Waymo is no longer principally a technical demonstration. The evidence shows a substantial, operating U.S. robotaxi network: roughly 500,000 paid rides a week, 15 million rides during 2025, and more than 20 million rider-only trips. 1,9,11,12,13,14,15,20,22,23,24,25,28,29,30,33,34,37,38,40,41,55,56,10,19,45,59,53 Its vehicles drive approximately four million miles weekly and have accumulated more than 200 million fully autonomous miles. 21,60,2,16,29,35,36,45,59 In late September 2026, the service footprint had reached 15 U.S. cities, up from Phoenix, Los Angeles and San Francisco a year earlier. 42
That is the material conclusion for Alphabet: Waymo has crossed the proof-of-concept threshold. The investment question is now whether it can turn a demonstrated autonomous service into a repeatable, safe and economically durable mobility business. The proof is in the performance, not the promise—and operating scale is better evidenced than financial conversion.
Scale Has a Physical and Geographic Shape
Waymo’s fleet is plainly measured in thousands, although the precise count is unsettled. A three-source estimate put the fleet at about 3,000 robotaxis, while more recent reporting cited about 4,000 autonomous vehicles nationwide and a 16-source claim stated that Waymo had more than 4,000 vehicles. 5,7,57,61,26,27,28,31,32,36,38,47,49,60,63 The discrepancy is not resolved in the supplied record; it may reflect different reporting dates or vehicle definitions. It would therefore be unsound to treat either point estimate as definitive. What the evidence does establish is a large fleet by robotaxi standards, supporting descriptions of Waymo as the largest U.S. operator. 61
The fleet is not yet broadly dispersed. About 80% of Waymo robotaxis are located in California and Texas. 42,48,50 Texas is the clearest example of concentrated scaling: it had 1,102 registered autonomous vehicles on September 24 after a 49% increase over the preceding three weeks. 42,50,57,42 Waymo began commercial Austin service through Uber in March 2025, then expanded to Dallas, Houston and San Antonio. 42,50,57,63,42 This looks less like indiscriminate geographic expansion than an effort to establish repeatable operations in large markets before distributing equipment more widely. It also means that state-level operating and regulatory conditions retain outsized importance.
Vehicle supply is central to that effort. The Ojai, a modified Zeekr RT minivan with Waymo’s sixth-generation driving system, represented roughly one-third of the Texas fleet in late September. 42,50,42 Waymo’s stated purpose for the vehicle is to reduce costs and support profitability. 42 Its Mesa factory is being scaled toward annual output in the tens of thousands of Waymo-enabled vehicles, and one September report said the company was on track to bring 5,100 Ojais to the United States by year-end. 63,42,50,57 This is a credible route to more fleet capacity, but it is not a safety valve against industrial risk: the vehicles are imported from China and retrofitted in Arizona, exposing the ramp to trade-policy risk and tariff-related import costs. 42 A larger depot does not by itself make a transport system economical; utilization, durability, maintenance and vehicle cost remain operating requirements rather than proven outcomes. 49
Safety Advantage Does Not Eliminate the Edge-Case Burden
Waymo’s comparative safety case is the strongest support for its commercial thesis. The company reports 82% fewer injury-causing crashes than human drivers over more than 270 million miles, equivalent under its benchmark assumptions to 841 fewer such crashes. 60 It also reports a 90% reduction in serious injuries. 33,59,60 Importantly, the evidence is not confined to company reporting: an IIHS analysis published in July found 68% fewer police-reportable crashes per mile than human drivers, based on roughly 50 million Waymo miles from 2021 through 2024. 60
Taken together, these measures support an aggregate safety advantage against the reported human-driving benchmarks. They do not settle the full safety question. A service operating at greater scale will encounter more unusual conditions, and those conditions are where public legitimacy and regulatory confidence are won or lost. Waymo reported 94 new crash involvements in September 2026, after 104 in August and 102 in July; the September reports included one fatality and three serious injuries. 43 The material supplies neither fault findings nor miles-adjusted rates for those monthly reports, so they cannot directly overturn the broader comparative evidence. They do, however, demonstrate that absolute incident exposure rises alongside deployment.
