Walt, it's becoming a trend here in Autonomy Markets. Another week, another big week for Waymo. There's news reports that Waymo has their supply. The question we'll find out later is Walt happy. Then we had earnings from Aurora, Lyft, and AV ride owner Nebius. And there's some news happening on the foreign autonomous we're going to dive into. But let's start with the big news. It looks like Waymo's got their supply of vehicles. Is Walt happy? I don't know if supply and demand are going to quite be met because, as you know, I'm very bullish on the demand for safe, clean, reliable, and predictable rides from autonomy. But yes, Waymo got cars. There's a publication I've never heard of before called, I think you pronounce it Gusu. Hopefully they're reliable. No one's come out to really deny this, but apparently there's a deal in the works between Waymo and Hyundai to purchase 50,000 cars by 2028. So, look, we already knew that they were getting cars from Hyundai. It was, you know, in some of my expectations for the number of cars that Waymo was going to have in 2026, we didn't really expect them to hit, you know, production, meaning, you know, line-fit production of these cars till the very end of the year, early 2027. So, this was definitively a new data point and a lot of cars, 50,000 by 2028.
A lot of cars. And this week also, there's been multiple photos on X and Reddit of Waymo testing the Hyundais in San Francisco. So, that's a really good positive sign. What I'd love to see is us get clarity on, are they going to be line-fit? Are they going to go to Magna? Are there going to be sensors? Are they going to basically get shells? I'd love to get an official release from Waymo of how these vehicles are going to start coming off and eventually, will they be fully factory-produced and being able to go into service minutes after coming off the line?
A lot of great questions that popped into my mind as soon as I saw this story. I still don't think that you're going to see many out of them in '26. I think this 50,000 is mostly going to hit in '27 and '28. I mean, for Waymo, I'm only expecting about 4,000 new cars for them to receive for this year. And again, only a portion of that to be Hyundai, the rest being Zeekr, and I think some residual volumes from JAG. And just look, to put this 50,000 in context, I have been forecasting for the entire industry in the United States, the entire autonomous industry, that by the end of 2028, there'd only be 55,000 cars total at the end of '28. And some of our peers on the street, I think, were even lower than that. And by the way, my estimate assumes very little from Tesla. That's something I'm holding back on until I see how Austin starts to scale. That's obviously going to be the turning point for Tesla. So that 50,000 alone, you know, from Hyundai is a big number relative to street expectations.
I'll summarize it this way. Once we get an official confirmation, probably in the form of a blog post from Waymo, I'll declare that Waymo just threw the gauntlet down on the industry. That 50,000 is a big number. The next domino to fall to further strengthen their thing is expand on that Toyota relationship. Now, Waymo's cooking with a full tank of gas, or they're electric, so a full charge.
Cooking with ions. There were some financial data in there as well. They talked about a $2.5 billion deal for these 50,000 cars. Simple math would indicate that's $50,000 per car. I'm guessing that's just what they're paying Hyundai and that that wouldn't include the cost of their sensors. What do you think? What we do know, Waymo has been actively working to reduce the cost of their sensors and compute. And if you look at, you and I were to go buy a Hyundai Ioniq 5 on the lot today with incentives, you could probably get one around 32 to 38,000. So based on that, there's a very good indicator that the sensors are built into that. Again, we have no confirmation, but based on the retail pricing, there's a very good confirmation because you have a plus or minus $20,000 buffer. I mean, if that's the case, then what we're talking about here is more likely a line-fit situation where the sensor kit is effectively purchased by Hyundai, integrated into the car, and embedded into that $50,000 cost. Again, no idea. That would be certainly an interesting detail. And at that price point, I mean, I've run enough models to know that for all those that were questioning the ability to drive a profitable business, if you're talking 50,000 a car with the sensors, obviously, if it's line-fit, you don't even have to deal with that Mesa plant incremental cost there. That is a great number for Waymo to exploit.
And what Waymo did, there's a certain faction of retail investors that said Waymo can never get it sub-$1,000. They just blew that narrative up overnight.
