It took 15 years. Then it happened all at once.
The autonomous vehicle industry spent a decade in permanent pilot mode. In 2026, it broke loose. Here's what changed and what it means for the physical world.
We predicted Uber's play with SpotHero before the market saw it coming. Now we're looking at what Uber's autonomous vehicle strategy means for the next wave of fleet infrastructure, and making a prediction on what's to come.
Since the start of 2026, Uber has announced autonomous vehicle partnerships at a pace that borders on absurd. A $1 billion deal with Waabi to deploy 25,000 robotaxis. A joint venture with Nuro and Lucid to build 20,000 purpose-built robotaxis launching later this year. Amazon's Zoox joining the Uber app in Las Vegas this summer. Wayve, Nissan, and Uber teaming up for a Tokyo pilot. A commercial robotaxi launch with Avride in Dallas. A planned fleet of thousands of autonomous Volkswagen ID. Buzz vans starting in Los Angeles. Partnerships with WeRide operating live in Abu Dhabi, Dubai, and Riyadh. Deals with Baidu, Momenta, Pony.ai, and May Mobility across the U.S., Europe, and Asia.
At the time of writing this article, Uber formed more than 25 autonomous vehicle partnerships spanning robotaxis, freight, and delivery. The company expects to offer AV rides in 15 cities globally by year-end. It already facilitates millions of autonomous trips annually.
Meanwhile, Waymo went from six cities to ten in a matter of months. It's targeting one million rides per week by December. It raised $16 billion. It's heading to London and Tokyo.
Morgan Stanley called 2026 an "inflection year" for private AV companies, projecting 33 consumer launches and expansion into nine additional cities. The robotaxi market was valued at roughly $5 billion in 2025. Some forecasts have it north of $600 billion by 2033.
Something broke open. The question is why now.
The decade of perpetual testing
For years, autonomous vehicles lived in the gap between promise and product. Billions poured in. Timelines slipped. The technology was always "five years away."
The poster child for this pattern was Cruise. GM's robotaxi unit burned through more than $16 billion in total funding before GM pulled the plug in late 2024 following safety incidents and regulatory problems. Argo AI, backed by Ford and Volkswagen, shut down in 2022 after six years and billions spent. Apple quietly abandoned its decade-long car project. The industry was starting to think that full autonomy was a money pit.
Waymo kept going though. It started offering paid driverless rides in Phoenix suburbs as early as 2020. But even Waymo looked like a science experiment more than a business. Expensive vehicles. Tiny service areas. Heavy lidar rigs on every roof.
Then something shifted...
What changed: five forces converging at once
1. Waymo proved the commercial model works
This is the most important thing that happened. Waymo went from 200,000 paid rides per week in early 2025 to over 400,000 by year-end. It's now live in ten U.S. cities. Its safety data shows a tenfold reduction in serious injury crashes compared to human drivers across more than 127 million autonomous miles.
That data gave cities the confidence to say yes. Dallas, Houston, San Antonio, Orlando, Miami all opened up in rapid succession. Public opinion followed. A San Francisco survey found 67% of residents now support robotaxis, up from 44% in 2023. Net favorability swung from negative seven percent to positive 38.
Waymo proved that robotaxis are not a theoretical exercise. They're a product people use and like. That changed the conversation for every regulator, investor, and fleet operator in the country.
2. Capital flooded in at a scale nobody expected
Waymo raised $16 billion at a $126 billion valuation. Waabi raised $1 billion in a single quarter, oversubscribed by hundreds of millions. Uber committed over $100 million just for AV charging infrastructure. Avride secured $375 million. Nuro and Lucid pulled $300 million from Uber alone.
Given the rounds, the investors are now betting on who gets to market fastest, rather than writing speculative seed checks.
3. AI reasoning models cracked the long-tail problem
The hardest part of autonomous driving was never the straightforward stuff. It was the weird stuff. A mattress in the road. A ball bouncing into traffic. A construction worker waving you through a closed lane. The "long tail" of rare scenarios that rule-based systems couldn't handle.
Two breakthroughs changed this.
First, end-to-end AI models replaced the old approach of bolting together separate systems for perception, planning, and control. A single neural network now processes sensor inputs and outputs driving decisions, keeping context intact through the entire chain.
Second, reasoning models brought something closer to judgment. NVIDIA's Alpamayo family of open-source models introduced chain-of-thought reasoning to autonomous driving. The vehicle reasons through cause and effect instead of just seeing an obstacle. It can explain why it made a decision. This is the difference between a system that handles 99% of scenarios and one that handles 99.99%.
