Hey, Jaan here.
I’m back with our biweekly robotaxi newsletter to keep you up to date with everything going on in this very fast-moving industry, which, granted, is still in its early days of scaling up.
I might have to make this newsletter a weekly instead if this pace continues — there’s just too much happening to fit in this one email for you.
A quick reminder: you are getting this Robotaxi Report by default as complementary to our EV newsletters, but if this really isn’t your jam, you can easily opt out of this Robotaxi Report by clicking here (opt-out link).
In today’s Robotaxi newsletter, we’ll take a look at:
Uber to invest in Verne, which switches from pure-play to partnerships
Zoox and the future of Robotaxi Providers: Facilitators vs Providers and how they are actually competitors;
Waymo hits 500k paid rides/week & shows new safety data
Uber orders 20,000 Rivian R2 Robotaxis
Bolt & NVIDIA partner for data & learning
Robotaxi Resource of the week
… let’s dig right in. Enjoy!
IN PARTNERSHIP WITH 🤝 VOLTERA
This newsletter couldn’t happen without our long-time partner Voltera. Here’s a quick intro in Voltera’s own words:
One of the Voltera sites
“Voltera develops and operates charging infrastructure engineered for autonomous mobility.
Our sites deliver reliable high-power charging in key geofenced zones, helping robotaxi fleets maximize uptime and scale from pilot programs to full commercial deployment.”
Important: this newsletter does not go through any editorial overview of Voltera’s team, nor can they influence any of the reporting — the words in this newsletter are to be blamed on me alone.
VERNE: UNEXPECTED PARTNERSHIP IN EUROPE
Verne, the Rimac-founded robotaxi startup that was kind of supposed to be the European pure-play vertically integrated robotaxi answer to Tesla and its Cybercab, announced a deal with Pony AI and Uber. Uber also intends to invest. (link)
The details of the partnership:
The vehicle used is an electric Arcfox Alpha T5 Robotaxi equipped with Pony AI robotaxi suite (the one pictured above, produced by BAIC).
Note that this is not Verne’s own robotaxi.It will be deployed in both Uber and Verne’s own networks.
Uber “intends” to invest in Verne as part of the strategic partnership.
In Zagreb, Croatia, deployment is “already underway, including public-road validation”, before expanding to additional cities in Europe and beyond.
Verne will act as fleet owner and service operator.
Verne will lead the process of ensuring market readiness and obtaining European regulatory approval for these launches.
Partners plan to scale to a fleet of thousands of robotaxis over the next few years.
Interestingly, zooming in, I see the side AV test text still says “Project 3 Mobility”, which is a former name of Verne. The official entity still, perhaps.
There was no mention of its own Verne Robotaxi (below), but I’ve been told this bespoke vehicle is still the end goal, and details on that are coming later.
The status of this vehicle program is currently unknown — the most recent update we got was directly from Mate Rimac about 3 months ago, when he showed us the 61 prototypes on a lot that Verne had produced to date.
A few of my notes and/or opinions on this:
This Uber’s intended investment is an important bit here: a ‘necessary evil’ (not that evil) to scale up the company has likely come at a right time, and seems to be either a prequisite to a Series B (the $100M Series A was raised in February 2024), or somewhat of a replacement depending on how well the talks have been going.
Verne will also benefit from the easier scaling-up of all the operational sides where it wants to be vertically integrated itself eventually:
the operations & service (gets utilization for its fleet management hubs),
the customer-facing hailing app (gets to ramp up service for customers quicker, not dependent only on its own service slow ramp-up),
The utilization for its own robotaxi fleet — as soon as the first Verne’s own robotaxis hit the streets, they get rides through the Uber app, while its own customer-facing app might not have the reach
a technology switch? I have a gut feeling that we are seeing Verne drop Mobileye as the AV partner for its robotaxi, and opt for the “NVIDIA x everyone” or partner with Pony AI for the sensor suite instead. Remains to be seen, no official info yet.
Verne had also received EU funding of €179.5M in a project “covering the entire concept and production cycle of an autonomous electric vehicle, from “formulated technological concept” to “system completed and qualified”.
I found that the end date of this EU project is March 31st, 2026 — is this why we saw the announcement and test vehicles go live now?
