Arbe Robotics Ltd. Q4 FY2021 Earnings Call

Good morning and welcome to Arbe's Fourth Quarter and Full Year 2021 Conference Call. I will now hand it over to Miri Segal, CEO of MS-IR. Please proceed.

Thank you, Andrew, and everyone, for joining us today. Welcome to Arbe's Fourth Quarter and Full Year 2021 Conference Call. Before we begin, I would like to remind our listeners that certain information provided on this call may contain forward-looking statements, and the safe harbor statement outlined in today's earnings release also pertains to this call. If you have not received a copy of the release, please view it in the Investor Relations section of the company's website. Today, we are joined by Kobi Marenko, Arbe's Co-Founder and CEO, who will begin the call with a business update. Then we will turn the call over to Karine Pinto-Flomenboim, CFO, who will review the financials. We will then open the call for the question-and-answer session. With that, I'd like to turn it over to Kobi Marenko. Kobi, please go ahead.

Thank you, Miri. Good morning, everyone, and thank you for joining us today for Arbe's first earnings call as a publicly traded company. It's an important time for us at Arbe, and I want to begin by expressing our deep appreciation to our investors, customers, and partners who have collaborated with us till now. I'm also incredibly proud of the Arbe team who drove our 2021 achievements through their commitment and hard work. The recent completion of our SPAC merger and becoming a public company was a major milestone. In fact, Arbe is the first automotive imaging radar company to go public in the U.S. Since this is our first earnings call, I'd like to take a few moments to share our story, explain the core value proposition, and why we believe that we will lead the automotive sensors market. Arbe is the leading 4D imaging radar company in the market and the only pure-play radar public company traded on NASDAQ. We are driving a revolution, paving the way toward fully autonomous driving. We are a semiconductor company, and our imaging radar chipset offers the best performance in the industry, enabling free space mapping for safe Level 2+ and Level 3 autonomy that relies solely on radar. Arbe maintains significant product leadership, and we believe that our technology today is superior to the performance competitors projected that they will have even in 3 years from now. Our mission starts with safety. The first goal of our radar is to improve the safety of the entire vehicle system to allow automatic braking in any condition, to enable safe adaptive cruise control, lane change assist, and many other advanced features, scaling up with the industry's evolution towards full autonomous driving. Radars in cars need a change, and the entire industry understands that performance needs to improve dramatically. In 2025, the total market of radar will be around $11 billion, and a major part of it will belong to imaging radar. When we founded Arbe in 2015, the general industry consensus was that radar had already achieved as much as it could. It was widely believed that LiDAR was more promising for advanced applications. However, as time went by, it became clear that achieving high-performance LiDAR at a reasonable price was really unrealistic. Also, LiDAR is unable to reach performance targets at necessary ranges and does not function well in bad weather or when the sun shines directly on the sensor. In the past few years, the market has shifted back to radar. The majority of OEM car companies are beginning to understand that imaging radar is the key for any kind of Level 2+ and Level 3 applications and is mandatory for higher autonomy. We believe that in the next few quarters, we will start to see announcements from the largest companies choosing to base the majority of their stack on imaging radar. This is where we specialize, in imaging radar, which is a completely new sensor that drives all the advantages of traditional radar, while enhancing them with real, physical, high resolution in 4D on the azimuth, on the elevation, on the doppler and on the range. This allows our technology to support the development of perception algorithms and to deliver much higher value to the industry. Currently, and for some time now, we have the only ultra-high resolution solution in the market. In fact, we believe that our radar has the best performance in terms of resolution, dynamic range, and sensitivity even when it comes to future radars that are still in an early development stage announced by competitors. Our assumption is based on the specification of those radars which were released by the competitors. Today, we can proudly say that we added 30 customers from companies that have purchased a low volume of units to preproduction projects and full designing. Based on the subset of those clients, we expect more than $300 million in revenues in 2025. As we look back at 2021, I'd like to summarize our main achievements. AutoX selected imaging radar to enable their fully autonomous RoboTaxi fleet. Our preproduction project with a top 5 global OEM is on schedule, progressing towards selection in 2022. We were selected by a top 10 global OEM for an imaging radar-based perception project. BAIC Group