Uber aims to transform its 2.5 million global drivers into a massive sensor network for autonomous vehicle (AV) companies. The plan, revealed by CTO Praveen Naga at TechCrunch’s StrictlyVC event, involves equipping human-driven cars with sensors to collect real-world data for AI training. This initiative, part of Uber’s AV Labs program launched in January, seeks to address the industry’s data bottleneck by leveraging Uber’s vast driver network.

Currently, AV Labs uses a small fleet of sensor-equipped vehicles, but Naga envisions scaling this to millions of cars. The company’s goal is to create an “AV cloud” — a centralized library of labeled sensor data accessible to partners. This would allow AV firms to train models on diverse scenarios without deploying their own fleets. Uber’s partnerships with 25 AV companies, including London-based Wayve, highlight its strategic pivot from direct self-driving development to a data-centric model.

Naga emphasized that data, not technology, is the key barrier to AV progress. By democratizing access to high-quality training data, Uber positions itself as a critical infrastructure provider. The company has already invested in AV startups, and its proprietary data could give it leverage over competitors reliant on Uber’s ride-hailing platform.

The move underscores Uber’s shift from abandoned self-driving ambitions to a supportive role in the AV ecosystem. While competitors race to build autonomous fleets, Uber’s sensor grid could redefine how AV companies access and utilize real-world data. The company’s focus remains on collaboration, not competition, as it seeks to shape the future of AI-driven transportation.