There are a lot of interested parties. If this hadn’t been a Uber + Pony AI deal, we would have very likely seen Verne partner with Baidu + Lyft for this, dropping the Baidu RT6 robotaxis in instead. The two sides might have been at the table together.
The part of their press release saying “Verne will lead obtaining European regulatory approval for these launches” is not insignificant — it’s something that Pony AI really needs as it expands through Europe via such partnerships. This is the way into Europe for the Arcfox robotaxis too, because most of its other partnerships seem to be tied to Stellantis.
I’ll spare you from all the other angles my brain went to for now. If you’re interested in taking this discussion further, drop me a note.
If you want to share this Verne news as an article, I’ve put it out as an EVwire article here.
ZOOX & THE FUTURE OF ROBOTAXI PROVIDERS
Photo by Aicha Evans, CEO of Zoox
Zoox announced it is now testing its service in Austin, Texas.
It has been mapping the streets with the test fleet (Toyota Highlanders) since 2024, and now to start with the actual robotaxis, rides will be available to Zoox employees and their family and friends. The early outside testers in the Zoox Explorers program will get access later this year.
Aicha Evans, CEO of Zoox, shared on X: “Today, I took the inaugural ride around Austin and livestreamed it to our crew. This city is where my career as an engineer began, so riding here felt truly full circle.”
Zoox also recently announced it is expanding testing to Phoenix and Dallas, which means it is now testing or giving rides in 10 US cities — rides currently in two of them.
It has also now launched three Zoox Fusion Centers, which are their ‘mission control’ hubs for teleguidance, supporting riders, and doing remote assistance, similarly to Waymo. All robotaxi providers have hubs similar to this in one way or another, and Zoox just announced its third in Scottsdale, Arizona. The other two are in the Bay Area and Las Vegas, where it already operates a publicly available fleet.
Fun find — Bill Gurley thinks all robotaxi players will end up going for the same form-factor as Zoox, and Steve Jurvetson (he led the seed round in Zoox btw) wonders if the boxcar form can pass a highway speed crash testing or is low speed only, all in this thread.
Zoox in Uber wasn’t always meant to happen
What has been interesting to watch in light of the news we covered in our previous newsletter about Zoox partnering up with Uber to have the rides bookable on its platform, starting this summer in Las Vegas and next year in Los Angeles, Uber riders can get matched with a Zoox robotaxi.
Zoox's CEO has been very adamant before about not wanting to be on anyone else’s platform. Aicha Evans has said this on numerous occasions publicly for years, and as late as November last year:
“We don't want to be a feature in someone else's platform. We want to be the platform."
It seems to me that Uber The Facilitator* has a way of changing the minds of these initially pure-play vertically integrated platforms — as we now see with both Verne and Zoox folding and letting their robotaxis go on the grand platform.
*Sorry, got all Foundation vibe here for a moment. Some of you know will know what I mean.
Why this sudden shift is more important than you’d think
Reality is, there are both wins and losses to this move.
In the short term, Zoox will get a lot more utilization and doesn’t have to acquire its own customers for the robotaxi service, making it easier to scale to other cities.
On the other hand, and especially for the long term, Zoox doesn’t have to acquire its own customers and is now dependent on the Uber robotaxi platform for its demand.
It is not at all easy to decouple from that later, and is essentially sacrificing the long game for the short-term benefit. Why? Let me explain.
I’ve seen it all before. Some of you working in shared mobility for longer have as well.
Let me bring you along for a quick personal story, as I was there when it happened.
The year was 2013, and in a small country called Estonia, an app called Taxify was launched. Nowadays, you know this company as the ride-hailing giant in Europe and beyond called Bolt, in 850+ cities worldwide. Here’s how it unfolded:
The idea was simple: let anyone find a suitable taxi for them with nothing but one single app, instead of working the queues and looking for the best choice or personal preference.
The taxi companies were excited to join — I know, because I was working at one of the first (maybe even the first) companies to join, a fully electric taxi company.
Boosted visibility. Your own customers could use the app to still find you (most taxi companies did not have their own app back then), and you got a trickle of extra people booking just because you were on the app. What’s not to love? Taxi companies were directing people to use the app, so Taxify got bigger.