Well, look, I mean, that's the type of stuff that I heard in Starlink. They're never going to get their CPE down below a thousand dollars. I don't, you know, you follow technology for enough decades, you know how prices come down as obviously that you scale up. And 50,000, great number, you know, still we're going to see materially larger scale over the years. So, these are numbers that I think can probably continue to drop, but again, no confirmation of that. For all we know, the 50,000 does not include the sensor stack. A little more context there. Our friends at Nuro, I think their sensor stack is probably 10, maybe $15,000 from what I understand. I mean, I think, you know, this is the kind of incremental cost, the ballpark that people should think about if they're trying to question whether autonomy can challenge the economics of the human-driven ride-share market. If you, another thing that needs to be looked at, if you look at multiple analysts and banks have published this data, what's the cost of LiDAR now? Down 98, 99% since the early days when David Hall founded it. So you and I know, and you know from your experience, this cost, technology starts here, goes down, eventually gets commoditized.
Well, that wasn't the only news from Waymo this week, Grayson, as it relates to this topic that we've been, maybe more so me than you, but I think both of us have been focused on volumes. You know, Waymo had a post on their website. I guess you would call it a blog post, an update, really just announcing sixth generation. There was a lot of stuff in there that we'll get into, but let me first focus on what stood out to me immediately. Tens of thousands of cars are going to go through that Mesa plant again. So, this is implying upfit. This is different than I think what we believe is the case with Hyundai. Tens of thousands of cars is what they're ramping to. So, in my mind, this is probably the Zeekr, which is now they're coined, spelled with a J. These are Chinese vehicles. So I think there are some restrictions by Commerce and about tightening, you know, vehicle supply chain and not allowing some of this US technology to be installed there. So these are going to be purchased, you know, by Waymo, taken to that plant, upfitted with these sensors. But to get to tens of thousands of vehicles incrementally now with Zeekr again just adds to this addressing the supply questions that we and I think many have had in terms of Waymo.
I mean, I'll use a simple, silly term. Waymo's open for business. It's very clear that Waymo is focused on scaling. I still do feel that the Zeekr, it's not the Ojai, it's the Zeekr rebranded, was still an unforced error. But I really love the bullishness coming out of Waymo. And the unsung hero that's really enabled Waymo to scale over the years, in my opinion, doesn't get enough credit is Magna. Magna is one of the world's best contract manufacturers, one of the world's best outfitters. And the work that they've done on the JLR Jaguar I-PACE has been absolutely incredible. So no matter which way Waymo goes, if all factory, I'd love to see some sort of relationship with Magna continue because the engineering that Magna's done has been top-notch. So there's no timeline on either of these things, right? So tens of thousands, I don't think your Zeekr is going to hit tens of thousands in '26. Like I said before, I don't think Hyundai, those 50,000, I think a very small percentage of that will hit in '26. With that said, with these two announcements, do I need to rethink my expectation for Waymo? Right now I'm at, I think I have them adding 3,500 cars in '26. More than doubling the number of cars they ended 2025, which was I think about 2,500. You think I need to rethink that and take that number higher, Grayson?
I do based on the announcements. I do. But I will say that with the caveat you could get some executive order from the Trump administration that effectively bans the Ojai. That could affect your numbers. But the way it's currently written, I would consider putting your number to 4,200 to 4,500. I will say I'm not going to, I'm not willing to do it yet. I want to start seeing those Ojais actually turned on with customers, which has not yet happened, right? We see them, plenty of videos and pictures of them on the road, but hasn't really been turned on. But like Tesla, you need that kind of hit, that step before you start to scale up the sensor installment.
