VERTEBRATE.ai

Tesla's robotaxi fleet has grown eightfold in recent weeks, Elon Musk told followers online, a figure offered with the usual Muskian lack of footnotes — no baseline number, no city-by-city breakdown, no third-party verification. In the same breath, Musk teased a thruster-equipped variant of the Roadster, a vehicle that has now been "coming soon" for the better part of seven years. The announcements landed together, as Tesla announcements often do, one dazzling, one unverifiable, both engineered to keep the stock-watching public looking up rather than down at the pavement.

Down at the pavement is where the trouble is. Tesla's robotaxis, it turns out, are not simply humming along on autonomous software while engineers in Austin and Palo Alto sip coffee. Many are being steered, at least part of the time, by remote human operators — a teleoperation layer that was supposed to serve as a safety net for the rare edge case. Instead, according to reports tracking the fleet's operations, these remote-piloted vehicles are racking up a notable number of collisions: scraped bumpers, curb strikes, the unglamorous vocabulary of fleet-damage reports rather than headline-grabbing catastrophes, but a pattern nonetheless.

That pattern matters more than the eightfold-growth claim, because it cuts against the entire premise Tesla has sold investors for a decade: that the company's camera-only, AI-driven approach would leapfrog competitors who lean on lidar, high-definition mapping, and yes, human remote operators as a crutch. If Tesla's own fleet is now depending on teleoperation — and crashing while doing so — the company isn't leapfrogging anyone. It's running the same experiment as Cruise and Waymo did in their messier years, with the same unglamorous failure mode: a person in an office, watching a screen with latency, making split-second calls about a car they cannot feel or hear.

The stakes extend beyond Tesla's balance sheet. Robotaxi fleets operate on public streets, among pedestrians and cyclists who never agreed to be test subjects for a remote-driving interface. Regulators who have already slow-walked Tesla's robotaxi permits in multiple states will take note of collision data more readily than they take note of Musk's thruster jokes. And for an industry still trying to convince a skeptical public that autonomous vehicles are safer than human drivers, every scraped quarter-panel is a data point the skeptics will cite for years.

There's also a simpler story here, one about attention management. A thruster-powered Roadster is the kind of detail that generates headlines and forum threads; a spreadsheet of remote-operator fender-benders does not. Tesla has long understood which of those gets covered more breathlessly. The gap between the two — between the fleet Musk describes and the fleet that is actually, haltingly, bumping its way through city traffic — is where the real robotaxi story lives, and it's a story investors and regulators alike would do well to read past the headline.