The numbers are finally in, and they are not good for Tesla's autonomous vehicle ambitions.

NHTSA filings reviewed by Electrek reveal that Tesla's robotaxi fleet has experienced 14 crashes since launching in Austin last summer, with five new incidents recorded in December 2025 and January alone. Based on Tesla's own reported mileage data - approximately 800,000 cumulative paid miles through mid-January - that works out to roughly one crash every 57,000 miles.

Here's why that number matters: Tesla's own Vehicle Safety Report states that average U.S. drivers experience a minor crash every 229,000 miles. The math is straightforward and uncomfortable. Tesla's robotaxis are crashing at approximately four times the rate of human drivers.

The technology is impressive. I've said that before about Tesla's hardware stack, and I'll say it again. The camera-only approach to autonomy - which many experts argued was insufficient - has gotten further than critics predicted. But "further than predicted" and "safe enough for commercial deployment" are not the same thing.

The crashes themselves tell an important story. The five incidents from the reporting period involved collisions at low speeds - 4 to 17 mph - with fixed objects, a bus, and a truck, plus two backing incidents. These aren't high-speed catastrophic failures. They're low-speed urban driving failures. That's actually more concerning for robotaxi deployment, which is precisely the low-speed urban driving environment where the vehicle is supposed to excel.

Then there's the transparency problem. Tesla has redacted the incident narratives in NHTSA filings, citing confidential business information. This stands in notable contrast to Waymo, which has been far more forthcoming about its incident data. When a company building autonomous vehicles argues that safety incident narratives are trade secrets, that's a regulatory argument worth scrutinizing.

For context: Waymo recently had NHTSA open investigations into separate incidents involving a child pedestrian and a school bus. That's not a clean record either. But Waymo's operational data suggests dramatically lower collision rates per mile than Tesla's robotaxi numbers, and the company has been more transparent about what happened when things went wrong.

Elon Musk has been promising full autonomous capability since 2016. The robotaxi commercial launch was supposed to be the proof point - the moment the technology graduated from demo to product. A four-times-higher crash rate than human drivers is not a graduation. It's a grade that says "needs significant improvement before the next evaluation."

The regulatory question is now unavoidable. Texas and other states have given Tesla considerable latitude to operate its robotaxi service. The data now exists to evaluate whether that latitude was warranted. Regulators who see these numbers and choose not to act will have a difficult case to make if the incidents escalate.

The technology is impressive. The question is whether it's safe enough - and right now, the data says no.