Waymo is taking its robotaxi service to San Diego, Las Vegas and Detroit, running a mixed fleet of all-electric Jaguar I-Pace cars on the fifth-generation Waymo Driver and next-generation Zeekr RT vehicles on the sixth.
Two of those cities are routine. The third is the interesting one.
The numbers Waymo publishes
The company's safety claims are specific, and they have grown with the mileage. At over 100 million fully autonomous miles, Waymo reported 88% fewer serious-injury crashes than human drivers. A later study across more than 220 million autonomous miles reported:
- 94% fewer crashes causing serious or fatal injuries
- 82% fewer crashes in which an airbag deployed
- 82% fewer crashes involving any reported injury
On scale: the company was past 250,000 rides a week as of April, and co-chief executive Tekedra Mawakana has said the goal is one million trips a week by the end of 2026.
What those figures do and do not establish
They are a genuinely strong result and should not be waved away — 220 million miles is a large sample, and the reductions are far too big to be noise.
Two qualifications belong next to them. First, they are published by the company, using its own methodology, against a human-driver baseline it constructs. That is normal in this industry because no regulator collects comparable data, but it is not independent verification.
Second, and more substantially: almost all of that mileage was accumulated in Phoenix, San Francisco, Los Angeles and Austin. Those are places with mild winters, little snow, and no sustained road salt or ice. San Diego and Las Vegas extend the same conditions.
Detroit does not. Snow covers lane markings, which is how these systems locate themselves laterally. Slush and salt spray coat the sensors. Ice changes stopping distances in ways a vehicle must anticipate rather than discover. A safety record built on dry tarmac does not transfer automatically, and Waymo has far less public autonomous mileage in genuine winter.
That makes Detroit the most informative deployment on the list — and the one worth watching, rather than the headline about three cities.
The regulatory step nobody reported
Neither Nevada nor Michigan is cleared yet. Waymo needs testing approval from Nevada's Department of Motor Vehicles and a transportation network company permit in Michigan before operating. The announcement is an intention, not a launch date, and in this sector the two have diverged before.
What the million-a-week target actually implies
Going from 250,000 to a million weekly trips in under a year is roughly a fourfold increase. That is not primarily a software problem. It is vehicles, depots, charging, cleaning, remote-assistance staff and maintenance — the unglamorous operations layer that decides whether an autonomous fleet is a business or a demonstration. The Zeekr RT appearing alongside the Jaguars is a sign of where the constraint actually sits: you cannot scale a fleet faster than you can buy and fit out cars.
Sources: Waymo's safety reporting; TechCrunch; Automotive World; eWeek; Claims Journal; and Prothom Alo's original report.
