Waymo's getting ready to launch commercial service at SFO for the public. But down in Santa Monica, they've got an incident they're dealing with. And as some of the critics say, LiDAR didn't bail them out. Meanwhile, Tesla is no longer sexy as they shut down the Model S and the Model X to make room for Optimus the bot. And Uber's inching closer to possibly rebooting ATG, funding Waabi to go and build robotaxis in addition to trucks. There's a lot to unpack this week, and also Mercedes is there. So, where do you want to start?
Well, I think we should start with Waymo, the airports. We've been waiting for this for a while. It's not totally new, right? They had the kiss and fly lot that I love to talk to. Now, we're getting, I guess, a little closer to the rental car lot. I don't know where that is relative to the garage that I am willing to walk to to get Ubers, which I think is a great facility there. But you and Rich Greenfield, my partner over at Lightshed, would prefer what Uber gets if you're taking a black, which is curbside. So, we're getting closer. But do you know where the rental car lot is relative to where you get a typical Uber in that garage?
I have to give Rich the honorary clappy hat. The clappy hat's not coming out today, but I'll give Rich an honorary clappy hat for that because he's right. We do want curbside service. Yes. To go to the rental car, which is commonly referred to as the RAC, you have to take a train. That's right, Walt. 10 minutes. Not only do you have to walk, but you have to get on a train, and it's a pain in the you know what. So, you got to take a train.
At JFK, it's kind of a little screwed up right now. And one of the terminals that I came into, I had to take a train. It was good because I got past some really bad traffic. But I tend to agree that once you're asking people to get on any type of train or tram or something like that versus what I think is a pleasant walk to where the Uber garage is, that's kind of a bridge too far. Give us a little inside baseball if you can. You've dealt in these situations before. Like what is it? How do they get from kiss and fly to the rental car lot, and what's it going to take to get to where the Uber Lyft garage is, and then ultimately authorization for curbside?
It's going to take a lot of political maneuvering. It's going to take the mayor of San Francisco taking bold action to say this is what we need because it's good for the tourism economy of San Francisco. If we look back at policy battles, the taxi union when they were strong, they banned Uber and Lyft from certain parts of the airport, and over time that's come. It's going to take a profound lobbying effort on Waymo's part to be able to overcome that. And then the question is, when Waymo starts operating the Uber and Lyft area, if that's what they do, does that cause more tension between Uber, Lyft and Waymo? Because a lot of those rides will probably go to Waymo. So overall, Walt, long-winded way of saying it's going to take an increased lobbying effort on Waymo's part to make it a reality.
Well, I'm actually impressed that in this short period of time, they got from kiss and fly to rental. So, congrats on that. Hopefully, it's a similar timeline to getting to the lot. And I think, you know, especially in California where the people of Northern California and Southern California get to experience Waymo on a daily basis. Hopefully, there'll be more pressure on their politicians or bureaucrats, whoever is making these decisions, to get there. And you're right, you know, the downside for Uber is these are very valuable trips. And I think the potential economic impact to Uber starts to rise when people start to think about using Waymo for more than just getting around San Francisco, but more of their regional needs. Those are higher value and greater market share opportunities for Waymo in that market. And you know what else happened that wasn't discussed a lot this week? Waymo expanded the OD in the peninsula to go to downtown San Jose. So now you can go from the San Francisco airport to downtown San Jose. So if you and I go out on an Autonomy Markets mission, or you go out on a business trip, you can take Waymo the entire time. That completely changes your journey and gives you a really great experience.
Yeah. I mean, that's awesome. And I think in a short period of time, Waymo has really transformed themselves in California from being just kind of in San Fran to more regional. You know, we're going to be out there a couple times in the first half of this year. So, I'm looking forward to being able to use these Waymos in that market and that expansion. And I think maybe we should think about setting up a tracker where we just look at square mileage overall where Waymo is in the United States as kind of a reflection of where their footprint is relative to any of their peers in the marketplace. You know, you'll see like May Mobility or whoever launching in Atlanta. We'll show what that looks like in stark contrast to where Waymo is today in terms of that square mileage. And by the way, as they expand to London, we're still going to use square mileage.
Yes, we're not going on the metric system there.
Listen, Walt, Hooked on Phonics obviously didn't work for me, so I'm sorry about that one. But you know what would be better? We take the population that lives in those areas and overlay on that map, and then we'll get a really big understanding of what percentage of the US population is being served today.