announced that Arbe's technology will be installed on its mass-produced Level 2 plus, Level 3 models. We are shortlisted with 17 opportunities, including 8 OEMs, 5 Robotic Taxis, and new mobility players as well as full trucks manufacturers. Five Tier 1s, including a leading radar player that recently started working with us, are building production radars based on our chipset. And we made low-volume sales to more than 30 customers during the year. Our 2021 development and commercial objectives were fully achieved, and we're ahead of schedule for full production of our chipset by the end of this year. At this year's CES Conference, Arbe was selected as a 2022 Innovation Awards Honoree in the vehicle intelligence and transportation product category. Our perception imaging radar was recognized for driving a revolution in autonomous vehicle sensing at an affordable cost for the mass market. The highlight of our innovation was the introduction of free space mapping. This is the first time an automotive radar can contribute to free space mapping, a critical function of autonomous driving that requires great accuracy and redundancy. The imaging radar complements camera sensors at long range and wide field of view, sensing the environment in all weather, lighting, and visibility conditions. We believe that the ability to generate a free space map and enhanced perception algorithms is truly game-changing for the future of autonomous driving, and we are enabling greater safety than even before for drivers, pedestrians, bicycles, and anyone sharing the road with autonomous vehicles. I'd like to highlight some of our recently announced collaborations with major car manufacturers who plan to adopt our radar. Recently BAIC, one of the largest players in China and one of Arbe's investors, announced that our sensor will be brought to the market by hiring a Chinese Tier 1 and will eventually be in every one of BAIC's Level 2 plus and Level 3 cars. AutoX, the leading RoboTaxi in China and one of the first movers in the United States announced that all of their cars will use Arbe's radar and anticipated 400,000 units in the next 5 years. A few months ago, one of the top 5 car manufacturers selected Arbe and Valeo as Tier 1 for a preproduction project that will lead to design selection next year. In November, Weifu High-Technology Group, a major Tier 1 in the Chinese market, announced that it began the road pilot phase of its Arbe-based radar system and is scheduled to be in full production by the end of 2022. We believe that it could become one of our flagship projects. Hyundai Mobis recently announced that it is conducting a large-scale pilot with Arbe's technology for Level 2+ and Level 3. Hyundai is also an investor in Arbe. This pilot was successfully completed and the final radar selection will take place in the coming months. Additionally, Qamcom, Arbe's Tier 1 for the nonautomotive vertical, is making impressive progress, holding pilots with Robotic Taxis, autonomous trucks, and delivery robots. As we look at 2022 and beyond, we are optimistic about the long-term future and market needs to improve safety by using Arbe's superior solution. While experts assume that the chip shortage and supply chain challenges will continue to impact global markets in 2022, it's important to understand that one of Arbe's major advantages is our relationship with Global Foundries. As one of their strategic customers, we are the first company to go to production with chips that are automotive-grade based on the 22-nanometer process. Global Foundries has taken full responsibility for chip production and supply chain management. They are guaranteeing the supply we promise to our customers for the coming years. Despite the current silicon shortage, we will have the capacity necessary to meet our commitments. In terms of our goals for 2022, we aim to close 8 design wins with OEMs, autonomous trucks, delivery robots, and nonautomotive companies this year, and we are laser-focused to achieve this target. Based on current market trends, we estimate that Tier 1s will show increased demand for Arbe's chipset. Our goal was to establish a relationship with 2 leading Tier 1s this year, and one of them is already working with us, and we are only in the beginning of 2022. A major strategic goal for 2022 is to go to production with our chips. This is an ambitious plan and is supported by Global Foundries. We believe we can reach that milestone. Our Tier 1 partners are following closely, and they are in the advanced stages of a plan that will take a full radar model to production in parallel. In China, we will be in production by the end of this year, as our Tier 1s have announced. And in Europe and the U.S., we believe it will be in the second half of 2023. To summarize, we are confident in our vision for this market. Everything we see leads to the following conclusion: high-definition imaging radar is about to change the auto market, and then it is going to work its way into many other markets and change them as well. We believe that our current solution is well ahead of the competition. We intend to gain a significant market share as leaders and innovators in this space. I'd like now to turn the call over to our CFO, Karine.

