Omer Keilaf:
Now, um, le- let’s split between, uh, premium, level 3, high speed, to level 2+, low speed, level 3, uh, and- and adoption. So, may- maybe I’ll- I’ll ski, I’ll- I’ll jump to, um, slide that might help me here. So, okay. So, the way that we see the market is that, you know, initially the level three premium cars are- are able to, and willing to, sell ... and to buy LiDAR at around $1,000 in volume. Uh, when it goes to level 2+ and level 3. And then in 2030, uh, the- the price would need to be around $500. And you need to include here that it’s not only the technology. You need to take into account also, uh, the- the assembly, the Tier-1, the warranty, the liability, all of those, uh, (laughs) less, uh, le- less, uh, appealing topics in automotive that nobody likes to talk about, uh, but are important.
Omer Keilaf:
So, um, the- the bump out of- of these figures should be around, uh, 30% to 40%. Uh, because, uh, on top of it you need to- to add, uh, the margins for, uh, for the, uh, for the company, for the Tier-1, uh, and you need to add the- the assembly cost, um, et cetera. Um, the way that we work with the market, and again, uh, you know, our business model is which we sell components, uh, to the Tier-1 is very much like what Mobileye did. We sell, uh, the ASIC, the MEMS, the detector. Um, those are a- a certain portion of the, of the, of the bump and that for those we do take 50% gross margin, but that’s for a cheap model, it’s not for a LiDAR model. Uh, for a LiDAR model, the gross margin would be, could- should’ve been, uh, lower, obviously, in order to meet with these, uh, price, uh, targets.
Jeff Osborne:
Got it. And then, um, you know, how- how important ... and I think there’s a debate about, uh, at least in my mind there’s a debate about software, uh, for LiDAR companies relative to some of the companies that I cover. So, for example, uh, Aptiv is an investor in your company but also is working on sensor fusion in their own right. Uh, and so, how do you think about what the expectations are for your own software relative to what Tier ones, which you have four of, uh, what they’re developing, uh, in-house for their own needs around using software. Do you, do you see over time, the- the democratization of hardware and software decoupling? Uh, certainly, you know, Mobileye has locked vendors in, into their walled garden of buying their camera and the re- result in software.
Jeff Osborne:
I’ve heard a lot of complaints about not getting the access to the data a- at the speeds that they want or, you know, in the format that they want. So, I guess, do- do you see any, uh, debate about, uh, the decoupling of hardware and software or not really?
Omer Keilaf:
So, I- I- I would split that answer to the different, um, parts of the, of the, I- I would say the vertical. So, when you talk about consumer vehicles, uh, there is a need to provide a LiDAR with the software. Of- of ... when- when I talk about the software, I talk about the computer vision of the LiDAR, on- only. Okay? So ... And- and the reason behind it is because car makers wants to meet with the ISO 26262 and- and to have an ACLB system, which means they want to have a complete a-a very clear separation between the decision making of, of each sensor. So this way they get, uh, the redundancy. So f-for this reason, uh, th-they are asking the computer vision to be relying solely on, on the LiDAR, and doesn’t, don’t take any information from the camera. Otherwise, uh, the validation process is just, uh, would be crazy and very expensive. Uh, when we talk about robo-taxis and kind of applications that, uh, probably, uh, I would say, less of automotive grades, that kind of mind thinking.
Omer Keilaf:
Uh, on those we do see, you know, companies that want to do the low level fusion, meaning that they take the road data in the different sensors and make, uh, one good solution. For the passenger cars, I believe it will, uh, be, uh, as, as, as I explained earlier long term, even if they would turn into low, high level fusion, they would need, they would still, uh, would like to have a secondary kind of um, uh, fall back, uh, uh, architectural to deduce the redundancy with software. Uh, for a, uh, kind of mobilization, of, of, of data, which is an interesting question. So I, I think that the car makers are worried to, to be in a position where they are highly dependent on the, on the sensor company to, to b, to b, to be actually, uh, providing them all of the feature. Uh, but I think that for them to really develop internally their own computer vision just on the light, that is not economically correct. Uh, we, uh, you know, the, the level of investment that we need to do in order to, uh, collect all of the data, and update, and, and, uh, you know, bring it to the right maturity, uh, we can upload that on several programs and for the same reasons that car makers do not develop their own LiDARs because they can not provide sufficient volume to bring it to the right price. Uh, they would need to rely on LiDAR companies to do it for them. What they are trying to do is to take ownership on kind of like the driving decision and, and do the integration of those sensors. This is where we see most car makers do not want to kind of, uh, um, outsource that kind of, um, feature.