The Ojais, we did get an announcement, I believe it was an X post, that they are testing them with employees and trusted friends, if you want to use the term trusted friends, I don't know the proper term to use, that Waymo has invited to go in the vehicle. The other thing that we need to watch for with the Zeekrs, the Zeekrs will not be deployed in every public market that Waymo has, I believe based on the political narratives around the country, the Zeekr will be placed in more left-leaning states than right-wing states because of the China issue, though. So, that's something to watch there. I'll give you an example. If you look at the official announcement that Waymo made for announcing the Denver market, it was announced specific language and photograph of the Zeekr. Does Denver become a Zeekr-only market? Do we start to see that evolve? That's something to watch as well.
You know, I said before in my coverage of the wireless industry in the early days, pre-Bid, pre-revenue, even to a certain extent. The early signs on things like units give you greater confidence in your longer-term estimates. So, for example, like if I went back to, you know, '95 and I tried to forecast what I thought the number of, you know, phones that people would have in their hands 10 years, I way undershot it, right? And it turned out being much bigger. And some of that came to realization as they started to roll out. So, let's turn back to these car levels again. Going 55 was my estimate for end of '28. Clearly, Waymo alone is going to top that number. You know, getting, if you just kind of model it out, getting to 750,000 cars in less than a decade doesn't seem like heavy lifting. And if you think about 750,000 cars in the US, which the context is ride-share companies, Uber, Lyft, you're talking about a 2.2 million-ish range of drivers, obviously aren't on the road all day. So that 750 equates two to three times that size based on their availability. Now, there's peak hour things that have to be addressed, but I mean, that is a substantial number of cars. It could probably take, you know, half the ride-share market in terms of miles driven. Half right now. Look, the ride-share market, as bulls have pitched, could expand dramatically. So, maybe it's less than that in a much larger market, but this is all happening in less than a decade. And I just want to underscore, sorry to pick on it, but like, you know, in the last couple weeks and months, you've had Darren and David Richer out there saying that they couldn't scale, the autonomous industry couldn't scale for decades, plural decades. And you know, I think the trajectory that they're on right now is they can get to decent scale that takes a substantial portion of a reasonable expectation of the market in less than 10 years.
You're seeing, and this is just me, I believe you're seeing a massive disconnect in the market between myth and reality. And then when I say that, you're seeing Waymo scale at a pretty incredible pace. You have Toyota lurking over here, which everybody knows can scale faster than any company in the world when it comes to scaling vehicles. And to say that this technology is going to be delayed, it's not. I'll give you a wireless analogy. I'm going to go way back in wireless world with you. I don't know why I'm going there, but I'm going there. I remember Omnipoint way back in the day with Omnipoint and the GSM. And there was a CDMA that Verizon had, oh, they can never scale GSM. It's a European thing. And you know this better than I do. I was just a customer. Boy oh boy did that Omnipoint GSM network in America scale faster than I think anybody really imagined. The same thing is going to happen with autonomy. It's not just Waymo, that Lucid vehicle. They have the capability to produce thousands of those vehicles at the plant in Arizona. And when these vehicles come online, despite what the market thinks and what certain individuals on certain podcasts think, autonomy is going to disrupt the traditional ride-share market. It's not necessarily going to expand the pie. It's going to really disrupt it and I think it is an existential threat to the traditional ride-share business.
So as our listeners can see, we can barely contain our enthusiasm that we got this news out of Waymo. Hopefully both of the things we're reading properly. There was one small disappointment in this context of, you know, what came out of Waymo this week. The language in there made it pretty clear to me that Waymo has no intention to make their own car. I forget the exact quote. I think I tweeted, if you can look at my Twitter account for the exact quote. So, my hopes for maybe Waymo going to a contract manufacturer and making their own car is dashed. So, I think what's next for Waymo is, you know, getting an OEM. And you know I can already hear the criticism. Well, they just got 50,000 and they just got Zeekr. I hear you. But you know, starting a process with a new OEM now is not going to yield cars obviously for at least 18 months, right? Two years maybe, especially if you're looking for low-cost line-fit type implementations with Western manufacturers. I don't think we want to keep relying on this Mesa plant in Phoenix when you really want to hit the scale that's going to get to the 750,000 cars that I'm expecting in less than a decade, much less than a decade. So, that's what I'm hopeful. I'm disappointed that they won't just make it. That's fine. Let's get another OEM on the tape. You know, they had that Toyota announcement that you referenced earlier. Nothing there that has come of it as of yet, but there's plenty of other OEMs that maybe are starting to feel incremental pressure that hey, they got to start thinking about autonomy even though EV was a kind of a bust for them, that autonomy could be their future.