At one point, Taxify reached critical mass to start making its own decisions. Click to order, and we’ll send you a taxi. Based on proximity, price, or whatever metric other than just company choice.
Then, at one point, the taxi companies became unchooseable, anonymized; the rides were just given based on the platform’s own metrics for the customer (which makes sense from the platform view).
Oh, and the prices are now determined by the platform, not the taxi company.
For the taxi companies? Uncoupling meant pretty much death.
Because by then, they had pushed their own customers to get used to using someone else’s app… which now became their competitor.
On our own e-taxi company side, we luckily recognized this early & got out of the app quickly to rebuild our customer base, to live and fight another day.
But especially with the introduction of private drivers and not just taxi companies on the platform, it got worse for the taxi companies that stayed on the platform, a lot of them just withering away over time and now out of business.
There’s a whole other story that pushed the Bolt founders to pursue private drivers btw, they actually wanted to fight this as it is something that Uber was bringing to the market and was essentially illegal back then… story for another time.
To be clear — all this is not meant to put Bolt in a bad light here. I think this has been somewhat of a natural progression of the market and competition (taxi companies) not realizing who their actual competition was. There are no sour grapes on my end, and I am eagerly watching Bolt make its robotaxi moves in this industry.
HOW I THINK THE ROBOTAXI FACILITATORS VS PROVIDERS WILL PLAY OUT
Zoox CEO Aicha Evans ← → Uber CEO Dara Khosrowshahi
The potential parallel of the early ride-hailing app vs taxi days with these robotaxi companies today is hard to ignore. Especially as one party in this mix is the same one.
For clarity, I’ll call the ride-hailing apps “Facilitators” and the robotaxi companies “Providers”.
Here’s the step-by-step of what I think will happen.
Facilitators start partnering up with all Providers they can.
Uber as the Facilitator (and Lyft & Bolt, and other ride-hailing giants) will partner with all these robotaxi providers to bring them on their “demand platform”.
It is an initial win for the likes of Zoox, Verne, Pony AI and all other providers as they get the utilization, and often even investment.→ WE ARE HERE ←
Partnership fleets are increasing, but both are still winning.
More robotaxi players join these demand platforms and the fleets are finally scaling up, increasing the choice on apps between autonomous players. At first, the Providers are separated enough by regions (cities), and each player gets all or most of the demand.→ Everyone is winning ←
Then, partnerships start to overlap in a city.
The same robotaxi-hailing platforms will have multiple robotaxi players in the same city. Maybe there will be a way to toggle which robotaxi you get from the Uber app. Or maybe it’s already mixed from the start.
→ First sign of trouble for the robotaxi provider. ←Facilitator, the demand platform makes the ‘unifying’ move.
There are several robotaxi providers on the platform with large enough fleets in each city. If it didn’t do so before, the demand platform will not distinguish between the robotaxi players anymore; it will just match the rider with “autonomous vehicle” and that’s it.→ There is no edge for a single robotaxi provider anymore ←
It is fully at the whim of the robotaxi-hailing app and its algorithms. Competition increases, and importantly, the prices are now pushed by the platform.
Facilitator, the demand platform launches the optimized competing product — its own fleet.
It’s optional, but very likely. This is what the private drivers brought to the ride-hailing platforms before, and what killed most taxi companies. In our new case, it is the platform introducing its own product. Even if the Providers decouple from the platform, they have to build most of the customer base from scratch.→ The Providers who didn’t realize & adapt will go out of business ←
Now, you can see indications of this already. Baidu launched its own robotaxis from the start, as did Tesla. Uber has done large deals with Lucid/Nuro; Rivian; Stellantis. Bolt with Stellantis. And so on.
The Facilitator becomes the Provider.
What started out as a marketplace is actually going to create the core demand, service, and do as much of it itself as it wants. It dictates what the early Providers need or do not need to do to stay on the platform. It makes the kings (or really, becomes one), and decides the outcasts.
Result: true robotaxi providers like Zoox, Verne, Pony AI, WeRide, and all contestants in this robotaxi race are better off taking the harder route — building solo. This is why Waymo made the smart move and is on Uber app only in the first two cities, and none of the rest.
