Waymo at 500,000 Rides a Week: How a 4,000-Car Fleet Scaled to 15 Cities

Waymo is now averaging 500,000 paid driverless rides per week across 15 U.S. cities, with about 4,000 vehicles in service as of September 24, 2026, reported by TechCrunch. The figures describe continuous paid service at national scale.

In September 2024, the network ran in three cities: Phoenix, Los Angeles and San Francisco. It now runs in 15 cities, five times that city count.

The fleet is concentrated. About 80% operates in California and Texas, leaving about 800 vehicles for the rest of the country, including Arizona and Florida.

In Texas, Waymo had 1,102 autonomous vehicles registered as of September 24, 2026, after a 49% increase in the prior three weeks. TechCrunch based its Texas counts on state vehicle registrations and data from the Texas Autonomous Vehicle Fleet Tracker, which makes Texas one of the few states where fleet size can be checked from public records.

The Texas fleet stood at about 600 vehicles in June 2026. It was above 700 by the end of August 2026, then rose in September to 1,102. That September rise accounts for much of the net national growth to the current 4,000-vehicle total.

Paid service in Texas began in Austin in March 2025 through a partnership with Uber, which lets riders hail Waymo cars through the Uber app. Waymo later expanded to Dallas, Houston and San Antonio. On July 28, 2025, Waymo said it would start autonomous ride-hailing in Dallas in 2026, according to Reuters, a plan that lines up with the multi-city Texas service now reported.

About one-third of the Texas fleet now consists of Ojai minivans. The Ojai is a modified Zeekr RT minivan built on Zeekr's SEA-M platform, the base electric chassis, with Zeekr owned by China's Geely Holding Group. It carries Waymo's sixth-generation self-driving system, the current set of sensors, computers and software.

Base Zeekr vehicles are shipped to the U.S. without Chinese connected-car technology, the wireless hardware that can send data outside the car. Waymo installs its self-driving system at its factory in Arizona. Waymo spun out of Google and is majority-owned by Alphabet.

For context, Waymo had more than 2,500 vehicles as of January 31, 2026, when it was seeking to raise about $16 billion in a round valuing it at nearly $110 billion, according to Reuters. By May 28, 2026, it had 577 automated vehicles registered in Texas, more than 13 times Tesla's Texas total at that point, according to Bloomberg. Tesla had registered 42 automated vehicles in its driverless Robotaxi service in Texas in May 2026, according to CNBC.

The broader context here is operations rather than tests. Weekly paid rides, registered fleet size and state additions measure availability, maintenance throughput, charging and cleaning, mapping and validation, and keeping cars in paid service. The constraints that matter now are dispatch efficiency, matching the closest free car to a rider, vehicle utilization, incident response, and unit economics per mile, whether each mile brings in more than it costs.

In my view, three details deserve more attention than the headline ride count. First is the Uber channel in Austin. Third-party distribution lowers the cost of finding riders and tests whether robotaxis can fit into existing demand instead of needing a separate app habit. Second is the Arizona installation step for the Ojai. Installing the self-driving system in the U.S. on vehicles shipped without connected-car hardware addresses supply-chain and data-governance questions that grow as fleets scale. Third is concentration. Keeping most cars in California and Texas simplifies operations, staffing and regulatory work, but performance in those states will shape public views nationally.

Worth flagging is what scale enables next. A fleet doing half a million paid rides per week produces a steady stream of construction zones, weather changes and passenger needs. That volume supports steady improvement in routing, pickup and drop-off behavior, rider support and fleet-management software, the routine systems that decide whether a service feels reliable. My kids learned to trust new technology not when a demo impressed them but when it worked the same way twice in a row. Robotaxis face the same test, and the Texas growth suggests Waymo is now optimizing for repetition.