Uber Sold Its Self-Driving Unit in 2020. Now It Plans to Commit More Than $10 Billion to Robotaxis

On August 5, an analyst on Uber’s second-quarter earnings call asked about the company’s relationship with Waymo in Austin and Atlanta. Chief Executive Dara Khosrowshahi said Waymo remained an important partner and that Uber expected the relationship to continue into next year. He also made clear that Uber did not want to depend on a single autonomous-vehicle provider.

That is the contradiction at the center of Uber’s robotaxi strategy. Uber does not build the self-driving technology it is now committing billions to put on its network. It needs the companies that do, and it cannot assume they will always need Uber. In its second-quarter filing, Uber identified Waymo, Tesla and Zoox among companies it expects to use autonomous technology to compete with it.

The question is whether a company that sold its own self-driving unit can become a leading commercialization platform for robotaxis when its suppliers can also become rivals.

Why did Uber sell its self-driving unit in 2020?

On December 7, 2020, Aurora announced that it was acquiring Uber’s Advanced Technologies Group (ATG), the unit that had spent years developing Uber’s autonomous driver. The transaction combined a transfer of ATG with a $400 million investment by Uber in Aurora. Uber CEO Dara Khosrowshahi joined Aurora’s board, and Uber expected to hold a minority stake of about 26 percent in the combined company, according to contemporary reporting and regulatory disclosures.

The transaction was not a $4 billion cash sale. Contemporary reporting described Aurora as being valued at about $10 billion after the deal, while ATG had been valued at $7.25 billion following a 2019 investment from Toyota, Denso and SoftBank. Those are valuation figures, not cash proceeds paid to Uber for ATG.

The reasons for the decision were layered. Fortune reported that ATG was burning roughly $500 million a year and that some investors had urged Uber to reconsider its strategy. Uber’s filings showed that its advanced-technology programs had recorded substantial losses during 2020. The pandemic had also weakened the ride-hailing business, while the road to a commercially viable self-driving service remained long. The unit was shadowed by a 2018 Arizona crash in which one of its test vehicles killed a pedestrian.

What Uber gave up was direct ownership of the technology team and its autonomous-driver program. What it kept was an equity stake, a board connection and a stated path to bringing Aurora’s technology to the Uber network. The 2020 transaction was a retreat from building the driver itself, not from the idea that autonomous vehicles could eventually be dispatched through Uber.

What does Uber’s $10 billion commitment actually mean?

In its prepared remarks for the second quarter of 2026, Uber said it expects to commit more than $10 billion of capital over the coming years across equity investments, infrastructure and vehicle offtake commitments. That is a forward-looking expectation, not money already spent and not a single cheque.

Chief Financial Officer Balaji Krishnamurthy described the commitment as a multi-year total that largely reflects partnerships already announced. The structure combines equity stakes in autonomous-driving partners—often tied to milestones—with spending on fleet operations, facilities and vehicle purchase commitments. Uber has also said it is exploring financing structures that could bring outside capital into fleet deployment and help preserve balance-sheet flexibility.

Uber said its partners had committed approximately 120,000 vehicles to its network over the coming years. That number refers to partner commitments across the ecosystem. It does not mean Uber has bought all those vehicles or that they are already on the road.

The strategic argument is that a good autonomous driver is only one part of a commercially viable robotaxi service. The business also needs a steady flow of riders, dispatch systems, vehicle integration, fleet operations, charging, financing, insurance and relationships with regulators and local communities. Uber says its ambition is to become the world’s leading commercialization platform for autonomous mobility. That remains an ambition, not an established position.

Uber has also built AV Labs, which operates hundreds of sensor-equipped vehicles to collect driving data for partners. The company says this is not an attempt to rebuild its own autonomous driver. It is intended to help partners identify and understand the driving scenarios that matter to ride-hailing and to move development and deployment forward more quickly.

How Uber’s robotaxi strategy works in practice

Two partnerships show the model. In March 2026, Uber and Rivian announced that Uber would invest up to $1.25 billion in the automaker through 2031, subject to autonomous-driving milestones. An initial $300 million investment was committed following signing, subject to regulatory approval. Uber or its fleet partners are expected to purchase 10,000 fully autonomous R2 robotaxis, with an option to negotiate the purchase of up to 40,000 more beginning in 2030. Initial deployments are planned for San Francisco and Miami in 2028, with expansion to 25 cities through 2031 if the milestones are achieved. The vehicles are intended to be available exclusively through Uber.