I saw someone using the word coopetition and I think it’s exactly right for this purposes.
The moment these Providers decide to give their customers away for the platform, they are signing away their true company potential and risk remaining a small cog in the big machine.
This is also why Dara, CEO of Uber, has been openly inviting Tesla robotaxis to its platform, and why Elon Musk will never accept the invite.
Now, there are a lot of reasons this will play out in this way, even if the players involved might all realize this today. One is that the robotaxi Provider thinks it can overcome and be the best/largest provider on the platform. After step 5, it’ll start declining, but until then, it is the king.
The other part is that current Facilitators and thus, future Providers are not saying any of this aloud currently, quite the opposite — because they would create their competition for the future Step No. 5 where they need to dominate.
They need to partner with all of these robotaxi providers to gather the service under one roof. The more the better, any exclusives regionally would hurt them in the long run.
So, if we look at this deck from Uber again from Q4 (with recent news of Rivian and Zoox not even in it), you can see how it might play out in the odd case that I am right about this.
Another fascinating layer to all this is the fleet management and operations — the green box, which, for example, Verne has now decided to join per the announcement above. We’ll tackle that topic soon.
As you can see from my rambles, I don’t outsource any of my thinking, nor writing, to AI yet. So I’m inviting you to discuss this topic and the potential future of the industry.
DISCUSS: I’ve just put this thought process on the 6 steps up as a post on LinkedIn and as a post on X, please leave a comment on what you think.
WAYMO HITS 500k RIDES PER WEEK
Waymo hit the 500k mark, now serving half a million paid rides per week. That’s 26 million customers served a year. It is also currently a bit over half of all robotaxi rides provided on the planet.
Let’s look at the other news around Waymo these past two weeks:
Waymo released new safety data
Looking into 170.7M miles driven through Dec 2025, the Waymo Driver was involved in 13x fewer serious injury or worse crashes than human drivers in the same cities, per its latest Safety Report: (link).
The safety data also always gives us a good ratio of the numbers the overall fleet does based on cities:
Phoenix has logged 68.61M miles
San Francisco Bay Area 53.52M miles
Los Angeles 37.86M miles
Austin 10.72M miles
The absolute key safety benefits of Waymo can be summed up with just this one overview:
And to see the absolute biggest effect of the crash type:
As we’ve said before, Waymo has massively removed vehicle-to-vehicle intersection crashes, at least in the types of roads and cities it operates on. The numbers here go 262 incidents as human benchmark vs 10 incidents for Waymo across the 170M mile dataset.
You can find the methodology and datasets for download on their Safety site linked above, and I’ll also provide an interesting viewpoint: not all human-driven incidents are reported to NHTSA, which is what the human side of this data is based on — on the robotaxi side, however, every single smallest incident is reported on for maximum transparency.
UBER INVESTS UP TO $1.25B & ORDERS UP TO 50,000 RIVIAN R2 ROBOTAXIS
Uber announced that it is investing up to $1.25 billion in Rivian Automotive to deploy up to 50,000 R2 robotaxis. (EVwire link)
The companies plan to deploy an initial 10,000 Rivian R2 robotaxis, with commercial operations expected to begin in San Francisco and Miami in 2028.
The program is then set to expand to as many as 25 cities across the U.S., Canada, and Europe by 2031.
Rivian's R2 is unveiled, but isn't in production yet (expected in June).
Uber, or its fleet partners, will purchase the vehicles, with these R2 robotaxis operating exclusively on the Uber platform.
Uber’s investment will be tied to Rivian achieving key autonomy benchmarks, with an initial $300 million tranche committed, subject to regulatory approval.
Rivian’s R2 platform is central to the partnership, thanks to its third-generation autonomy system. This includes a multi-modal sensor suite featuring 11 cameras, five radars, and one LiDAR, supported by dual in-house RAP1 chips delivering up to 1600 TOPS of compute.
Quite the fleet customer for the R2 for Rivian.
