Uber Returns to Self-Driving Race with Data-Focused AV Labs Division

Uber's Strategic Return to Autonomous Vehicles

Uber Technologies has announced the creation of AV Labs, a new division dedicated to gathering real-world driving data for its network of autonomous vehicle partners. The move marks the company's return to hands-on involvement in the self-driving space, years after selling its in-house robotaxi operation to Aurora Innovation in 2020.

A Data Play, Not a Technology Play

Rather than developing its own self-driving technology, Uber is positioning itself as a data infrastructure provider for an industry that increasingly relies on massive datasets to train AI systems. The company now has more than twenty autonomous vehicle partners, including Waymo, Waabi, Nuro, and Lucid Motors.

AV Labs will deploy sensor-equipped vehicles across Uber's six-hundred-city global footprint to capture edge-case driving scenariosÔÇöthe unpredictable situations that remain the primary bottleneck for autonomous vehicle development. Partners will receive processed data as a semantic understanding layer rather than raw sensor feeds, and Uber will run partner software in shadow mode to flag any divergence between autonomous decisions and human driver behaviour.

Scrappy Beginnings with Big Ambitions

The division is starting with a single Hyundai Ioniq 5 outfitted with lidars, radars, and cameras. Uber's VP of Engineering Danny Guo acknowledged the modest starting point, noting the company plans to grow AV Labs to several hundred employees within a year.

The announcement comes amid broader scrutiny of autonomous vehicle safety, with Waymo currently under investigation after its vehicles were caught illegally passing stopped school buses on multiple occasions. Such incidents highlight exactly the kind of edge cases Uber hopes to help address through its data collection efforts.

Part of a Larger Autonomous Strategy

AV Labs complements Uber's existing partnership with NVIDIA to build a robotaxi data factory, with plans to collect more than three million hours of driving data. The company aims to deploy one hundred thousand autonomous vehicles globally by 2027 and expects to offer robotaxi services in more than ten countries by late 2026.