Yeah, for sure. That'll be interesting. Now, you talked about, or we both talked about, you know, getting public support for greater pressure in the airports. The people of Santa Monica don't seem like they support Waymo too much. We've already had complaints about their cars beeping because they didn't have anywhere to park, or maybe their back office was not suitable. We have a new incident this week as far as I can tell. There was a school situation. The Waymo was driving approximately 17 miles per hour. That's what was reported. And a child ran out from behind a car. The Waymo slowed to six. I actually, in my car this week, drove at 17 miles an hour. It seemed like more than an appropriate speed, you know, in a school type area. So, it doesn't seem like that's a speed that would be too fast for a school zone. Someone can feel free to correct me and say, 'Oh, the appropriate number is 15.' But honestly, the difference between 15 and 17, not much of a difference. And six, like six literally is a number where, again, look at your car if you have a digital readout. You get to six, I'm not even sure most people are getting slower than that before they can, you know, move out of a stop sign. We've talked about this thing. So, I think at the end of the day, my guess is this Waymo performed better than a human. And fortunately, this child was not, you know, seriously injured, which I can't imagine they would be at 6 miles an hour on contact.
And what happened was that this was in the NHTSA report. I just want to highlight this is in the NHTSA report. A child darted out from behind a double parked SUV. A human wouldn't see it. And if you want to look at the National Safety Council statistic database, the percentage of individuals that are distracted on their phone are high. So there is a higher percentage that that individual, if it was a human, would have been going over 17 miles an hour. To put it in perspective, the school district where I live, the school zone is 20 miles an hour. And I can't tell you they have a cop out there with a radar gun every day. Boom, boom, boom, boom, people ignore it. So there was a high likelihood in my opinion, and there's a massive amount of data to back it up, that a human would have caused a tragedy potentially or a worse situation than the Waymo did.
So it's obvious that the media likes to, you know, write stories about this because it's a robot, whatever, AI. I mean, there was also Zoox had a minor incident in, I think it was San Fran as well, in the Mission. So in California, not in LA, in the Mission, where someone's hand was struck in their like 1977 Cabriolet or something like that. Again, you know, big headlines in TechCrunch, but like minor dust-up. Here's the silver lining to our listeners who I believe are probably mostly pro-autonomy and seeing the benefits. This has not changed anything in terms of Waymo's service or other companies' service. Right? So you've had these incidents. Thank God they weren't serious. I think, you know, these are not, you know, this hasn't changed the trajectory. This is always the kind of the freak-out that people have, like does an incident change the trajectory or the willingness of these companies to move forward. Nor has there been, at least not that I'm aware of in Santa Monica, some outcry. I mean, I'm sure there's other, you know, interested parties that will drum up these incidents to try and achieve their goals to slow autonomy for their own benefit. But I think the positive side of this is, you know, the consumers have been able to digest this news, you know, in a normal fashion, which is, you know, kind of rare these days to be honest. It tells you that the technology is maturing and it's becoming part of the local community. There's, I can't tell you how many former NTSB individuals and board members that I've spoken to and said we just become numb to death on the roads. And if it was involving a human, we wouldn't be talking about it right now. But this is a positive sign that there hasn't been a major outcry. And it also tells me that the public acceptance of this technology is only growing.
Now, the flip side of this is you can't help but start to question because of the debates that exist with LiDAR and frankly some of the videos that Waymo themselves, Dmitri himself, has presented in terms of why LiDAR is so much better, with an exact example of a human, I think it was a kid actually, from behind a bus that jumped out. Now in this case they were describing the LiDAR I guess having seen below the bus and didn't have a direct line of sight, but certainly there has been this belief and certainly tests that have attacked Tesla, some of them, you know, questionable in how they were run, where they have like a mannequin that they're throwing out in front of the FSD whatever version that they happen to be using then. But the whole narrative was specifically that LiDAR with this specific instance would be better. And yet, if you look at the literal quote from Waymo from that release, it says Waymo said its vehicle, quote, 'immediately detected the individual as soon as they began to emerge from behind the stopped vehicle.' So, that's telling you if they're emerging from the vehicle, then you have sight, right? It is the camera that is seeing, you know, that child see it. So like, so to me, like it is, you know, if it doesn't, you know, kind of underscore like, yeah, sure, LiDAR for long distances, like if I'm traveling on a highway doing 70 miles an hour, is LiDAR going to see something 3 miles or 2 miles down the road? Sure. But do I need to see something 3 miles or 2 miles down the road? So like, in the inner debate that I continue to have where I agree that, you know, LiDAR provides greater information in some scenarios, but does it restrict the ability to create a driving experience that is materially better than a human? Like how is this one not in the column of like, shouldn't LiDAR have done better in this situation?