The more consequential warning signs concern operational resilience at the boundary of the operational design domain. Recent scrutiny has included robotaxis disrupting traffic, entering crime scenes, impeding emergency services and sometimes requiring first responders to drive them manually. 61 Waymo recalled 3,791 vehicles after a robotaxi entered floodwater in San Antonio and 3,871 vehicles after vehicles drove into active freeway construction zones; it later paused freeway service because of construction-zone behavior. 60,52
This is not a contradiction in the simple sense of “safe” versus “unsafe.” It is the more difficult engineering tension between favorable aggregate results and the particular failures that emergency responders, cities and riders must confront. Safety engineering is what happens between the edge cases. Waymo’s position that remote assistance leaves the self-driving system in control, together with its commitment to comply with California Senate Bill 1246, indicates that governance and emergency interaction are becoming strategically material alongside driving performance. 61
Economics Remain the Unverified Link
Investor confidence is substantial. Waymo, an Alphabet subsidiary rather than a separately listed security, raised $16 billion in February 2026 at a reported $126 billion valuation, a figure supported by 11 sources. 3,4,6,8,15,16,17,18,54,10,19,33,35,38,39,53 Yet valuation is a measure of expectations, not a substitute for an operating statement. The material provides no standalone revenue, profitability, financial-statement, cash-flow or unit-economics data for Waymo. 64,63,64,51,63,64,46
One later excerpt reports a $350 million revenue run rate and an implied $152.72 billion valuation, but it supplies no revenue split by city or product. 52 Those figures cannot establish profitability, particularly because other recent material characterizes Waymo as still loss-making. 65 The financial evidence is therefore materially weaker than the operational evidence.
This distinction should discipline the commercialization narrative. More rides can create more driving data and, potentially, more revenue, but neither outcome establishes positive unit economics. 65 Bank of America estimated that robotaxi bookings could reach roughly $6 billion by 2028, or about 5% of the U.S. ride-hailing market, and identified fares below $2 per mile as an important adoption threshold. 63 Such projections describe the size of a possible market, not Waymo’s realized margin structure. Certification should be a floor, not a ceiling; likewise, operational scale should be a precondition to evaluating economics, not a proxy for having achieved them.
Waymo’s distribution choices make this conversion question sharper. Uber enabled demand access in Austin, but Waymo plans to launch its own app in Austin and Atlanta in January 2028, ending Uber’s exclusive arrangement in those markets. 42,63 The company is thus balancing a partner’s demand aggregation against its own control of customer experience and economics. That tension is structural: Uber can place partner-operated autonomous vehicles alongside human drivers on the same marketplace, maintaining coverage when autonomous supply is incomplete, while autonomous operators can in turn build proprietary networks that bypass Uber. 44,62
International Expansion Is a Validation Program, Not Yet Financial Guidance
Waymo’s international agenda should be read as a deliberate localization sequence, rather than as a simple replication of its U.S. service map. Tokyo provides the clearest example. Waymo began operating on public roads in 2025, initially using human-driven Jaguar I-PACE vehicles for data collection. 58 It subsequently said its vehicles had driven autonomously under trained Nihon Kotsu personnel supervision while adapting to Tokyo roads. 58 The target is a fully autonomous commercial taxi service in 2027 with GO and Nihon Kotsu, beginning with a small fleet and expanding toward about 100 vehicles in selected neighborhoods. 63
The qualification matters: further technology validation and regulatory approval remain explicit conditions to that Tokyo objective. 63 Local operating knowledge and institutional permission are therefore part of the product, not peripheral administrative steps.
Singapore follows the same logic on a longer timetable. The plan calls for mapping during 2027, adaptation to local traffic patterns, and work with the Ministry of Transport and Land Transport Authority before a planned fully autonomous ride-hailing launch in 2028. 53 The record gives no Singapore fleet size, operating area, pricing or revenue opportunity. 53 International plans expand the commercial ambition, but they do not yet provide a reasonable basis to forecast international economics.
What Alphabet Must Prove Next
Waymo has established the difficult first proposition: it can operate fully driverless passenger service at meaningful scale. The evidence for paid rides, autonomous mileage, fleet deployment and U.S. expansion is unusually strong. 1,9,11,12,13,14,15,20,22,23,24,25,28,29,30,33,34,37,38,40,41,55,56,21,60,26,27,28,31,32,36,38,47,49,60,63,42,49,56 The next phase is more exacting. Alphabet must show that the system can expand without transferring unresolved edge-case risk into new cities, that vehicle supply can lower rather than merely relocate costs, and that ride volume can become profitable economics rather than a costly measure of technical progress.
The practical test is not whether autonomous mobility is possible. It is whether Waymo can make each new market a reliable, locally accepted and economically repeatable operation. Favorable comparative crash evidence gives the company a meaningful foundation. 60 But recalls, emergency-response scrutiny and conditional overseas approvals show that safety validation and public legitimacy remain binding constraints. 60,61,63 Every marketed capability carries a corresponding duty of care; for Waymo, that duty will determine not only the speed of expansion, but the quality and durability of the value Alphabet can ultimately realize.