ICYMI we’re about two weeks away from the Ride AI summit in Hollywood, California on April 2nd, a premium one-day event spotlighting autonomous vehicles and AI mobility (check out the stellar schedule). As a reminder, ticket prices are increasing by 10% every week from now until the event, so if you plan on joining us, now is the time to book.

In the latest episode of the Ride AI podcast, our hosts Edward Niedermeyer and Timothy B. Lee previewed the aspects of the event they are most excited about.

‍

Today’s newsletter is supported by… Gridwise Analytics

Autonomous Vehicles Are Transforming Rideshare—Here’s What You Need to Know

Waymo, Zoox, Tesla, and other robotaxi firms are reshaping ridesharing, but challenges remain. To drive mass adoption, AV fleets must reduce per ride pricing below $1 per mile while maintaining profitability—requiring strong fleet efficiency and optimized pricing strategies. In a new whitepaper from Gridwise and S&P Mobility, explore the hurdles AV firms face and how data-driven strategies can help overcome them.

• Download the Report

Now, Here’s What’s Happening.

Waymo has launched a new service area in California. This time, it’s Waymo’s own birthplace: Silicon Valley. The new 24/7 zone includes Mountain View, Palo Alto, Los Altos, and parts of Sunnyvale. At first, Waymo’s robotaxis will be available only to a select group of Waymo One customers in eligible zip codes within the service area, although the company plans to gradually open up access to more riders. Also, the CA DMV has given Waymo approval to operate fully autonomously in expanded South Bay areas, including “almost all of San Jose.” While the expanded area won’t be immediately available for riders, based on previous launch schedules, it should be within a few months.

In addition, Waymo and the San Francisco Airport Authority have reportedly struck a deal to finally allow Waymo to map the airport road environment for “future autonomous use.” Notably, the permit expressly restricts Waymo from engaging in any commercial activity at the airport, and compels Waymo to share “some” data with the city government. The nature and extent of the data shared is not currently known.

The permit went into effect on March 14th, and will remain valid through April 13th, at which point the airport may choose to extend it another 30 days.

If insider knowledge (i.e. indications from employees and executives) are to be believed, my personal prediction is that Waymo will finally launch highway service by the end of the year in order to connect Silicon Valley to San Francisco and SFO. Waymos are now a routine sight on the freeways around San Francisco, and there has even been first-look coverage from news media with reporters being able to actually take a Waymo on the highway. With the time to general availability reducing every time Waymo launches a new service area, it really does seem like highways will happen pretty soon.

Some bad news however, as Waymo received almost 600 parking tickets last year in SF… way-mo than I would have thought. The parking violations ranged from blocking traffic to street cleaning. Now, I will say, the city does love to hand out parking tickets. I personally have received more here than any other place I’ve lived, mostly due to the incredibly confusing street parking signs, although I will admit I do sometimes just simply forget to move the car.

But this begs the question: Is it time for Waymo to train its own street parking sign reading models? Is this part of the long tail of issues we must solve? There might be way-mo merit to that idea than previously thought! Alright I’ll stop now.

Are Zoox vehicles due for a recall? A previously undisclosed safety report from NHTSA alleges the Amazon-backed company failed to comply with eight safety rules… although the rules were mostly intended for human-driven vehicles. The list of rules Zoox broke includes requirements for side mirrors and brake pedals, which Zoox claims its vehicles don’t need. Apparently, NHTSA officials showed up at a Zoox facility in Las Vegas in 2024 with measuring equipment in an attempt to call the company’s bluff after it stuck stickers on all its cars that say the vehicles meet NHTSA’s rules.

Nothing seems to have come of the report, however. Neither the California nor Nevada DMV has taken any action, despite having had access to the report since December. It remains to be seen what will happen when NHTSA concludes its review.

In some Tesla news, Baidu has now dispatched mapping engineers to Tesla’s offices in Beijing in an apparent bid to integrate Baidu’s mapping data more strongly into Tesla’s assisted driving features. This comes after customers in China racked up hundreds in fines due to their usage of Tesla’s assisted driving features.