Well, it goes back to the Elon debate where he says the sensors create noise. I'm not saying that's what it is, but LiDAR is not this end-all be-all miracle drug or miracle sensor that certain individuals make it out to be. Is it a component that does a job? Yes. Is it a component that you need to drive a car? No. N-O spells no. And to give you an example on that, do you know where you can't deploy a LiDAR? I want to have, we'll play a little game here. Where, when you're deploying an autonomous vehicle, can you not use LiDAR because it will harm your safety? Where do you think that is?
You cannot put it in military applications because it gives off pulsing lights so the enemy knows where it is.
Well, thankfully, well, some people might debate this. We don't have enemies on our streets. I'm not willing to say no, you don't need LiDAR. I think it's not binary, Grayson. Like there is, you know, there is nuance. Like there are certain situations where maybe I'm willing to consider that LiDAR, and I think it's a reasonable consideration, is going to do better. The question is like, does that mean that camera is not enough to save tens of thousands of lives? I don't know. Like it seems like, you know, we're going down a path where that's a possibility, and I'm still leaving that option open, but I'm not saying it's like, no, you definitively don't need it. There's going to be, I think, situations where you do on those edge cases. You know, from my standpoint, I have an open mind. I'm not forced into one side or another. I own a Tesla. You own a Tesla and I drive it every day. And it's validating to me that it can be done. And you're right, it's specialty applications. Of course, you probably need LiDAR, but in the day-to-day what we experience, no, it's been proven that you do not fundamentally need LiDAR.
So, I want to throw this thought piece out there to our listeners, which I got, by the way, awesome positive feedback this week. Thank you so much for all the messages that you sent from last week's episode. Appreciate that. But I want to throw this out to our self-appointed experts, which is, is it possible that, you know, when you think about the added data from the additional sensors that are on a Waymo, including that LiDAR sensor, that the additional processing time that it had to take for taking all these sensors in had an incremental, whatever it is, half a millisecond, quarter millisecond of processing time that maybe the car would have stopped, you know, that quarter second, that quarter millisecond faster, that would have taken six down to two in terms of the miles per hour on impact. Now who knows, compute's powerful. I don't think, I don't know, I don't think Waymo is using Nvidia, so they're using their own kind of purpose-built, like is theirs better than Nvidia or faster than Nvidia? I don't know. But I just know when I have in our LLMs that I use, when I go to thinking mode and it's trying to get more inputs, the answers come back slower. Like so there seems to be some basic logic in that maybe there is a slight fraction performance differential that might exist when you have more sensors. I don't know, again, not making a statement. We have a lot of self-appointed experts. Let me know your thoughts on that as a concept. And do you have any thoughts, Grayson?
It is very, very possible. So when you're doing deep research in your LLM, you're triggering a larger percentage of the database, mostly running on an HDD, not an SSD. So there is time there. So there is a lag on that standpoint. The way that the Waymo system is architected, there could be a lag of a millisecond, half a millisecond. We don't know because Waymo has never made the specs public, but based on the architecture of these models and how they work, you are very right. They're possibly, I'll use the word possibly, could be. Talking about the LLMs, you're getting access to data centers, right? Massive compute. Remember, these decisions are being done based on the compute that's on the vehicle. They're not relying on the amount of compute that exists elsewhere. So, you have some limitation on the compute that's in the car. And again, you're giving it all of this extra data. I don't know. It just stands to reason that additional data would take more time to process. Maybe the answer is like, no, dude, like that incremental processing time is like one, I don't know, what is the smallest measure of time, a picosecond or whatever. And that's a completely irrelevant statement, but it seems to me that like latency and, you know, the tie-in between compute and the ability to handle different amounts of data has some impact.
100%. The question is what impact, and that's something that maybe Waymo should get more transparent about.
Let's go to Tesla earnings. It kind of validated what I've said on this podcast before in terms of like, does anyone really care about the number of cars they're making? It's all about autonomy. And I guess to a lesser extent or maybe to a greater extent, robots. And in fact, Tesla is getting rid of models S, X, and Y to focus on autonomy and robots, including an increase in capex which includes some more compute as well as some of the factories to build out this stuff. So I mean, it's a complete pivot to be like, hey, this is not about us being an EV company anymore. It's fully, you know, embracing autonomy as the future. Not that that's like a massive revelation, but that's the facts. And we got a new market timeline where you've got a handful of markets, Texas, Arizona, Florida, Nevada, all expected to be delivered in five months. What does delivered mean, Grayson? Because, you know, that could mean anything. Let's just put it that way. Why don't you say what you think this means in terms of these markets being launched in the first half of 2026?
