Tesla's steering-wheel-free Cybercab begins public robotaxi rides in Austin

Cybercab picks up public riders in Austin

Tesla has put its purpose-built Cybercab into operation as part of the robotaxi service it runs in Austin, Texas, with members of the public now taking rides, according to The Verge. The vehicle, first unveiled by Elon Musk at the 2024 "We, Robot" event, is a gold-tinted two-seater with gull-wing doors and no steering wheel, pedals or mirrors, and it embodies the company's plan for autonomous transport at scale.

Austin is where Tesla's robotaxi effort is most mature. The company has operated there since June 2025 with roughly 110 Model Y vehicles in service, according to Robotaxi Tracker data cited by The Verge. Elsewhere the fleet's size and supervision status have fluctuated for months, with only a handful of vehicles doing paid passenger trips at any given time.

A modified FSD, watched from afar

The Cybercab runs a modified version of Tesla's Full Self-Driving software. For the millions of Tesla owners who have bought FSD, the system remains a Level 2 driver-assist feature that requires human supervision. In the Cybercab there is no human fallback inside the car. Instead, remote operators monitor the vehicles and can, in theory, step in when something goes wrong. The Verge reports that Tesla has installed Starlink connections in every Cybercab to support that teleoperation link, which becomes the primary failsafe when a vehicle is disabled or stuck, since there is no wheel for a recovery crew to take hold of.

Cameras versus lidar

The launch is the biggest test yet of Musk's long-running argument that self-driving cars need only cameras. At Tesla's inaugural Autonomy Day in 2019, he called lidar "a fool's errand" and said anyone relying on it was "doomed," reasoning that humans drive using their eyes, so vehicles should too. Tesla dropped radar from its sensor suite in 2022, and The Verge reports that the company's own engineers objected, warning that cameras obscured by rain or sunlight could produce basic perception errors that lead to crashes. They were rebuffed.

Rivals take the opposite view, and the results so far favor them. Waymo, which pairs lidar with cameras and radar, operates a fleet of about 4,000 fully driverless vehicles across as many as 14 US cities, the largest such deployment in the country. Waymo co-CEO Dmitri Dolgov said at Y Combinator's Startup School that cameras have become good enough to approximate human driving, but that if the target is superhuman safety, "weak sensing just leads to a safety curve that flattens out way too early." Camera-only systems must infer depth from flat, two-dimensional images and are vulnerable to glare, fog and snow that lidar and radar can often see through.

Musk's counterargument is cost. A camera-only system is cheaper to mass produce, he contends, letting Tesla deploy Cybercabs faster than Waymo. The Verge notes the savings also come from omitting traditional controls and mirrors, and Musk promised at the 2024 event that consumers would eventually be able to buy a Cybercab for under $30,000 and run small fleets of their own.

Liability and regulatory hurdles

The stakes for Tesla have shifted. FSD and Autopilot have been linked to dozens of fatalities and thousands of crashes, and the company has spent years blaming drivers. With no driver aboard the Cybercab, Tesla now bears full legal responsibility when something goes wrong, as The Verge points out.

Selling the Cybercab to ordinary buyers would additionally require a federal exemption from safety rules that mandate steering wheels and pedals, a process that held up Amazon-owned Zoox for years before it won approval with heavy conditions. Tesla also lacks the California DMV permit needed to run fully driverless vehicles there, having not yet demonstrated safe operation. With no federal policy on driverless deployment and related legislation stalled in Congress, Tesla must fight for permission state by state, and sometimes city by city, leaving the Cybercab effectively confined to Austin for now.

Why it matters

This is the first real-world test of whether a camera-only, neural-network-driven system can handle full autonomy with no human in the car, the bet on which Musk has staked the company and his trillion-dollar pay package. If it works, Tesla could undercut sensor-laden rivals on cost and scale robotaxis far faster. If it does not, Tesla inherits the full liability for every failure, its regulatory climb gets steeper, and the industry consensus that full autonomy requires more than cameras is vindicated.