AUSTIN, Texas — Tesla's driverless robotaxi network has crossed a milestone that reframes the entire debate about its rollout pace: more than 380,000 fully unsupervised miles with, by the company's account, zero notable incidents. The figure, disclosed alongside Tesla's Q2 results, points to a service scaling on the strength of its safety record rather than raw speed.
A spotless record at scale
Tesla AI chief Ashok Elluswamy told investors the program has "an impeccable safety record," logging over 380,000 miles of unsupervised driving across six cities in two states. Counting rides with an in-vehicle safety monitor, cumulative paid robotaxi mileage is approaching 2.5 million. Musk added that mileage is now compounding at more than 10% a week, a pace that turns a modest base into a large fleet footprint quickly.
That clean record is not an accident of luck but of design. Tesla has deliberately kept its fleet small and its expansion measured, prioritizing zero harm over headline numbers. The same discipline shows up in the software: robotaxis are already running early builds of the next-generation stack, building on a clean safety record on early FSD v15.
Why the cautious pace is a feature
Elluswamy explained that expanding across many cities, rather than saturating one, proves the software generalizes without heavy per-city tuning. In practice, that means Tesla's system is learning to drive anywhere rather than memorizing a single map, an advantage as the company targets rapid multi-market growth. Musk framed the constraint plainly: a single injury would draw worldwide scrutiny, so Tesla is scaling "as fast as humanly possible" while refusing to compromise on safety.