A second program brings together Lucid, which builds the vehicles, and Nuro, which supplies the autonomous-driving system. The companies have described a minimum commitment of 35,000 vehicles worldwide. Lucid’s regulatory filing explains that the total comes from a 2025 agreement covering 20,000 Gravity vehicles and an April 2026 agreement covering 25,000 midsize vehicles over six years after production begins. Contractual offset provisions bring the combined minimum commitment to at least 35,000 vehicles. Lucid targets late 2028 for the start of production of the midsize model.

The Bay Area launch is planned for 2026, with Houston expected to follow in mid-2027. Reuters, citing two people familiar with the matter, has reported that Uber’s commitment to Nuro is close to $500 million; that figure is based on reporting rather than a public contractual disclosure. Uber has secured a 50,000-square-foot Houston depot and charging facility to support charging, maintenance, repairs and cleaning. The companies say the vehicles will be owned and operated by Uber and its fleet partners and offered through the Uber network.

These are not purely asset-light arrangements. Uber is putting capital into partners and preparing to take responsibility for parts of the physical fleet operation. That could give it greater control over utilization and service quality, but it also creates execution risk if vehicle supply, cost or regulatory approval falls short.

In its second-quarter remarks, Uber said autonomous vehicles were live on its network in seven cities and that it was on track for as many as 15 by year-end. Since then, Uber and Wayve launched supervised autonomous rides in London on September 3, 2026. Uber’s city count was therefore a snapshot of the second quarter, not a complete count at the time of publication. The 15-city figure remains a company target, not a guarantee.

What happens when Waymo competes with Uber?

In September 2024, Waymo and Uber announced that Waymo’s service in Austin and Atlanta would be available only through the Uber app. Uber would manage and dispatch the fleet and handle cleaning, repairs and depot operations, while Waymo remained responsible for testing and operating the Waymo Driver.

That arrangement is expected to change. In July 2026, Uber told reporters that Waymo had notified it of plans to launch service through its own app in Austin and Atlanta in January 2028, alongside its existing deployment with Uber. Uber said the end of exclusivity would allow it to add other autonomous providers in those cities and that Waymo vehicles would remain on its network through at least May 2028, when the contract ends. These dates were reported from Uber’s account; Waymo did not immediately comment in the initial reports.

Uber has framed first-party and third-party distribution as models that can coexist, as they do in travel, food delivery and e-commerce. An autonomous vehicle can serve its maker’s own app while also accepting trips through Uber, which argues that its rider base can provide incremental demand and help improve vehicle utilization. Uber management has cited trips-per-day figures for vehicles launched on its platform, but these are company-reported measures rather than independent proof that the model will work across markets.

The strategic distinction is between owning the customer relationship and owning the technology. Uber controls its app, its rider relationship and dispatch on its network. In some markets it is also building depots, charging access and fleet-management capabilities. It does not control the autonomous-driving software, production schedules set by manufacturers, whether partners develop competing apps or the regulators that determine where driverless vehicles can operate.

A broad partner roster does not automatically create leverage. Uber has leverage only if no single supplier becomes indispensable and riders continue to use Uber when a partner offers a competing app. The company has said its category share in Los Angeles, San Francisco and Phoenix has improved year over year, but that is management’s own measure. Stronger evidence would include partners continuing to supply vehicles after their own apps launch, sustained ride volumes through Uber, and commercial terms that leave Uber with attractive economics.

How does the capital-allocation trade-off look?

Uber is making other large capital decisions at the same time. Its proposed takeover offer for Delivery Hero helps show the scale of the commitments competing for attention. On July 16, 2026, Uber announced an offer of €41.50 per share, implying an equity value of $14.8 billion for the company before adjustments for Uber’s existing stake. Uber later published the offer document, and the acceptance period was scheduled to run through November 5, 2026. The transaction remained subject to the offer conditions and regulatory approvals; Uber’s filings said closing was expected in the second half of 2027. The company said it planned to fund the offer with existing cash and new debt, supported by a €14.2 billion bridge credit agreement.

The robotaxi commitment has a different shape. It is spread over years and linked to investments, facilities and vehicle commitments, with some spending conditional on partners hitting milestones. Delivery Hero is a proposed acquisition supported by debt financing. Neither transaction funds nor proves the other, but both compete for capital and management attention.