Thank you, Kobi, and thanks to everyone for joining us today. I would like to review our financial results in more detail. Total revenue in the fourth quarter was $0.5 million compared to $0.2 million in the fourth quarter of 2020. For the full year of 2021, total revenue was $2.2 million, an increase of 577% compared to $0.3 million in 2020. Backlog as of December 31, 2021, was $2.3 million, consisting mainly of services, chipset orders, and NRE from several leading car manufacturers and Tier 1s. Gross margin for Q4 2021 was 37.7% compared to 41.8% in the same period in 2020. For the full year of 2021, gross margin was 36% compared to a negative 2.4% in 2020. The gross margin increase was primarily related to economies of scale. Moving on to expenses. In Q4 2021, we reported total operating expenses of $14.2 million, an increase from $3.9 million in the fourth quarter of 2020. The increase was primarily related to additional investment in R&D towards production, labor costs, share-based compensation, and costs resulting from the additional legal, accounting, and general overhead resulting from our status as a public company. Operating expenses for the full year totaled $34.1 million compared to $15 million in 2020. The increase was in line with our expectation, as we continue to grow the company and add to our employee base to support our future growth. The company continues strengthening its research and development investments, with R&D expenses totaling $11.6 million for Q4 2021 compared to $3.3 million in Q4 '20. R&D expenses for the full year were $28.6 million compared to $12.8 million in 2020. Looking at adjusted EBITDA in Q4 of 2021, which excludes expenses for noncash share-based compensation and for nonrecurring items, was a loss of $11.9 million compared to a loss of $3.7 million in the fourth quarter of 2020. EBITDA for the full year of 2021 was a loss of $30.4 million compared to a loss of $14.2 million in 2020. Net loss in the fourth quarter of 2021 was $15.8 million compared to a net loss of $4.5 million in the same period of 2020. Net loss for the full year of 2021 was $58.1 million compared to $15.6 million in 2020, which included $24.8 million of financial expenses mostly related to the revaluation of convertible loans and warrants. Moving to our balance sheet. As of December 31, 2021, Arbe had $100.8 million in cash and cash equivalents. Most of the cash is a result of the $118 million in gross proceeds following the business combination with ITAC, which was completed in October 2021. With respect to our guidance for 2022, revenue is expected to be in the range of $7 million to $11 million. Adjusted EBITDA is expected to be a loss in the range of $34 million to $38 million. We anticipate rapid growth following 2022 as Tier 1s move to mass production and OEMs scale up order volumes. We are on track to reach our $312 million revenue goals for 2025. Now I will turn the call back over to Kobi, and then we will be happy to take your questions.

Thank you, Karine. As we move into 2022, I'm confident in our strategic plan. Our product is leading the market and our customers are fully engaged. This is the right time to join us on this ride. We look forward to updating you on our progress next quarter. Thank you, and we are now ready to open the call for questions.

The first question comes from Joshua Buchalter with Cowen.

Congratulations on the results. In the press release, you mentioned the project with the top five OEMs moving towards selection in 2022. Could you provide more details about the process, particularly regarding the timeline and how competitive it was?

Yes. So basically, we are not really allowed to elaborate on this project since we are under NDA with the customer. But what I can say is that, as you mentioned, the project is on track. We, together with Valeo, need to provide the customer with 2 kinds of radars, one for the front and another for the side. This should be done during this quarter, Q1. We believe that selection is going to be in the second half of '22.

Thanks for the detail there. And I understand, on that note, there's only so much you can share. But any more you can give us on the perception project with the top 10 global OEM? Is it a similar type of application and volume potential as the top 5 preproduction project?

Yes. Essentially, the top 10 OEM is aiming for similar goals. They are looking to develop a radar capable of complete free space mapping. For the project with the top 5 OEM, our role is limited to providing the radar itself, which facilitates this free space mapping. In other projects, in addition to the radar, we are supplying our software stack for free space mapping as at least a reference code for the OEM. The radars are already integrated into their test vehicles, and they are operating with our software. This allows them to incorporate our software into their full system and evaluate how free space mapping from the radar enhances the overall system, particularly through fusion with cameras, and how it addresses the challenges associated with Level 2 plus and Level 3 automation. We anticipate that the selection process will be completed by the end of this year.

The next question comes from Suji Desilva with ROTH Capital. The radars are already installed in their testing cars, and they are driving with our software. They can take our software and connect it into their full stack to see how free space mapping based on the radar is improving the total system, especially with the fusion with the camera and how it helps to solve the entire Level 2 plus, Level 3 problems. Again, we believe that selection is going to be by the end of this year.

Congratulations on the progress in calendar '21 here. So Kobi, on the Tier 1s, the 5 that you have, do you expect all of those will be in production readiness at the end of calendar '22? Or will they phase in at different times? And I'm just curious, are they all leveraging Global Foundries' supply chain?

So, first of all, by the end of '22, we believe that the Chinese Tier 1 will be in production. The European and American will be during '23. Yes, basically, everyone that is buying chips from us is buying the chip from Global Foundries, actually from us, but the supply chain is managed by Global Foundries. So Global Foundries is our only software production, and the chipsets will be shipped to the Tier 1 from Global Foundries. Technically, by the way, it will pass through our Israel office, but just touch and go to the customer.