Fines aren’t the only bad news for Tesla. Pony.ai CEO James Peng publicly expressed doubts about Tesla’s upcoming robotaxi deployment this year. And he’s not the only one. Waymo’s former CEO John Krafcik also implied that Tesla might “fake” a robotaxi service. This caused some drama in the robotaxi sphere, as news of competing autonomy CEOs throwing shade at one another tends to do. To be clear, Peng stopped short of outright saying that Tesla’s robotaxi service won’t happen at all.

Krafcik, on the other hand, was much more direct: “There are many ways to fake a robotaxi service.” Here’s the full quote.

If a company were serious about building a safe robotaxi business, the robotaxi wouldn’t look anything like this prototype. A serious robotaxi would demonstrate the primacy of safety; the manufacturer would place sensors in optimal positions—on the roof, as well as on the sides and corners of the vehicle. These sensors would also have cleaning and drying functions—windshield wipers, compressed air nozzles, and so on. A serious robotaxi also wouldn’t have a low-slung coupe body design. This design makes it difficult for people to easily get in and out; not everyone will be able to use these robotaxi vehicles comfortably.

A new video from famous YouTuber Mark Rober shows a Tesla on Autopilot crashing through a Wile E. Coyote-style wall painted to look like a road. The video pits a car with LiDAR against a Tesla with Autopilot through a series of tests. The results of the tests are… not entirely favorable for Tesla. See for yourself.

The video has caused significant controversy in the self-driving space, with some accusing Rober of faking the results on behalf of his sponsor, Luminar. Others accused Rober of conducting the tests in an unrealistic scenario, vis-à-vis the fake wall. A significant portion of the complaints hinged on a pivotal moment in Rober’s video that showed Autopilot seemingly being disengaged right before the Tesla made impact.

In response to the accusations, Rober released an unedited clip of the testing. This version showed that Rober had accelerated the Tesla to 40mph and enabled Autopilot. However, moments before the car runs into the wall, Autopilot seemingly disengaged by itself. This behavior has been documented previously by NHTSA in a report, but NHTSA was unable to find malicious intent. Watch this space, as they say…

Rivian finally launches hands-off highway driving in the US and Canada. In the launch video, RJ Scaringe, Rivan’s CEO, revealed that data from the fleet is continuously collected and fed into larger offboard models, which are then distilled into more efficient onboard models through over-the-air updates. And “Enhanced Highway Assist,” as the system is known, is just the beginning for Rivian’s new intelligence capabilities. Scaringe mentioned in the launch video that automatic lane changing is coming “soon,” and that access to surface roads and turn-by-turn navigation are also on the roadmap. He touched on higher levels of autonomy as well, noting that the current perception architecture is not so different from that of an L4 perception stack, suggesting that hands-off eyes-off features would be coming within the next year. This would likely put Rivian’s current system at L3. Concerns around practical L3 autonomy aside, I’m looking forward to seeing what Rivian can achieve with their perception stack.

In Other News...

• Nissan is now testing driverless vehicles on public streets in Yokohama.
• What it looks like inside a Waymo depot in a multi-level parking garage
• Waymo’s sensor fusion avoids pedestrians climbing over wall on dark, divided highway

One Last Thing…

Travis Kalanick thinks Uber screwed up by abandoning its in-house self-driving project.

Look, [new management] killed the autonomous car project we had going on. At the time, we were really only behind Waymo but probably catching up, and we were going to pass them in short order … I wasn’t running the company when that happened, but you know, you could say, ‘Wish we had an autonomous ride-sharing product right now. That would be great.

I do wonder what it would be like if Uber had continued its autonomous car project. However, hindsight is 20/20. It’s easy to forget that at the time, there were big problems with Uber’s AV efforts. The company was burning a ton of cash on a program that had seemingly mid-tier results at best—and they may not have been as close to catching up to Waymo as Kalanick likes to think. Safety standards were apparently lax, resulting in an Uber test AV, driven by a safety driver who was distracted on their phone, crashing into and killing a pedestrian. NTSB reports showed that, at the time of the accident, Uber’s test vehicles were not able to track pedestrians outside of crosswalks. Not to mention the issues of IP theft and “ask forgiveness, not permission” culture.

I do think Waymo needs strong rivals. While I’m a fan of the service, competition would be beneficial for the entire industry and for consumers alike. Maybe Zoox will emerge as a serious challenger? Or dare I say… Tesla?

Alright, that’s it from me… until next week!

‍