At the end of June, Uber reported $5.4 billion in unrestricted cash, cash equivalents and short-term investments. That is a different measure from its $12.5 billion in investments, most of which were publicly listed, and different again from the $10.1 billion of trailing-twelve-month free cash flow reported for the period. Free cash flow measures cash generated over time; it is not cash sitting in a bank account available for immediate spending. In the second quarter, Uber also increased its economic interest in Delivery Hero through shares and derivatives. These categories should not be added together as if they were one pool of cash.

The company’s ability to finance robotaxi deployment depends on more than the headline free-cash-flow number. It must maintain the core ride-hailing business, manage insurance and regulatory exposure, fund other strategic priorities and avoid paying too much for vehicle capacity before demand is proven. Milestone-based deals can limit some of that risk, but they do not eliminate it.

What has Uber proved, and what remains unproven?

Uber’s core business gives it room to make long-term bets. In the second quarter of 2026, gross bookings reached $58.0 billion and reported trips reached approximately 3.9 billion. Autonomous rides remained a small part of that scale. On the earnings call, Khosrowshahi described autonomous vehicles as completing hundreds of thousands of trips a week against roughly 300 million trips across Uber. That comparison suggests a small share of overall volume, but the autonomous-trip estimate was a management estimate, not a separately reported audited metric.

Much of what would validate the strategy lies ahead. Rivian’s planned robotaxis are due to begin deploying in 2028. Lucid’s midsize robotaxi is not expected to enter production until late 2028, and the Gravity is a premium vehicle. That raises an important question: can a relatively expensive vehicle earn acceptable returns on ordinary ride-hailing fares once insurance, cleaning, charging, maintenance, depreciation and fleet financing are included? Uber has not yet published enough deployment-level economics to answer that question.

Vehicle supply is another uncertainty. Partnerships can announce large commitments, but actual deployment depends on manufacturing, autonomous-system performance, safety validation and local approval. The agreements’ milestone provisions matter because they tie some capital to progress. They also mean the announced fleet sizes should be read as plans and contractual minimums, not as vehicles already available to riders.

Regulation and safety bring further costs. Uber’s filings warn that crashes involving partner vehicles could expose the company to liability and that new rules could limit the services it can offer. A service may be technically capable of operating but still be constrained by city rules, community concerns or the need to show that it can coexist safely with pedestrians, cyclists, emergency vehicles and conventional traffic. Uber management has acknowledged that deployment may need to proceed at a pace that earns regulatory confidence.

The company has said it will provide more visibility into profit and cash-flow effects as deployments scale. Until those figures emerge, investors cannot judge the economics from fleet announcements alone. Trips per vehicle, vehicle uptime, cost per mile, contribution after fleet expenses, customer retention and the split of demand between Uber and partners’ own apps will be more revealing than the total number of vehicles committed.

What can other executives learn from Uber’s bet?

The 2020 and 2026 decisions rest on the same thesis: a company can capture value without owning the core technology if it controls a scarce layer that technology providers cannot easily replace. In 2020, Uber stepped away from developing its own driver while retaining a financial stake and a route to future commercial use. In 2026, it is committing capital to several providers while trying to own more of the demand, dispatch and operating infrastructure around them.

The balance sheet has changed, too. Uber now reports more than $10 billion in trailing-twelve-month free cash flow, giving it more room to make staged commitments. The lesson is not simply to outsource difficult technology or to buy scale through partnerships. It is to define which layer of the value chain the company intends to control, then structure contracts so that capital follows evidence of delivery rather than promises alone.

Austin and Atlanta will be a revealing test. If Waymo adds its own app in January 2028 and riders continue to book meaningful volumes through Uber until and after the existing contract ends, Uber’s distribution thesis gains support. If demand migrates to Waymo’s own channel, Uber may have helped build a market for a supplier that no longer needs it.

Uber has not yet proved that it will become the leading commercialization platform for autonomous mobility. It has shown that it can assemble a broad set of partnerships, commit capital in stages and use an existing marketplace to accelerate deployment. Whether that becomes a durable advantage will depend on what happens after the announcements: the cars must arrive, regulators must allow them to operate, unit economics must work and riders must keep choosing Uber.