The Cybercab launch will settle whether Tesla is an AI company or just a carmaker.

Tesla is putting a car with no steering wheel and no pedals on public streets in Austin today. It seats two, it’s gold, and it navigates entirely on cameras and AI. The company never said how big the launch would be. Then Elon Musk posted an AI-generated image of Cybercabs swallowing the city in a dust storm and captioned it “A Storm of Cybercabs.”

For months Tesla executives talked about caution and safety, though at times the concern read more like worry about how a public failure would look. That messaging has flipped. Ashok Elluswamy, who runs Tesla AI, wrote on X that streets won’t be the same anymore. When a user said Tesla was about to flood Austin, Musk replied: “Yes.” The state’s automated vehicle tracker listed 420 registered autonomous vehicles in Texas at publication time, and Tesla has staged Cybercabs across Austin and other U.S. cities.

The launch theatrics matter less than what follows. A working robotaxi network built on the Cybercab would be the first hard proof that Tesla has become an AI and robotics company rather than an automaker with expensive hobbies.

Musk claimed in 2016 that every Tesla rolling off the line could go fully autonomous with a software update. That was wrong. The hardware has been reworked several times, and Musk now admits millions of cars will need retrofits. Full Self-Driving (Supervised) got better, but liability still sits with the driver.

The Cybercab came out of a cheaper next-gen platform meant to halve Model 3 and Model Y build costs. Musk killed the manually drivable version. In April 2024 he approved mass layoffs to go, in his words, “balls to the wall for autonomy.” Everything else, he wrote, was a variation on a horse carriage.

The Model Y-based pilot launched in Austin in June 2025 with a safety employee up front. Since then Tesla has logged a few dozen incidents, mostly minor collisions with stationary objects, plus some teleoperator interventions at low speed. In July the company disclosed that paid miles on the network were shrinking.

Three things still cut against the bull case:

• Geofence problem: Musk once said geofenced areas mean you don’t have real self-driving.
• Edge case gap: Waymo runs 4,000 robotaxis and hits far more scenarios.
• Visibility risk: Model Ys blend in; gold Cybercabs will not.

If you’re modeling AV timelines into a logistics, fleet, or mobility bet, watch paid miles and geofence size, not launch-day volume. Tesla’s cost advantage per vehicle is real, but cheap hardware only compounds if the software holds at scale. Judge this launch in ninety days, not ninety minutes, and price in the possibility that the network stays small.

Failure in a Model Y is a footnote; failure in a gold two-seater is a headline.