Uber's Play to Become the AV Industry's Data Layer: 25 Partners, an AV Cloud, and Millions of Drivers as Rolling Sensors
Uber CTO Praveen Neppalli Naga spent a few minutes at TechCrunch’s StrictlyVC event in San Francisco on May 1 describing what Uber actually wants to become in the autonomous vehicle era: the company that owns the data everyone else needs to build self-driving cars.
The ambition is not subtle. Naga wants to outfit Uber’s millions of human drivers with sensors, turning the ride-hailing network into a distributed real-world data-collection platform that any AV developer can tap. “The bottleneck is data,” he said. “Companies like Waymo need to go around and collect the data, collect different scenarios… The problem for all these companies is access to that data, because they don’t have the capital to deploy the cars and go collect all this information.”
Where It Stands Now
The sensors-on-driver-cars vision is not live yet. AV Labs, the programme Uber launched in January 2026, currently runs a small dedicated fleet of sensor-equipped vehicles operated by Uber directly, not its driver network. The regulatory groundwork isn’t there yet either — Naga acknowledged state-by-state clarity on what sensor deployment and data sharing actually means legally doesn’t exist in most jurisdictions.
What does exist: partnerships with 25 AV companies, including Wayve (London operations), and what Naga described as an “AV cloud” — a library of labeled sensor data that partner companies can query. Partners can also run their trained models in shadow mode against real Uber trips, simulating autonomous vehicle decision-making without putting an AV on the road. Uber is also making direct equity investments in AV companies in its partner network.
The AV cloud is already a meaningful infrastructure asset. Shadow-mode testing at Uber’s trip volume — millions of rides per day across hundreds of cities — provides AV developers with scenario exposure that custom fleets operating in a handful of permitted locations cannot match.
The Strategic Logic
Uber sold its Advanced Technologies Group to Aurora in 2020. Co-founder Travis Kalanick has since called that a mistake. The data-layer pivot is the company’s answer: capture the infrastructure position without the capital cost of building the cars.
Naga’s framing is that being a data supplier is more defensible than being a robotaxi operator. Every AV company competing on the ride-hailing platform needs training data. If Uber controls a uniquely scaled source of that data, it holds leverage over a sector that already depends on Uber’s marketplace to reach passengers.
“Our goal is not to make money out of this data,” Naga said. “We want to democratize it.” That positioning is familiar for early-stage infrastructure plays in tech. The commercial calculus will become clearer once the sensor fleet is deployed and AV developers have established dependencies.
What It Takes to Activate the Driver Network
Deploying sensors on driver-owned vehicles involves consent, compensation, and regulation that Uber hasn’t resolved. Drivers didn’t sign up to operate data collection platforms. State and federal rules on passenger privacy and commercial sensor use vary. The company is treating AV Labs as the proving ground to work out sensor integration before attempting to scale into the driver network.
If it works, the asset Uber is building is significant: the geographic spread of Uber’s 10,000+ city footprint, including cities and road conditions that no AV company’s dedicated test fleet has covered, accessible via a queryable data API. Specific scenario retrieval — edge cases, particular intersections at specific times of day, unusual weather conditions — is exactly what AV model training requires and what no individual operator has the scale to collect on demand.
The competitive pressure is real. Lyft has equivalent assets. Automakers with connected-vehicle fleets — GM’s OnStar, Ford, others — already sell driving data. What Uber has is density in urban environments where robotaxis will first deploy commercially, plus an existing relationship with 25 companies that need exactly what Uber is building.
The sensor grid is years away at minimum. The AV cloud is operating today.