Uber is pursuing an ambitious long-term strategy to turn the network of millions of drivers globally into a mobile sensor system, aiming to collect real-world data to serve the self-driving car and artificial intelligence industries.
This information was revealed by Mr. Praveen Neppalli Naga, Uber's Chief Technology Officer, at the StrictlyVC event organized by TechCrunch in San Francisco on April 30, 2026.
According to Mr. Praveen Neppalli Naga, this is a natural development from the AV Labs program, an initiative that Uber announced earlier this year to research and collect data to serve self-driving cars.
Currently, AV Labs operates a specialized vehicle fleet equipped with dedicated sensors, completely separate from the Uber driver network.
However, the company's long-term goal is to expand its scale by integrating sensor devices directly into driver's cars. If only a small part of these millions of vehicles are converted, Uber can create a huge data collection system, far beyond the capabilities of any self-driving car company.
According to Mr. Naga, the biggest bottleneck of the self-driving car industry today is no longer the foundation technology, but the data itself.
Companies need to collect information from countless real-life situations from busy intersections to school areas at different times to train AI models.
However, deploying enough vehicles to collect this data requires very large costs, exceeding the capacity of many businesses.
In that context, Uber sees the opportunity to become a "data layer" for the entire self-driving car ecosystem.
This is also considered a strategic step, especially when the company has abandoned its ambition to develop its own self-driving cars for many years. This decision once caused controversy, as many opinions worried that Uber could gradually lose its position if self-driving cars became popular.
To strengthen its new role, Uber has partnered with about 25 self-driving car companies, including Wayve in London. Along with that, the company is building a platform called the "self-driving car cloud", which stores and labels sensor data so that partners can access, query and use it during model training.
Not only stopping at providing data, this system also allows companies to test their models in "glare mode". That is, algorithms can run in parallel with Uber's actual trips to simulate how a self-driving car will react, without having to bring the self-driving vehicle onto the road.
With the advantage of owning large-scale data and having invested in many self-driving car companies, Uber can completely create significant leverage in the industry.