It is going to be their future. I will say I know it was discontinued, but I was a fan of the Firefly car. You can go visit it at the Computer History Museum in Silicon Valley. It's really, really cool, really great design, but it was the right decision to move away from it. As Casser, the CEO of Applied, says, building and running a car company is a hard thing to do, and that's not what Alphabet's strengths are. And Casser is right. He was there for many years before starting Applied. So, I like that. If you look at the global automotive business, you have two companies that come to mind, both Japanese, that I believe could make moves. One, you have Honda. They were a former investor in Cruise. They like the space. They currently have a relationship with Helm AI that Vlad is running. But so Honda could very much be in play. And the other that needs to make a strategic move, Nissan. Yes, Nissan has the relationship with Wayve for the ProPILOT 2. That's an L2 system. However, Nissan does have excess capacity in certain Japanese plants to start ramping up autonomous vehicles. And why do I say Nissan, Walt? There's been reports over the last couple months coming out of the Nikkei out of Tokyo that Nissan was looking to sublease those plants and/or sell those plants to Foxconn to build frames for autonomous vehicles. So where there's smoke, there's fire.
Grayson loves to cite the Nikkei. I've had mixed success with some of their information in the past. No, those are great ideas, OENS. But Grayson, I'm going to take it a step further and say, look, if Uber's out there and others are saying, you know, the bull case is that this dramatically expands the ride-share market because it's lowering cost, expand the market, expand the market. That's the mantra, right? Whether it's short-term or long-term. If you're expanding miles driven ride-share, dropping the cost, like that is potentially taking away from private car sales, right? Reducing the fact that that second or third car in your garage is going to happen because you'll just use ride-share instead. Otherwise, how does that whole concept of ride-share market expanding? If you're whiteboarding this at the OEMs, and I know that there's OEMs that listen to this, like you have to consider that. That is another pressure point. It's not about just doing a Waymo deal to sell cars into the Waymo ride-share market, you know, which will represent like if you know the X million cars a year, X Y% are going to be autonomous. It's also selling to the end customers. Selling cars in the future, having, you know, some type of autonomy included even if it's only available on a regional basis is going to be necessary for you to drive revenue just from your base product. Do you agree or not?
Yes, because I'll cite the data to Cox Automotive, Motive, and Experian. Here you are seeing a trend in the United States. I've not looked at international data. So I want to cite this. This is based on US data. You are seeing consumers, US households move from three vehicles, two vehicles to one vehicle. You're seeing them downsize as a way to cut expenses. It is a very real growing trend. Now, let's move on to the more meat of the Waymo post, which was not the supply stuff that excites the two of us. There was a lot of discussion about the sensors that are on these Waymos. What were your takeaways from that part of the Waymo post?
The sensors were interesting. Reading between the lines, Waymo talked a lot about the camera upgrading to a 17-megapixel camera, saying they could do more with less because the high fidelity on that. That had a very interesting undertone to the way it was written. Did you pick up on that?
No, I didn't. What did you say?
The way it was written basically, and this is my opinion, they were taking a shot at Tesla and saying that their cameras are superior to the Tesla cameras.
I mean, that's possible. I think there were more obvious shots at Tesla. I mean, there was whole sections on how they have camera cleaners, and you know, as our listeners know and those that are following it closely, some of the cars in Austin have been spotted with cleaners on cameras that don't exist on my Tesla Model Y, right? So, maybe that was a shot there. And obviously, there was once again a full section on LiDAR. And then there was a tweet later by someone from Tesla, I think on the engineering side, talking about like, you know, how this blog post or this post by Waymo was talking about how you want to get multiple data points or multiple sensor inputs to make your autonomy work, which is obviously the opposite of, you know, what the Tesla mantra has been, is where some of these multiple data points can confuse and slow down the reaction. So yeah, I think I didn't pick up on the megapixel camera one, but I did pick up that kind of undertone on the camera cleaners and also the LiDAR section about it.
