expect, you know, their customers, their other customers will return to normal? You know, there's been sort of a decent amount of reporting around the auto chip shortage. I'm just curious if, you know, you're hearing anything, like has anything gotten worse in recent months? Any color on that would be great.
You know, I tried to talk to a lot of people in the industry to understand the situation. The consistent message I heard is that they don't expect the situation will improve until Q2, maybe even Q3 next year at best. I think, however, we do start seeing signs that things changing a little bit, right? We do see that some component catch up. Even yesterday, I see some news about microcontroller situation in automotive improved. So I do hope that the situation is not as bad as others, everybody thinking that Q2, Q3, we'll see. We have to wait until Q2, Q3 next year to see some improvement. At this point, the consensus from the people I talked to is that this is going to be a problem for next few quarters.
Great. That's all for me. Thanks, guys.
Thank you.
With that, I see no further questions. I would like to turn the call to Dr. Fermi Wang for closing comments.
Yeah. Thank you. We are excited to be returning to Las Vegas in January to host both our Capital Market Day and the CES event. We'll be demonstrating a wide range of automotive security, access control, and robotic technologies, and operate our level four autonomous vehicles, and also provide our first demonstration of Oculii Radar software. We really hope that we can see all of you there in person, and thank you for your attention today. Goodbye. I'll talk to you next time.
Once again, we'd like to thank you for participating on today's conference call. You may now disconnect.