Waymo now operates its driverless ride-hailing service in ten American cities. That sentence would have seemed implausible five years ago, when "robotaxi" was a category that existed mostly in investor presentations and on closed test tracks. The latest expansion — opening to select riders in Houston, Dallas, San Antonio, and Orlando — brings Waymo into Sun Belt markets that are geographically very different from its San Francisco home turf: wider roads, more highway driving, less dense street grids. If the service works there, it works almost anywhere in the American southwest, which is a meaningful technical statement.
The scale numbers are becoming harder to dismiss. Waymo completed fourteen million trips in 2025, and the company is on track for over a million rides per week by year end. Motional, a competitor that had gone through a major reset in 2024, just joined the Uber app in Las Vegas — another Sun Belt city, another data point that the market is real. Uber, meanwhile, has structured its entire autonomous vehicle strategy around being the platform others deploy on: its new Autonomous Solutions division handles the operational complexity of running a robotaxi fleet, and its AV Labs division aggregates driving data for partners. The bet is that Uber becomes the Android of robotaxis — the distribution layer that benefits regardless of whose hardware wins.
What's striking about the current moment is how it differs from the long trough that preceded it. For most of the period from 2019 to 2023, the autonomous vehicle industry was defined by retreats: cancelled projects, scaled-back promises, billions written off. The companies that survived — primarily Waymo, which had the advantage of Alphabet's patient capital — spent those years quietly accumulating miles and improving edge case handling. The result is a technology that, in a limited set of geographies and conditions, actually works at commercial scale. Ten cities is not everywhere, but it is also not nowhere.
The remaining limits are worth being clear about. Waymo's service areas within each city are still geofenced — it is not point-to-point anywhere in Houston, but point-to-point within a defined zone. Adverse weather handling remains a constraint; the Sun Belt cities chosen for expansion are not coincidentally among the driest and most weather-stable in the US. Washington DC, on the expansion roadmap for later in 2026, will be a more demanding test. And the economics, while improving, are not yet demonstrably better than human-driven rides at current scale — the capital cost of the vehicles and the sensor stack remains substantial.
None of that changes the basic fact of the milestone. Self-driving cars are a service that real people are using, in ten cities, right now, without a human backup driver. The technology went from "perpetually five years away" to "quietly operational" without the dramatic single moment that the coverage always promised. That is probably how most genuine technological transitions actually happen — not with a bang, but with an accumulation of city announcements until one day the number is ten and the question shifts from "will it work" to "how fast will it scale."