What it means is that Elon and Tesla, they're not getting their sexy back on. They're moving away from the Justin Timberlake days and they're moving to the robotics stage. That's what it says to me at a high level. But the common denominator in the markets that Tesla has announced that they are going to launch for the first half of 2026 are the regulatory environment. Each one of these states, Arizona, Florida, and Texas, has a regulatory environment where Tesla has the ability and they could have the authority, depending on permitting, which I do not see for the record as any potential hiccups, to deploy driver-out unsupervised robotaxis in Cybercabs, and Cybercabs will be pending NHTSA regulations and NHTSA approval, but they have a clear path to do this. And I'm not going to speculate on which markets will go unsupervised in that timeline, but I'm just saying the facts from a regulatory standpoint. Tesla's cleared for launch.
Let me focus on that last thing. You're not going to say which is unsupervised. So what is, when they say first half, when they show this status in first half of '26, what does that mean? That they'll just have a car with perhaps a safety attendant in those markets, like two or three cars with a safety attendant in these markets in the first half of '26. Do you think that's what this is communicating?
I don't know if it's communicating one car, two cars, three cars. What I think, in my opinion, it's communicating Tesla's getting ready to expand the robotaxi service. And it also says to me Tesla is emerging as the biggest competitor to Waymo. Let's look at the Miami market for example. I put on my Inspector Clouseau hat and I did some work. I was able to pull building records from Miami-Dade County and was able to uncover the location of Waymo's Miami depot, uncover the current status. I will tell the audience it is under construction. And here's the interesting thing, Walt. When you and I were in Austin, we visited the Waymo depot and that was the size of a Costco or maybe a football stadium. That thing was huge. The one in Miami, it only has a permit for 69 spots. And what we also know from the building permit, the energy is not there yet. So, is Tesla saying there's a strategic opportunity to go into Miami and gain market share? I don't know. But they're going into markets where Waymo is active or has been planning to go. So, it says to me they're getting ready to compete.
69 spots are going to need obviously more than that if they really want to service Miami. They talked a big game about as they're launching, they're claiming they're going to have all of the service centers there. I'm not sure, we didn't see it in Austin. They also claim that they've got 500 cars on the market. Obviously, this includes California where they still have a, you know, a safety driver sitting behind the wheel there. I'm going to say I think the first half of '26 is basically what it looked like in December in Austin, which, you know, a handful of cars, someone, you know, as a safety attendant, you know, in the passenger seat if that, you know, for the number of markets. Obviously, they don't have a supply issue, but like that there's a number of people in operations. That's a lot of markets if you're going to have to have, you know, drivers to ramp up in the first half of '26. I think that's, to me, that feels like an aggressive milestone for them given some of the recent performance. But we'll see. We'll see. And it's, and you know, again, more things to hold them accountable for. We're not going to jam them up, I think, too badly on a three-week deadline miss here, but you know, now you've really put the pressure on in terms of the number of markets that they hope to launch on. A bigger issue for me also is just taking Austin to the next level as well. And maybe we'll see that in the coming weeks, getting more fully unsupervised cars on that road so that when we head down there, you know, we'll have no problem getting an unsupervised ride in a Tesla robotaxi.
Next week, we could get some clarity and insight as a representative from Tesla is testifying in front of the Senate about autonomous vehicles. So we could perhaps get some clarity there for their plans. That is perhaps. We do not know. We haven't seen the testimony, but there is a public hearing on that. So perhaps we could. When you look at the amount of vehicles in the market, are you thinking the 30 to 40, 20 to 30? Is this kind of reminiscent of when you and I first went to Austin?
Well, I don't currently, I mean, let's do a shout-out to David Moss. This guy we hadn't heard about until he drove coast to coast in his Tesla. Guy's a tire salesman. And the amount of information that he was generating while trying to get a driver-out, attendant-out ride in Austin was phenomenal in terms of also the number of cars, like the counting that he was looking at license plates. So there's a lot more cars there than I thought. Obviously, he had to go through 58 rides where there was someone sitting in the attendant seat before he did get that fully driverless, attendant-out ride. For the record, he turned around, looked through the window. There was no chase car. That always was to me kind of a silly supposition that someone was, you know, remotely driving it from behind the car. I can't imagine just from just thinking about that technically how that would make any sense, but whatever. Everyone has a right to their opinion. In this case, David Moss was there. I think he's a pretty credible source from what I can tell. No chase car and was able to get an attendant-out ride and gave us a better sense of, you know, the amount of cars that they have in the market.