Europe: Bolt partners with NVIDIA for data & learning
Bolt and NVIDIA partner to combine Bolt’s extensive ride-hailing and carsharing fleet data with NVIDIA Omniverse libraries, NVIDIA Cosmos world foundation models, NVIDIA Alpamayo AV foundation models, and NVIDIA AI infrastructure to accelerate safe AV development for European roads. (link for more details)
The resulting AV platform will be deployed on the NVIDIA DRIVE Hyperion computer and sensor architecture to deliver a robotaxi service that meets Europe’s highest safety and performance standards.
ROBOTAXI RESOURCE OF THE WEEK
Here is our next visual robotaxi resource that I teased last week:
Originally built by Ethan McKanna, this tracker is built as a live tracker of the robotaxi fleet rollouts for Tesla, Waymo and Zoox. With Tesla specifically, he created a way to easily log all Tesla robotaxi vehicles based on the licence plates from the community, to get the most accurate potential view of the real fleet rollout. It’s not 100%, but it’s the closest we can possibly get.
For Tesla, this tool has logged 436 different Tesla robotaxi vehicles in SF Bay Area in use (with a safety driver there) and 3 Cybercabs in testing.
In Austin, it shows 42 rider vehicles (with a passenger side safety monitor), 52 more that haven’t been spotted for 30 days, and 8 that are confirmed fully driverless (no safety monitor). 26 different Cybercabs have been spotted in the test fleet.
It also shows 12 Zoox vehicles: 7 in the Bay Area and 5 in Las Vegas.
Now, Ethan has announced that he has accepted an internship position in Tesla’s robotaxi team for the summer, so he will be handing off the project to someone else for the time.
You can see by the viral response with 3.7k likes to my X post announcing his move just how many people really cared about this project Ethan has created.
Ethan wrote: “Turns out the team at Tesla was paying attention. This summer, I'll be joining the Vehicle Software team as an intern, where I'll have the chance to help build the Robotaxi experience.
It's surreal to go from obsessively tracking every detail of this industry from the outside to now getting the opportunity to directly shape it."
That’s two for two of the people creating robotaxi visual trackers being grabbed by robotaxi players, if you count Jackson from AVmap we shared in the last newsletter.
I guess if you’re looking for one of the robotaxi industry jobs… you know what to do ;)
…and that’s all I could fit in the email today!
This newsletter went out to exactly {{active_subscriber_count}} other EV geeks just like you and me.
What did you think of our first Robotaxi Report?
Thank you for the vote of confidence last week! Here are some of your comments from the previous report:
P said:
“Great newsletter, love the format polish. More like this please, Jaan :)”
Your wish is my command 🫡
Duco said:
“I guess you are aware of it, but Ark Invest has extensive intel on robotaxi strategy. Thanks, Jaan, great job.”
— Indeed, Duco, I think we’ll be sharing some of their reports along the way here, too.
A said:
“Good read. Really surprised at the level of covering has in the US and China. Waiting for some ac8here in Canada”
— The regional play is fascinating. Let’s see what happens in Canada, almost a free market up for grabs today!
Brian said:
“Good coverage of an important developing component of our transportation systems.”
— Thanks, it’s a very important component indeed.
Richard said:
“Excellent initiative Jaan. The key is to what extent the rise of autonomous taxis will affect the underlying demand for personal cars? As legacy automakers wind back their EV plans are they heading for the exit lane?”
— Good point, and remains to be seen. Logically, it should reduce the personal car demand… but it is also quite long until we can see this. Additionally, there will be personal autonomous cars (Tesla et al) that serve dual purpose.
One thing I am certain of, though: selling a non-autonomous ICE vehicle will be very difficult soon, and this will remain a very, very niche market only (some specific performance/nostalgia/design market). The legacy automakers winding back without a good plan will be in even bigger trouble very soon.
Rick said:
“I want the cyber cab, but until then, I am in a self-drive (supervised) Model 3”
— The personal ownership of Cybercab will be interesting to watch unfold, if and when it happens.
TJ said:
“With Tesla on the verge of expanding exponentially, and other companies throwing their hats into the ring, this is BIG. Great to cover it in detail!”
— Indeed, TJ, this is B I G
L said:
“Very interesting, you keep writing, I’ll keep reading. ;-). L”
— You keep reading, I’ll keep writing!
Thank you & see you soon!
— Jaan
