Uber's plan to own autonomy without building it

UBER TECHNOLOGIES has spent the past twelve months signing robotaxi partnerships with a frequency that would embarrass a diplomat at a summit buffet. On March 19th the ride-hailing behemoth announced its latest: a deal worth up to $1.25 billion with Rivian to build as many as 50,000 autonomous R2 SUVs, starting with a $300 million investment and a planned rollout in San Francisco and Miami by 2028. It was, by the company's own recent standards, a fairly quiet week — Uber had already unveiled partnerships with Amazon's Zoox, an expanded tie-up with Nvidia, and a three-way collaboration with Wayve and Nissan in the preceding days alone.

The cumulative picture is staggering. Uber now counts more than 25 autonomous vehicle partnerships spanning robotaxis, freight, and delivery across at least four continents. With Lucid and Nuro, it plans to deploy over 20,000 premium robotaxis built on the Gravity SUV platform, with the first rolling out in San Francisco later this year. Through Nvidia, it aims for 100,000 Level 4 vehicles across 28 cities by 2028, with Stellantis supplying the first 5,000 units. Wayve and Nissan are preparing a Tokyo pilot for late 2026. WeRide is already operating on the Uber app in Dubai and Abu Dhabi. Chinese AV firms Momenta and Baidu have signed on for Europe and the Middle East. By the end of 2026, Uber targets driverless rides in 15 cities. CEO Dara Khosrowshahi, who generated $52 billion in revenue and nearly $10 billion in free cash flow last year, has described the company's ambition plainly: to become the largest facilitator of autonomous trips in the world.

But the most interesting thing about Uber's strategy is not whom it has signed — it is whom it cannot control. The company's most important AV partner remains Alphabet's Waymo, which operates on the Uber app in Austin, Atlanta, and Phoenix. Waymo now completes over 400,000 weekly rides across ten U.S. cities and is expanding rapidly — to Dallas, Washington D.C., London, and beyond. Increasingly, it is doing so without Uber. In Dallas, Waymo chose Avis as its fleet partner. In London, where Uber plans its own Wayve-powered service, Waymo will launch independently with fleet operator Moove. The two companies are drifting from symbiosis toward direct competition, and London may be the first city where riders can choose between them.

This matters because it exposes the central tension of Uber's platform bet. The company sold its own self-driving unit to Aurora in 2020 — a move now widely regarded as shrewd, given the billions competitors have burned on proprietary autonomy stacks. In its place, Uber has built what amounts to an operating system for robotaxis: a suite called Uber Autonomous Solutions that handles routing, charging, in-car interfaces, data collection, and demand matching. The pitch to AV developers is compelling — Khosrowshahi claims vehicles on the Uber network achieve 30 percent higher utilization than those operating on standalone apps. With 202 million monthly active users and 40 million daily trips, the demand side is formidable.

Yet the bear case is equally coherent. Every robotaxi partnership means shared revenue and lower margins than a ride fulfilled by an independent human driver who absorbs all vehicle costs. BofA Securities has noted that the financial impact of the Zoox deal on profitability through 2027 is minimal — which is another way of saying the upside, for now, is also minimal. And the partners themselves carry significant execution risk. Rivian has not yet begun producing the R2; its Georgia factory remains under construction; and its autonomy platform, while promising, has not achieved the hands-off, eyes-off driving required for a true robotaxi. Similar caveats apply across the portfolio: Wayve's technology is nascent, Nuro has never operated at commercial scale, and Momenta has yet to deploy outside China.

The analogy Wall Street likes best is that Uber is building a toll road for the autonomous age — collecting a fee each time a robotaxi uses its network without bearing the cost of building or maintaining the fleet. Deutsche Bank recently declared Uber the long-term winner, drawing parallels to Microsoft's capture of PC-era value through software rather than hardware. The stock, trading around $78 at roughly 15 times trailing earnings, reflects cautious optimism. But toll roads work only when traffic is both high-volume and captive. If Waymo, Tesla, and eventually others prove they can attract riders on their own apps — and pricing data from San Francisco suggests the gap between Waymo and Uber fares is narrowing — then Uber's network advantage may be less durable than its partnership count implies.

San Francisco is shaping up to be the proving ground. By late 2026, it could become the only city where riders choose among four competing robotaxi services: Waymo, Tesla (operating under a charter permit, with prices significantly below competitors), Zoox, and Uber's own Lucid-Nuro fleet. The result will offer the first real market data on what consumers actually value — a guaranteed autonomous ride from a dedicated app, or the convenience of a platform that aggregates everything. Uber is betting the answer is the latter. It had better be right: with 25 partnerships and counting, it has staked its autonomous future not on any single technology, but on the proposition that in a world full of robotaxis, somebody still needs to answer the phone. ■

For more, join 75,000 subscribers getting tech's favorite brief here