Mr. Moss did a phenomenal job of putting on his journalistic hat. We've got great insights into that. The best insights that we've seen out of Austin. So Mr. Moss, thank you very much. The other thing that Mr. Moss talked about that you and I spoke about and did a podcast on was he said the vehicles blended in and the motoring public had no idea there was no driver in there. That was a really interesting takeaway that you and I discussed previously.
That was exactly our takeaway when we did our rides down in Austin. And also, you know, I remember reading someone's, you know, post saying like, 'Oh, you know, I was there. I didn't see any Teslas.' Like, no, because they don't have sensors all over them. Of course, you didn't see any robotaxis there. They blend in. In addition to that though, again picking up on what we were talking about last week with my drives to Philly and the sensor cleaning issue and the same thing that Alex Roy I think talked about in terms of his coast-to-coast ride. They've now added cleaning mechanisms on the side and the back sensor, which is new, right? That definitely doesn't exist on my car. So the question is, are these new cars? Is this going to be installed on all new Model Ys coming off the line? Was this an aftermarket addition? Like, what did you make of that? And by the way, most importantly, you know, can I go back to Tesla at some point and have my car updated so that I can get those little jets for my camera sensors as well?
Going back to Mr. Moss again, he posted close-up photos of the new cleaning things. It looked to me, and I'm not an engineer, but it looked to me, especially with the rear one, that it was not factored, that it was modified to perhaps during these tests in Austin, Tesla learned something and that perhaps in, I don't know the timeline, so just hear me out. Maybe three months, all new Model Ys start coming with it. It looked to me that it was not built on day one. It looks like it was updated after, but again, I have no factual stuff to prove that. I'm just basing that off of a photograph.
And the other issue with this though, Grayson, is like it's not just replacing a camera. You're going to need to pipe some plastic tubing to get your cleaner fluid to those locations. So if existing Ys that have already been sold with hardware need to get upfit, you know, with these new camera sensors, that is not, I can't imagine something easy to do. Again, maybe some of our listeners who are experienced in tripping out their cars will have some better sense than I would. And look, if it costs, let's say, you know, just to ballpark it, let's say if I took this to a detailer and said, you know, put this stuff on my car, let's say it cost $500, $1,000 with a $100, $200 a month payment for robotaxi or whatever it is. The payback on that's obviously going to be pretty quick if that's what it takes. But certainly an interesting development that David was able to uncover.
Yeah, I'll just say it this way, Mr. Moss. Thanks. It was awesome. You gave us great insight. And I'll say this for Walt and I, we see you in Austin, sir. Lunch is on us when we see you down there.
He doesn't live in Austin. He was just visiting it like we visited. I was actually, Grayson, going to warn him because we're familiar that we know there's someone that we've seen flying to Austin and filing complaints, following the robotaxis around and filing. I was going to warn him to like look for the guy in the bushes with the camera. Maybe they wanted to jump out, but I decided that maybe I was not going to do that. But that was, I love to refer back to your Inspector Clouseau of discovering who's actually, where did you...
Find that again? News, the police reports in Austin where they list who's making these complaints.
See, I got a — I've got a magnifying glass to find things. Now, we know who you are. We know who you are.
Just a good warning for people to think about the stats when they see these complaints. You know, where they're originating from and maybe what's driving that.
I didn't have to use my inspector glass to find it because it was in the public Austin police records. And furthermore, those police records linked to a YouTube video. I'll leave it at that. So, it's out there.
Beyond all that, perhaps the biggest news came after earnings when headlines started to hit about, you know, SpaceX merging with xAI, also thinking about merging with Tesla. And like, look, my takeaway from this is, as I've said much more on the Light Shed podcast, I've talked about it a little bit here. I was just interviewed on Bloomberg talking about Apple and my concerns about Google and their monopolistic power that they have within the AI realm. This is the new paradigm, right? You have to be these mega companies. So for Tesla to mash it all together and look on paper, you're talking about what, three, four billion dollar potential enterprise value for this company with a ton of free cash flow associated with that. A lot of synergies we don't necessarily get into here, but like I am all for — and by the way, from a Light Shed perspective, given all the work that we do on the space side of things, on connectivity as well as AI, and with now autonomy, would love to see this mega merger happen. What — how about you?
I'm all for it. I think it's wonderful. Now I want to throw this back to you and ask a hypothetical question from history. Is this the new — if the merger does go through — is this the new GE for the AI years?