I'll summarize it this way. The blog post was very crafty in terms of the language that we used and it wasn't just sensors. There was also language around compute where this is, I'm going to quote this here. Waymo's custom silicon chips rather than relying on multiple hardware components. That language was in there. That's interesting. Not one mention of Nvidia custom silicon chips. Interesting.
I mean, look, this is their thing, custom, right? Custom LiDAR as well. Obviously, they've claimed in this very blog post that they have, you know, even though their LiDAR is custom, that they've benefited from the drop in industry prices. I think some of the criticism has been like, oh, LiDAR is so cheap, you should have just bought a third party. They're claiming that they've benefited from those, you know, the reduced cost on the LiDAR. And then Nvidia stuff is like, obviously, you know, the counterpoint here is from Uber saying that the Nvidia system is going to effectively democratize and commoditize autonomy. So I think it was interesting for them to push their own custom silicon chips. Look, Google gets a lot of props given how they've turned around Gemini relative to the early threats of ChatGPT. So I think you got to cut them some slack and some benefit here about what they can do on the custom silicon side of things.
Well, that's an incredible custom silicon shop. I play with the TPUs, Walt, for training purposes or data. The TPUs are phenomenal. The engineering that went into the TPUs is great. The other thing, I'm going to put on my Walt hat here. When you're building custom silicon, the costs go down. You're not relying on the, if you want to call it, the Nvidia tax, their costs go down. So, everybody looks at, oh, the cost for the compute, the cost for this. I'm going to go on here and make a prediction. I think Waymo's cost to build and operate the Waymo vehicle today are a lot less than the market thinks.
I mean, sure, but then you're just ignoring the billions that they've spent over the past decade to get to this point to custom build everything. But sure, maybe it's cheaper to go look with all this tech, Grayson. I want to understand they can do the custom chips, they can clean the cameras, they can get the 50 million megapixel whatever, but they can't get a door that closes. They're now paying DoorDash to close the doors for them in the market. What's going on with that?
As much as I don't want to admit it, I don't like it. That's why they went to the Zeekr design so the doors can remotely close the way that, and you and I when we were in Austin, we tested this. If you walk away, the vehicle can't go through the door open. So they had to solve the door problem. So over time, I'm not going to give a prediction of when because I don't have any insight or any idea, but at some point those Jaguar vehicles are all going to be discontinued and so they're going to go to models where the doors can close.
Well, I would hope if you're spending $2.5 billion with Hyundai that Hyundai can figure out how to put some type of motor mechanism in there to close the door and perhaps a much larger reservoir for your camera cleaning fluids as well that is going to be needed. Some of those custom autonomy changes that are going to need to happen for some of these cars.
But that's the fun thing about this podcast. You and I talk about all the nuances or the, you're going to call the nooks and crannies that go into operating this from cameras to cleaning to fleet management. We go through it all and I think we do a pretty good job.
Now, with all this Waymo news, you know, we talked about Uber's earnings last week, miss and realities, but they were out there again. There was some podcast and stuff and really talking up the partners. So, I think it's good to kind of like go through, you know, some of the partners that they're talking about saying like, okay, Waymo's out there and they're great, but we've got, you know, WeRide and Pony and saying that, you know, like where they're all there. And just for context again, like today, Waymo has 2,500 plus cars, right? We think it's about a thousand square miles of service area. Pony, which was I think brought up, has a thousand cars. So, you know, 1,000 to 2,500. I think from based on what I was able to see online in China, you're talking about 1,600 square miles. Now, Waymo already gets complaints primarily pushed by Uber themselves about the wait times for a Waymo versus, you know, a human driver. So, how is it working with Pony with a thousand cars and 1,600 square miles versus Waymo's 2,500 cars with a thousand square miles? Like is that, do we have more qualitative feedback on how that service is going? And is that, I think Pony's talked about going to 3,000 cars at year end and I think by my estimates that Waymo is going to be more than twice that in this market.