That's helpful color. And then in your calendar '20 revenue, I know you talked about AutoX being a significant number of units over the next 5 years. Is any of that in calendar '22? Or is that just to answer that in the guidance for '22, or is that starting calendar '23?

Although it will start generating meaningful revenues already in '22.

Got it. And lastly, can you just talk about the decision to choose imaging radar versus the traditional radars? Just give us a thumbnail of the OEM's decision points. Are they just all doing studies to validate that imaging radar can work and be perhaps even a lighter alternative? Is that what's going on in the OEM and Tier 1 base now?

Yes. So I think that, basically, there are 2 trends here in the market that are moving in opposite directions. So the regular radar, what we call a 3x4, 6x8 radars, those radars with low number of transmitters and receivers, which has also a low amount of pixels and low resolution, those radars that are basically part of any vehicle manufactured more or less, those radars are basically going into commodity mode. So their price is going down. The chips behind it, the i-Chips and NXP and Infineon, the prices of those chips are going down. The processing power that is needed to process it is the same. It's a commodity today, and there's no real innovation in those kinds of radars. On the other hand, the car manufacturers are beginning to understand that those radars won't serve them for Level 2+, Level 3, and more and more OEMs are taking decisions to put more advanced hardware in their cars, even if the software stack for Level 3 is not yet there. So those radars will be there to enable the system to be in Level 3, or Level 2+ with just a software update over the year. So the advantage of 4D imaging radar as opposed to traditional radar is that the imaging radar is actually generating a picture of the environment almost as good as a camera. This is the picture that almost a human being can see and understand what is going on, as opposed to the traditional radar that basically is based on Doppler. So they see the Doppler, the relative speed of the object. And they basically have the ability to identify the objects based on the fact that they have the Doppler. So if they don't have a Doppler they won't be recognized. This also creates some kind of misunderstanding regarding braking phenomena with the post target that those radars are generating. I would say that, within the last few months or maybe till 2021, the radar was not part of the perception stack. So the perception stack was built mainly on camera, so vision. The radar was used mainly for blind spot detection and automatic braking. There was no real connection between those 2 stacks, which created a huge problem because the vision sensors, including LiDAR, which is a laser-based sensor, are vulnerable to weather. Also, they cannot see a target at long range. They are very good at 50 meters, 70 meters, sometimes even 100 meters, but after that, they are useless. To get to a real safe Level 2+, Level 3 must have a kind of sensor like the radar that can see up to 350 meters, can see a pedestrian at 200, 250 meters, a motorcycle at 300 meters and have good resolution and work in any weather and lighting condition. Now, there's a huge effort in integrating this type of sensor into the full stack. Our free space mapping is actually a shortcut for the OEMs to do that. We are coming with this stack, saving OEMs years of development from scratch for this technology because without it, they would just need to take our radar and start the development now because our radar is the only one that you can take the point cloud from and generate a full free space mapping. Earlier, we mentioned the top 10 OEM that selected Arbe for the perception project. We were competing against another 1 or 2 radars, with an order of magnitude better. Actually, all the other radars were not qualified for this perception task. I believe that '22 is going to be a year of imaging radar perception. The fact that it's doable, that you can take an image out of a radar, merge it or fuse it with the camera, and get a point cloud that is 100% reliable, 100% safe is a breakthrough for Level 2+, Level 3, and I think this will even accelerate the acceptance and readiness of these technologies to the market.

The next question comes from Matthew Galinko with Maxim Group.

Can you discuss the gross margin potential or the difference between selling just the radar and selling the complete perception stack?

Sure. So first of all, we are not selling the full radar, we are selling the chipset. We are a semiconductor company, and as a semiconductor company, we have around 60-something percent gross margin. Of course, the software stack is software, and the gross margins in software are much better than in hardware. But the price that you can get per car is limited, I would say, very few dollars. In our model, at least until 2025, the software is not part of the revenues. Right now, the software is just an expense, and we are just investing in R&D for that. Our entire model is based on selling chips. By the way, we believe that the software stack will find its way also into the silicon in the next generation of our processor that's going to be in the market in 2025. We will also support deep learning algorithms that will help with this perception, things that today we are doing on the central unit. By that, we will reduce the total cost and the total power consumption of the car.

Got it. And maybe just to go a little bit deeper into that, is there a significant latency benefit to integrating the software into the chip? And is that important in higher levels of autonomy?

