Uber is building an autonomous vehicle empire, and here’s every company it’s using to do it

Uber's autonomous vehicle strategy isn't a strategy. It's a hedge fund with a ride-hailing app attached. The company that once poached Carnegie Mellon's robotics lab and bought a self-driving truck startup for $680 million now owns pieces of more than 30 AV companies across four continents. It writes checks. It signs MOUs. It presses go on pilot programs from Munich to Dallas. What it doesn't do anymore is build.

The fatal Tempe crash in 2018 didn't just kill a pedestrian. It killed Uber's ambition to own the stack. Khosrowshahi's 2020 "reset" sold ATG to Aurora, Jump to Lime, Elevate to Joby. The press releases called it a divestiture. The cap tables told a different story. Uber kept equity in all three. It kept a seat at every table it supposedly left. That wasn't an exit. That was a converted position.

Aurora is the clearest proof. Uber holds 325.97 million Class A shares — 19.7% economic interest, 6.9% voting power — through a holding company named Neben Holdings. The name means "next to" in German. Fitting. Uber sits next to Aurora, not inside it. In June 2024 Uber Freight announced a multi-year collaboration. By May 2025 Aurora's trucks were running Dallas-to-Houston roundtrips on the Uber Freight platform. The investment pays dividends in data and route density, not just equity appreciation.

But Aurora is one bet among dozens. Waymo. Cruise. Motional. Serve. Nuro. Kodiak. The list reads like a cap table for the entire AV sector. Some are robotaxi plays. Some are long-haul trucking. Some are sidewalk delivery bots. Uber partners with competitors simultaneously. It integrates Waymo in Phoenix and Cruise in San Francisco while Aurora hauls freight in Texas. The OEM-agnostic framing with Autobrains in Munich — "vehicles will be equipped with Israel-based Autobrains' autonomy stack" — reveals the actual product: Uber's dispatch layer. The car doesn't matter. The brain doesn't matter. The network matters.

This is the platform play executed at planetary scale. Uber doesn't need to solve perception, prediction, or planning. It needs to solve allocation. Which vehicle serves which request at which price in which regulatory regime. The 30-plus partnerships create a menu. Regulatory approval in Germany? Pull the Autobrains card. Freight lane in Texas? Pull the Aurora card. Robotaxi in Arizona? Pull the Waymo card. The blast radius covers every geometry of autonomy.

Critics call this fragmentation. They're wrong. Fragmentation implies disorder. This is deliberate optionality. Every partnership generates proprietary data — pickup patterns, dwell times, refusal rates, weather performance — that feeds Uber's routing engine. The AV partners get volume. Uber gets visibility. The asymmetry compounds. Waymo knows Waymo's fleet. Uber knows everyone's fleet.

The financial engineering is equally deliberate. Equity stakes align incentives without requiring board seats. Commercial agreements lock in volume without requiring capex. Wenn Uber wants to acquire later, it already owns the due diligence. Wenn a partner fails, the portfolio absorbs the loss. The structure resembles a venture fund that also happens to operate the world's largest mobility marketplace. That marketplace is the moat. No AV startup can replicate 150 million monthly active users. No OEM can replicate the two-sided network. Uber rents the moat to the technologists. They pay in integration and data.

Regulators haven't caught up. They evaluate each partnership in isolation — safety case, operational design domain, insurance structure. They don't evaluate the aggregate: a single dispatch layer controlling heterogeneous fleets across jurisdictions with different liability regimes. When an Aurora truck hits a bridge in Texas and a Waymo robotaxi hits a cyclist in Phoenix, the legal entity behind the wheel differs. The entity behind the dispatch is identical. That gap will eventually force a reckoning.

Drivers are the silent counterparty. Every autonomous mile displaces a human mile. Uber's take rate on AV trips will exceed its take rate on human trips — no driver split, no incentive bonuses, no surge sharing. The economics are brutal and beautiful. The 30 partnerships accelerate the crossover. Uber doesn't need to win the autonomy race. It just needs to survive it. The portfolio ensures survival. The platform ensures capture.

Khosrowshahi gets credit for the pivot. He turned a moonshot into a marketplace. The press calls it pragmatism. It's predation. Uber feasts on the R&D of 30 companies while insulating itself from their burn rates. The fatal crash forced the sale. The sale created the portfolio. The portfolio built the empire. History rhymes. Uber's autonomous vehicle empire runs on other people's code, other people's capital, other people's risk. Uber provides the demand. That was always the only asset that mattered.