Uber is looking at a familiar asset in a new way: its enormous network of human drivers. According to TechCrunch, Uber CTO Praveen Neppalli Naga described a plan at TechCrunch’s StrictlyVC event to use drivers as a kind of distributed sensor grid for autonomous vehicle companies. The idea builds on Uber’s AV Labs program, which the company announced earlier this year, and points to a practical role for ride-hailing networks even as robotaxi services expand.
The logic is straightforward. Self-driving developers need constantly refreshed information about lane changes, construction zones, unusual traffic patterns and local road conditions. Uber’s drivers already traverse those environments every day. If the company can package trip-level observations, vehicle telemetry or other road-intelligence signals in a privacy-conscious way, it could become a data supplier to companies trying to improve autonomous driving systems.
That would also give Uber a different kind of leverage in the AV ecosystem. Rather than betting only on owning or operating autonomous fleets, Uber can position itself as a demand channel, logistics layer and now potentially a data layer. For automakers and AV startups, access to broad, frequently updated real-world data could shorten the feedback loop between simulation, testing and deployment. It may also make partnerships with Uber more attractive in cities where autonomous coverage is still early and uneven.
Why it matters
Autonomous vehicles are not just a software problem; they are an operations and data-refresh problem. Roads change faster than HD maps, and edge cases are often local, temporary and difficult to reproduce. Uber’s proposal suggests the next phase of AV competition may depend as much on real-world sensing partnerships as on model quality. It also shows how platforms with dense human activity can remain valuable even while automation threatens parts of their core business.
Source: TechCrunch, published May 2, 2026, 1:36 a.m. CT. Header image: original SysBrix abstract illustration generated for this post.