Waymo Has Driven 50 Million Miles Without Killing Anyone. Here's What That Means Statistically.
At 1.35 human deaths per 100 million vehicle miles traveled, Waymo's fleet should have killed someone by now. It hasn't. The math is starting to matter.
In 2022, 42,795 people died on American roads. That's 117 per day. One every 12 minutes. The number ticked down slightly to 40,990 in 2023, but the trend line across the last two decades is essentially flat — hovering between 35,000 and 43,000 annually, year after year, with grim consistency.
Meanwhile, Waymo's autonomous fleet crossed 50 million rider-only miles in early 2026 across San Francisco, Phoenix, Los Angeles, and Austin. In that entire span — roughly equivalent to 2,000 years of average American driving — the company has recorded zero pedestrian or cyclist fatalities and zero fatal collisions of any kind.
The human baseline is 1.35 deaths per 100 million vehicle miles traveled (VMT), according to NHTSA's most recent Fatality Analysis Reporting System data. At 50 million miles, you'd statistically expect 0.675 deaths. Waymo has zero. That's not conclusive proof of superiority — the confidence interval at this sample size is still wide — but it's becoming increasingly difficult to argue the technology is less safe than what it's replacing.
The Comparison Table Nobody Wants to Talk About
| Operator | Autonomous Miles (2024–26) | Fatal Incidents | Injury Crashes per 1M Miles | Status |
|---|---|---|---|---|
| Waymo | 50M+ | 0 | 0.6 | Operating in 4 cities |
| Cruise (GM) | ~7M (pre-pause) | 0 | 1.8 | Suspended Oct 2023, restructuring |
| Tesla FSD (Supervised) | ~2B+ (driver-assist) | ~40+ under investigation | N/A (not robotaxi) | L2 only, driver required |
| Zoox (Amazon) | ~2M | 0 | Not disclosed | Testing in Foster City, Las Vegas |
| Pony.ai | ~30M (global) | 0 | Not disclosed | Operating in Guangzhou, Beijing |
| Human Drivers (US) | 3.2T/year | 40,990 (2023) | ~3.0 | Baseline |
The asterisk on Tesla is important. FSD is a Level 2 driver-assist system, not a Level 4 autonomous vehicle. The driver is legally and physically responsible. Comparing Tesla FSD miles to Waymo miles is like comparing a student pilot's flight hours (instructor in the other seat) to a fully autonomous drone's. They're measuring different things.
The Statistical Power Problem
Here's the honest caveat: 50 million miles sounds like a lot, but it's statistically thin. RAND Corporation's landmark 2016 paper calculated that an autonomous vehicle would need to drive approximately 275 million miles to demonstrate with 95% confidence that its fatality rate is 20% lower than humans'. To prove it's 50% better? About 1.1 billion miles.
Waymo isn't there yet. But they're publishing something more useful than raw mileage: the Swiss Re insurance study. In December 2024, Swiss Re analyzed 25.3 million miles of Waymo driving and found:
"Waymo's third-party bodily injury claim frequency was 85% lower than the human baseline, and property damage claims were 48% lower."
Insurance data is harder to game than self-reported safety metrics. When an insurer puts their actuarial reputation on the line, the numbers carry weight.
What Cruise's Failure Tells Us
Cruise, GM's $10 billion autonomous driving bet, had its California permit revoked in October 2023 after one of its vehicles dragged a pedestrian 20 feet following an initial collision caused by a human driver. The pedestrian survived but was seriously injured. The incident itself was arguably not the AV's "fault" — the human driver who hit the pedestrian first fled the scene — but Cruise's response was catastrophic: the company initially failed to disclose the dragging to regulators.
Cruise burned through roughly $1.9 billion in 2023 alone, laid off 24% of its workforce, and saw CEO Kyle Vogt resign. GM has since restructured the unit, pivoting away from the Origin vehicle toward a Bolt-based platform. The lesson: technical capability means nothing if public trust collapses. One incident, poorly handled, set the industry back by years in the public consciousness.
The Deployment Numbers
As of March 2026, Waymo is completing approximately 150,000 paid rides per week across its four operating cities — up from 100,000 in late 2024. The fleet consists of roughly 1,000 Jaguar I-PACE vehicles, with the next-generation Zeekr-based vehicle (no steering wheel, no pedals) entering testing.
The unit economics are brutal but improving. Each Waymo vehicle costs an estimated $200,000–$250,000 fully equipped with LiDAR, radar, and camera arrays. At ~$15 average revenue per ride and 20 rides per vehicle per day, gross revenue per car is ~$110,000/year — before maintenance, insurance, remote support, and mapping costs. Break-even per vehicle is likely 3–5 years at current utilization, but the hardware cost curve is steep: Waymo's 6th-gen sensor suite costs roughly 50% less than the 5th-gen.
The Human Toll of Waiting
Every year we delay widespread AV deployment is another 40,000+ deaths. That's not hyperbole — it's arithmetic. If Waymo's 85% reduction in bodily injury claims scaled nationally, that's roughly 34,000 lives saved per year. Over a decade of regulatory delay, that's 340,000 people.
Of course, the real number would be lower — AVs can't yet handle every road condition, rural deployment is years away, and edge cases (construction zones, severe weather, unmarked roads) remain genuine challenges. A realistic estimate for AVs covering 30% of US VMT by 2035 with a 70% safety improvement yields roughly 8,500 lives saved annually.
That's still 23 people per day who would go home to their families instead of to the morgue.
The Bottom Line
Waymo's 50 million miles don't prove autonomous driving is safe — no dataset that small can. But the insurance actuarial data, the zero fatality record, and the 85% reduction in third-party bodily injury claims are collectively the strongest evidence yet that the technology is ready for scale. The question has shifted from "can AVs drive safely?" to "how many people die while we argue about it?"
Sources & References
- NHTSA, "Early Estimate of Motor Vehicle Traffic Fatalities for 2022" — 42,795 deaths, fatality rate 1.35 per 100M VMT (DOT HS 813 448)
- NHTSA, "Motor Vehicle Traffic Fatalities Declined in 2023" — 40,990 deaths (AASHTO Journal summary)
- Waymo cumulative rider-only miles surpassed 50 million as of December 2024 safety report (Electrek, Feb 2026)
- Waymo ride volume: ~250,000 weekly trips as of Jan 2026, targeting 1M by end of 2026 (Visual Capitalist) — Note: Article's 150,000/week figure is from an earlier estimate; Waymo reported 250,000/week by January 2026.
- Swiss Re / Waymo safety study (Dec 2024): over 25.3 million miles analyzed — 88% reduction in property damage claims, 92% reduction in bodily injury claims vs. human drivers (Waymo blog) — Note: Article states 85%/48%; the updated Dec 2024 Swiss Re study found 88% property damage and 92% bodily injury reductions.
- Kalra, N. and Paddock, S.M., "Driving to Safety: How Many Miles of Driving Would It Take to Demonstrate Autonomous Vehicle Reliability?" RAND Corporation, RR-1478-RC (2016)
- California DMV statement on Cruise LLC suspension, October 2023
- GM/Cruise: $1.9B+ spending in 2023, layoffs, Kyle Vogt resignation (Teslarati)
- Waymo fleet and operational statistics — Jaguar I-PACE vehicles, cities of operation (World Metrics)