Waymo challenges Tesla's driverless approach before Cybercab launch

Waymo used a blog post and an interview with Axios last week to argue that fully autonomous vehicles cannot work safely using only cameras and AI — an implicit rebuke of Tesla’s approach — just one week before Tesla was expected to formally introduce its two-seater Cybercab in Austin, Texas, on September 3. The Alphabet-owned company did not name Tesla directly, but the timing left little doubt about the target.

Waymo also announced three new markets on Tuesday, extending a robotaxi network that already serves customers in more than a dozen U.S. cities. The company currently operates around 4,000 robotaxis across 14 U.S. cities, delivering 500,000 paid trips per week.

Srikanth Thirumalai, the Waymo vice president who oversees the company’s driving software, wrote in the blog post that “cameras are incredible, but they aren’t enough,” adding that after more than 200 million real-world miles, the data is clear that safe, fully autonomous operation at scale requires combining cameras, lidar, and radar. He argued that Tesla’s approach — a pure end-to-end neural architecture that takes in raw pixels and outputs steering commands directly — risks black-box failures, and told Axios that even the most capable AI models, with trillions of parameters, still hallucinate, unlike software, which cannot simply be rebooted once deployed in a physical vehicle.

The claims triggered a weekend-long fight on social media, with Tesla-aligned analysts describing Waymo’s approach as an expensive, sensor-heavy setup that AI at scale would eventually make irrelevant. Tesla, for its part, has spent years betting on cameras and AI alone, with CEO Elon Musk long dismissing lidar as a “crutch.”

Tesla has registered Cybercabs — a two-seat vehicle with no steering wheel or pedals — with the Texas DMV ahead of its September 3 event, and a recent filing shows it aims to build more than 125,000 of them annually. Musk once promised 1 million Tesla robotaxis on the road by 2020; that never happened, and the company has spent the past year running a modest Robotaxi pilot with modified Model Y SUVs in Texas and Florida, only recently removing safety monitors from most of those vehicles.

The stakes are high, with a robotaxi market that could be worth hundreds of billions of dollars. Waymo’s approach is also the costlier one: it buys and outfits vehicles built by other manufacturers, including paying import taxes on cars like the Zeekr-made Ojai, while Tesla builds its own cars and is betting its AI can eventually get by on cameras alone.

Why it matters

If Tesla can prove the Cybercab operates safely at scale using only cameras, it would undercut Waymo’s costlier sensor-based model and could reshape the economics of the emerging robotaxi industry.