Waabi's CEO Raquel Urtasun called it "a massive leap in AI." The company built a single AI "brain" that drives both trucks and passenger vehicles, learning from both environments simultaneously. What earlier companies needed thousands of engineers and billions of dollars to attempt, AV 2.0 companies are doing with leaner teams and fundamentally better technology.
4. Federal legislation is finally moving
For a decade, the U.S. had no federal framework for autonomous vehicles. Companies navigated a patchwork of state-by-state rules. Some states welcomed testing. Others effectively banned it. The regulatory chaos made national scaling nearly impossible.
In January 2026, Representatives Latta and Dingell introduced the SELF DRIVE Act, the first federal statute dedicated to AV safety. It directs NHTSA to develop national safety standards for automated driving systems. It creates a centralized crash data repository. It preempts conflicting state regulations.
In the Senate, the Autonomous Vehicle Acceleration Act calls for updating Federal Motor Vehicle Safety Standards to accommodate Level 4 and Level 5 vehicles. The SELF DRIVE Act has already cleared committee.
The regulatory wind shifted because the political incentives shifted. Nearly 40,000 Americans die in traffic crashes every year. AV companies now have the safety data to make a credible case that their vehicles are meaningfully safer than human drivers. And the competition with China on autonomous technology gave both parties a reason to move.
5. Uber flipped the script on autonomous vehicles
AVs were always part of Uber's vision. Why pay for a human driver when a robot can do it cheaper? But, Uber abandoned its own AV development program in 2020 after a fatal crash and years of burning cash. Then, Uber waited. And now, Uber is doing something smarter.
Uber stopped trying to build the technology and started building the platform that AV companies need.
Think about what an AV company has to do to commercialize:
- Build the self-driving software. Integrate it with a vehicle.
- Get regulatory approval in every market.
- Set up fleet operations: charging, cleaning, maintenance, inspections.
- Build a consumer app.
- Generate demand.
- Handle customer support.
- Manage pricing and dispatch.
Most AV companies are good at one or two of those things. Uber already does it all (on one way or another) at global scale, with 202 million monthly active users, and 40 million trips per day.
So Uber became the demand layer. In February 2026, it formalized this with the launch of Uber Autonomous Solutions, a dedicated division selling software, services, and operational support to AV partners. The pitch is straightforward: you build the robot, we'll fill it with passengers.
The math supports the strategy. Uber CEO Dara Khosrowshahi told analysts that AVs on the Uber platform achieve 30% higher utilization than robotaxis hailed on standalone apps. That utilization gap is the whole ballgame. A robotaxi sitting idle is just a depreciating asset. A robotaxi running 30% more trips per day is a viable business.
The partnership roster tells the story. Waymo. Waabi. Nuro. Zoox. Avride. WeRide. Wayve. Baidu. Momenta. Pony.ai. May Mobility. Volkswagen with Mobileye. Uber isn't picking a winner. It's building the rails and letting every horse run.
The real estate angle: where the ground game gets literal
There seems to be a myth that robotaxis will never park. That they will always be driving. That's not true. That is not what we see in the real world at Mobility Places. That's not what Uber sees either.
Every robotaxi needs a place to park, charge, get cleaned, and get serviced. Every expansion city means new depot capacity. Every new AV partner launching on the Uber platform means another operator who needs real estate. Somebody has to find it, lease it, equip it, and run it.
Uber is trying to be that somebody. The company committed $100 million to build autonomous vehicle charging hubs across U.S. cities, starting in the San Francisco Bay Area, Los Angeles, and Dallas. It's acquiring real estate for depots. On its Q4 2025 earnings call, Khosrowshahi said explicitly: "We are setting up depots, acquiring real estate, making sure we have the charging infrastructure in place." Uber Autonomous Solutions, the division launched in February 2026, offers AV partners depot tooling, maintenance coordination, vehicle cleaning, sensor calibration, fleet insurance, and rider support. The stated goal is to let AV software companies focus on building the self-driving brain while Uber handles the day-to-day physical operations of running a fleet.
That's the ambition. The reality is messier. Not every partner wants or needs Uber to run their operations. Waymo builds and manages its own depots, runs its own app, and handles its own fleet ops. It partners with Uber in Austin and Atlanta for demand, but it also operates independently in eight other cities. Zoox will launch on the Uber app in Las Vegas this summer while continuing to run its own fleet and its own app in parallel. These are companies with billions in backing and full-stack operational capability. They don't need Uber to clean their cars.
Other partners absolutely do. Nuro builds self-driving software, not fleet infrastructure. Waabi is a 300-person AI company. Avride is a startup that went from sidewalk delivery bots to robotaxis in 18 months. For these companies, Uber's operational layer is the difference between launching in one city and launching in ten. The Uber Autonomous Solutions pitch is that a software team shouldn't have to become a real estate operator, a charging network builder, and an insurance broker just to get a robotaxi on the road.
