On a Tuesday evening in late May 2026, a white Jaguar I-PACE with no one behind the wheel pulled to the curb on San Francisco’s Valencia Street, unlocked its doors, and waited for its passenger. The rider tapped her phone, climbed in, and was ferried 3.2 miles to the Mission Bay neighborhood without exchanging a word with a driver. Scenes like this now repeat more than 250,000 times every week across U.S. cities where Waymo operates commercially, according to ride-volume milestones the company has disclosed over the past year. What started as a cautious experiment in a Phoenix suburb has become something far more routine: a driverless taxi service woven into the daily transit fabric of San Francisco, Los Angeles, Phoenix, and Austin.
How the numbers grew this fast
Waymo first disclosed that it was completing more than 100,000 paid rides per week in mid-2024. By October of that year, the company told reporters the figure had climbed past 150,000. Growth continued as Waymo expanded its approved service zones and added vehicles to its fleet. The 250,000-per-week threshold, referenced in company communications, reflects a service that roughly doubled its weekly output in about a year.
A peer-reviewed safety analysis published on arXiv examined Waymo’s crash record across 56.7 million rider-only miles driven through the end of January 2025. That mileage figure is the most concrete, publicly available measure of cumulative driverless commercial driving in the United States. By mid-2026, the total is substantially higher, though Waymo has not released an updated cumulative number.
The mileage alone does not tell you how many individual trips occurred. Translating miles into rides requires assumptions about average trip length, route mix, and how many hours each vehicle spends carrying passengers versus repositioning empty. Waymo has shared weekly ride counts at specific moments but has not published a continuous, auditable time series.
Where the cars are legally allowed to operate
California, the state with the densest concentration of driverless activity, has granted Waymo authorization for both testing and commercial deployment across expanded zones. The California DMV’s permit map lists approved areas of operation that now include San Francisco, Los Angeles, and the San Francisco peninsula. The state authorized the Los Angeles and peninsula expansions in a separate action, confirming regulators’ willingness to let commercial operations grow into new metro regions rather than stay confined to a single test corridor.
Arizona, where Waymo has operated in the Phoenix metro area since 2020, remains the company’s longest-running commercial market. Austin, Texas, joined the roster more recently. Each jurisdiction sets its own permitting rules, creating a patchwork of local requirements layered on top of federal oversight.
Federal oversight: what gets reported and what stays hidden
The National Highway Traffic Safety Administration requires companies operating automated driving systems to report qualifying crashes under its Standing General Order, now in its third amendment. The order applies to both testing and commercial deployments, meaning the same reporting rules that governed early experiments now cover everyday paid rides. Companies must file reports within tight deadlines, and NHTSA can take enforcement action when they fall short.
That enforcement power is not theoretical. In 2024, NHTSA issued a consent order against Cruise after the General Motors subsidiary failed to fully report details of an October 2023 crash in which a pedestrian was struck and dragged by one of its vehicles in San Francisco. “The failure to provide complete and accurate information to NHTSA is unacceptable,” said NHTSA Deputy Administrator Sophie Shulman at the time the consent order was announced. Cruise subsequently suspended all driverless operations nationwide, and GM restructured the unit. The episode demonstrated that the federal reporting framework carries real consequences, but it also revealed that compliance is not automatic. Whether other operators have similar unreported gaps is, by definition, unknowable from public records alone.
For the public, the practical limitation is that NHTSA publishes only summary-level crash data. The detailed telemetry and full crash narratives companies submit are not released in real time. Riders and researchers can confirm that a reporting system exists and that it has been enforced, but they cannot independently audit raw incident data across all operators and cities.
What the safety research actually shows
The arXiv safety study is the largest independent look at Waymo’s crash record published to date. It segments crashes by type and severity, compares Waymo’s outcomes to human-driver benchmarks drawn from national datasets, and reports confidence intervals so outside researchers can test how sensitive the conclusions are to choices about geography, exposure, and crash classification.
The study found that Waymo’s rider-only vehicles had significantly lower rates of injury-causing crashes compared to the human-driver baseline across the cities examined. That result held across multiple crash types, though the confidence intervals widened for rarer, more severe crash categories where the sample size was smaller.
Two caveats matter. First, the study relies on crash data that Waymo itself collected and shared; no regulator has independently verified the completeness of the underlying dataset. Second, the human-driver benchmark is drawn from police-reported crash statistics, which carry their own well-documented reporting biases. The comparison is informative but not perfectly apples-to-apples.
“We think the right way to evaluate this technology is against the status quo, which is human driving, and by that measure the data is encouraging,” said Trent Victor, Waymo’s director of safety research, in a company briefing accompanying the study’s release. Independent safety researchers have echoed the need for caution in interpreting the results. “Fifty-six million miles is a meaningful sample, but it is not the same as proving safety across every possible driving scenario,” noted Missy Cummings, a robotics and autonomy researcher at George Mason University, in a public commentary on the findings.
The financial picture behind the fleet
Alphabet Inc., Waymo’s parent company, reports the unit’s finances within its “Other Bets” segment in quarterly filings with the U.S. Securities and Exchange Commission. Alphabet’s 10-Q for the quarter ending March 31, 2026, available through the SEC’s EDGAR archive, shows continued substantial investment in Waymo’s development and deployment. However, the filings do not break out city-by-city ride volumes, per-trip revenue, or fleet utilization rates. Investors can see that Alphabet is spending heavily on autonomous driving, but they cannot easily connect those expenditures to specific adoption metrics.
That opacity is common among companies scaling pre-profit technology platforms, but it leaves a gap for anyone trying to independently verify the pace of commercial growth.
What riders, regulators, and competitors are watching this summer
For someone considering hailing a robotaxi in San Francisco, Phoenix, or Los Angeles this summer, the practical picture is straightforward but incomplete. The vehicles are legally authorized in mapped service areas. Operators face mandatory federal crash-reporting requirements. The largest available safety dataset shows crash rates that compare favorably to human drivers using methods open to outside scrutiny. Fares are generally competitive with ride-hail services like Uber and Lyft, though pricing and wait times vary by zone and time of day.
What remains harder to pin down is the full distribution of risk across neighborhoods, times of day, and traffic conditions. No public dashboard ties ride volumes to safety outcomes across all operators and cities in real time. NHTSA collects crash data. State DMVs track permits. Alphabet reports financials. Academic researchers benchmark crash rates. But those four streams have never been merged into a single, continuously updated picture.
The competitive landscape is also shifting. With Cruise sidelined, Waymo faces less direct rivalry on U.S. streets than it did two years ago, though Amazon-backed Zoox is testing in several cities and smaller operators like May Mobility are running limited commercial routes. Internationally, companies like Baidu’s Apollo Go are scaling rapidly in Chinese cities, raising the question of whether U.S. regulatory caution will slow domestic deployment relative to global competitors.
The gap between what is verifiable and what is claimed will narrow only if regulators or companies publish more granular, standardized data. Until then, the public picture of driverless service rests on a mix of solid but narrow evidence and informed inference about a technology that is, week by week, becoming a normal part of how American cities move.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
