Autonomous driving is broken. And insurance companies are the first major signal: they’ve been quietly pulling out of the market because the technology isn’t reliable enough to price risk accurately.
When insurance companies stop underwriting a technology, it’s not because they’re being difficult. It’s because the risk profiles are too uncertain to calculate. Tesla’s Full Self-Driving, Waymo’s robotaxis, Cruise’s shutdown — they all share the same problem: no one can reliably predict what their self-driving cars will do in every situation.
The Safety Data Nobody Wants to Emphasize
Tesla’s FSD has accumulated millions of miles of autonomous driving data. And still, human intervention was required far too often. Waymo operates robotaxis — but only in a handful of cities, in controlled conditions, with remote operators ready to step in. Cruise pulled out of the entire market after safety violations. The record isn’t promising.
Why Insurance Piled Out First
Insurance works because actuaries can calculate risk based on historical data. With autonomous driving, historical data doesn’t exist — the technology is too new, and the edge cases are too numerous. You can’t price the risk of a self-driving car making a novel error because novel errors are precisely what makes them errors.
The Reality Check
We’re years away from fully autonomous vehicles that can operate everywhere, any time, without insurance companies panicking. The companies selling them now are selling a technology that’s not quite ready for prime time. The future is real. The timeline is wrong.