I'll use the term that this audience is familiar with. It's behind the great firewall. What's truly happening behind the great firewall? We don't know. We did get some insight this week. There was an article published in Insider. Brian Walker Smith, who's a professor at University of South Carolina, went to the mainland to ride in vehicles. Well, the article was sparse on details. A couple interesting things. You had to be, and this is quoted in the article here, you have to be a Chinese resident to download the app to get in there. There's a lot of restrictions of who can go in those vehicles. Not everybody can go in those vehicles. Another interesting thing from the article is that you're limited to the geofence and they're smaller than the Western media thinks they are. They only operate at certain times of the day. The government could say, which I believe this one, okay, you're shut down. Too much traffic. You can't operate now. So, when you get down to the nitty-gritty or I'll use the term the ground truth, I don't think it's as impressive as the market believes it is to be because when you're getting this ground truth intelligence, it's a little different than what the market's interpreting.
Well, I don't know if it's the market. I think what we're specifically talking about is what Uber's pushing in their defensiveness about the Waymo progress, about where Pony is. And I think they've listed May Mobility. We're familiar with where May Mobility's coverage is in the markets. We've obviously been in those cars. WeRide, another Chinese maker, a thousand cars. AV ride,
which I think is 80 cars is what they disclosed on their earnings report again versus the 2,500 going to 6,000 going to tens of thousands a year, which is Waymo. The other thing that I think to me is curious is, you know, you're out there like basically knocking down Waymo to a certain extent and pumping up Chinese car makers, and you're a US company. And like, you know, at some point there is some sensitivity about, you know, the race to AI, which is like the race to 5G. And if you're a US company, I mean, they constricted Nvidia's ability to at least to a certain extent no longer to sell chips. Like you have a US company basically, you know, helping and promoting and partnering with these Chinese car manufacturers while at the same time like showing them in comparison to a great US company in Waymo and Google.
I would summarize what you said this way: it's an unforced error, and frankly, it's unfortunate where it goes. I don't know, but I do think that you will see geopolitical risks emerge both in the UK, the EU, in the US around this technology of Uber accelerating the adoption of Chinese autonomy. But then Waymo's not in the clear either. They have the Seeker. So it's, I mean, both parties have their own political risks based on strategy decisions they've made.
So let's transition to again the third leg of the stool that I've talked about: the servicing of the cars. Again, in Uber's defense, in many of these interviews, it made it sound like their, you know, the part to scaling autonomy would be like who's going to, you know, deal with all the software to figure out where the cars should go or how many they should have or clean them. And it almost made it sound like Uber owns the cleaning function. You know, as our listeners well know, these are all third parties, right? And there's incremental expense when you're going through an AOMA or Move or whoever it is as a third party. That's very different than Lyft. And Lyft actually highlighted this on their earnings call, saying that FlexDrive by having it in-house and focusing on this will provide 20% of incremental savings. Now, we're not saying like if it costs a dollar per mile or whatever it is for a human, 80 cents for autonomy, it doesn't take it down to 60 cents, right? It's 20% of the 20%. Like so it is small, but in a business where you know you're talking about bookings or IBITA being a single-digit margin on a bookings number, those little things matter. And I think this investment that Lyft is making in FlexDrive is very different than what Uber is doing in terms of third parties. And I tend to agree that if executed properly, it can be a strategic advantage and one that delivers greater margins. It's just funny or interesting that it almost seems like Uber's trying to leave people with the impression that they're actually doing this. They're not. Like these are third-party partners. And in some cases, I think from your video last week in Miami, one of their partners is servicing the Waymos in Miami. And that's not going to be an Uber partner for Waymo in Miami.