No. So the software is a different stack from the chips. If you take our chips, basically, we have 3 chips, transmitter, receiver, and a processor. The processor is generating a point cloud, something like a video. This is going to the central processor of the car, and then all of these post-processing tasks, like the free space mapping, are done on the central unit of the car. There's no latency here. It's Ethernet 1 gigabit per second. The data moves very quickly, and there's no latency caused by that. What we're trying to achieve with our next generation of the chip is to reduce the total processing power that is needed to enable autonomous driving. Right now, today, the central units like NVIDIA, Qualcomm, and what Mobilis is offering cannot really deal with the amount of processing power needed for full autonomous driving because they are doing some processes like we can do on our chip.

That's very helpful. And then last one for me. Just in terms of the guidance range for revenue for 2022, can you just cover how you get to the high end versus the low end? Is that just mostly timing on how things move in the pipeline? Or what gets us to the top or the bottom?

I would like to share some insights regarding our revenues for 2022. Essentially, this year serves as a precursor, as we anticipate full production in 2023. Once we reach full production, we will have specific timelines for deliveries, and our guidance will become significantly more precise. Currently, we are selling a limited number of evaluation chips and charging non-recurring engineering fees. Under these conditions, revenue recognition hinges on milestones achieved by the customer. For instance, if we receive a non-recurring engineering fee of $1 million that has already been paid, this amount is included in our backlog. However, if the customer postpones the production of radar units using our chipset, causing them to enter the market later than planned, the revenue will be deferred accordingly. When we provide guidance, we indicate that our sales team will handle the sales and our delivery team will manage the deliveries, but this year, due to the lack of full production, we are reliant on the customer's schedule for revenue recognition. Consequently, we have established a range for full and partial revenue recognition. Nonetheless, from a business standpoint, the difference between $7 million and $11 million this year is negligible because the sales and orders will be there either way. Whether we achieve $7 million with more bookings or $11 million with fewer bookings, there is no real impact from a cash or operational perspective; it’s primarily about how we recognize the revenue.

Just to add, I think closing the gap going forward to 2025 will be, of course, closed by mass production, and we anticipate, of course, to reach the revenues that we provided.

So, I guess, not to put words in your mouth for 2023, but we'd be seeing just more traditional product revenue and less NRE come next year?

Exactly. So as we move into production, there's no NRE. Just delivery...

Of course, towards, I assume, '24, '25 in the back end of Q3.

The next question comes from Jaime Perez with R.F. Lafferty.

We've been hearing many OEMs mention that they expect the supply chain shortage to extend into at least 2022. I understand you've started to ramp up production recently. Given that customer demand for automobiles remains strong, do you think there will be pent-up demand that will lead these OEMs to require your products due to their production backlogs? Do you think you will be able to manage that situation? That's my first question.

When we ship chips in 2023 to our Tier 1 customers, let's assume these shipments occur in Q2. By the end of 2023, they will deliver a complete radar model to the car manufacturers. These cars will be available to customers in 2024. Therefore, the supply chain issues we face won't impact our results in 2023 and 2024, as the chips we are shipping now are intended for vehicles set to be sold by the end of 2024, which aligns with part of the 2025 model year. In China, it may happen a bit sooner, but the timetable suggests that the current shortage isn't significantly affecting the overall number of cars produced. Additionally, high-end vehicles are where new technologies, like imaging radar, are focused, as these models offer higher profitability. Manufacturers are prioritizing their supply chains to meet the demand for these higher-margin cars rather than those with lower gross margins. Overall, while supply chain challenges exist, I don't foresee them impacting our results in the coming years.

So based on your business model and production ramp-up, you basically missed the next couple of quarters of the supply chain issues. Is your timing going to be pretty much on track? And the second question on, have you looked at any nonautomotive applications?

Sure. So there are new automotive and nonautomotive applications. Delivery robots are, I would say, new automotive. It's not really clear yet today if those robots need a full automotive-grade component, but we are actively in this market. The same goes for other applications like shuttles. We are also in some projects of nonautomotive applications like smart cities, mining, heavy machinery, and agriculture. In all of those markets, there is potential. We have a partnership in this market with a large Swedish company called Qamcom, and they are basically supporting the customers in those low volume, high-margin markets.

Alright. Due to your lower competitive cost, it is likely very profitable to incorporate Arbe Robotics radar into projects such as robotics or drones compared to other technologies that can cost a couple of thousand dollars.

Yes. When we started our vision, it was to enable robotics and drones, and we are still committed to this mission. We believe that those verticals are valuable for us. However, automotive remains our main market because of the volume and because of the value that our technology brings to automotive.

This concludes our question-and-answer session. I would like to turn the conference back to Kobi Marenko, CEO, for any closing remarks.

So thank you, everyone, for joining us today. We look forward to seeing you in our next conference call. Have a great day. Thank you.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.