So the landscape is a spectrum. On one end, vertically integrated operators like Waymo that control everything from the AI stack to the parking lot. On the other end, lean AV software companies that plug into Uber for everything except the driving brain. In between, every possible hybrid arrangement. But regardless of who manages the depot, somebody still has to source the physical space, get the power connected, and make it work.
That's the bottleneck. The technology is scaling faster than the infrastructure to support it. The AV industry is entering the same phase that the EV industry hit two years ago: the hardware works, but the physical network isn't ready. Charging. Parking. Staging. Maintenance bays. These aren't software problems. They're real estate and operations problems. And they need to be solved in 15 or more cities simultaneously, by a mix of operators who have wildly different levels of capability and experience in sourcing and managing physical space.
This is the gap where companies that understand both real estate and fleet operations become essential. Whether it's Uber building depots for Nuro, Waymo leasing its own facilities through Avis, or a third-party operator providing turnkey fleet staging for the next wave of AV 2.0 startups, the industry is going to need hundreds of new facilities optimized for autonomous fleet operations in the next three to five years. The landlords and operators who figure out this asset class early will own a structural advantage that compounds with every new AV deployment.
What happens next: a story about P2P marketplaces
2026 is the year autonomous vehicles cross from tech demo to daily reality in more than a dozen cities. The technology inflection already happened. The capital is deployed. The legislation is moving. The consumer acceptance curve is steepening.
The next body of work now shifts to physical operations. Vehicles. Depots. Charging. Maintenance. Permitting. The companies that built the AV software stack spent fifteen years and tens of billions getting here. The companies that build the physical infrastructure stack will determine whether any of it actually scales.
Prediction: Right now, Uber's platform model looks like the fastest path to commercialization for most AV companies. They plug into Uber, get access to 202 million users, skip the decade of demand-building. They get access to infrastructure. It makes sense today. But, our predictions is that, once mass adoption hits, AV companies will start to go direct much harder, and need their own infrastructure much more.
We've seen this movie before. In the early days of peer-to-peer lodging, small hosts depended entirely on Airbnb and VRBO for distribution. The platforms provided the demand, the trust layer, the payment processing. Hosts grew. Some became professional operators managing dozens or hundreds of units. And then they started to chafe. Platform fees ate their margins. Algorithm changes buried their listings. They had no direct relationship with their customers. The biggest operators eventually built their own direct booking channels, their own brands, their own tech stacks. They didn't abandon the platforms entirely, but they diversified away from total dependence. You can say the same about Turo hosts, and virtually any other marketplace.
The same dynamic will play out in autonomous mobility. Today, Avride and Waabi and Nuro need Uber. They don't have consumer apps with 200 million users. They don't have operations teams in 15 cities. Uber's platform is the difference between a pilot in one city and a business in ten. But as these companies scale, some of them will want to own the customer relationship. They'll want to control pricing. They'll want margins that don't include a platform cut. The ones that survive the next five years will have the fleet size, the safety record, and the brand recognition to go direct. Some already do. Waymo runs its own app. Zoox runs its own app. The Uber partnership is additive for them, not existential.
This matters because it means the AV infrastructure layer can't be Uber-dependent either. The depots, charging hubs, and staging facilities that get built in the next three years need to serve a fragmented market that will only get more fragmented as it matures. The fleet operator who launches on Uber in 2026 may be running a hybrid distribution model by 2028 and a direct operation by 2030. Their infrastructure needs don't change when their platform strategy does. They still need a place to park, charge, clean, and stage vehicles. They still need power. They still need a landlord who understands the use case.
The AV future everyone predicted is arriving. It just doesn't look like the futurists imagined. It's not one company with a vertically integrated robotaxi fleet in one city. It's a fragmented ecosystem of technology providers, vehicle manufacturers, platform operators, and fleet managers, all converging on the same set of cities at the same time, all needing the same thing when they get there. Whether they're on Uber's platform or building their own, whether they're AV 1.0 survivors or AV 2.0 upstarts, the common denominator is physical space.
Somebody has to find it, negotiate it, equip it, and manage it. That's the ground game. And it's just getting started.
About Mobility Places: Mobility Places sources and manages parking for fleets. We've spent years translating between two industries that rarely speak the same language: fleet operators who need physical space, and property owners who don't understand fleet operations. Our clients today are microtransit companies and mobile service fleets. The next generation of clients looks like AV depot operators, robotaxi fleet managers, and the third-party companies being stood up to run autonomous vehicle operations in new markets.
