The material establishes that Waymo has moved from an Alphabet-funded experiment into a scaled commercial ride-hailing business 8,9,11,13,20,21,22,23,35,44,63,77,81, operating through Waymo One 77. It also establishes that the operating proof is outrunning the financial and safety proof. Waymo now reports roughly 500,000 paid rides per week 1,3,4,14,16,17,18,19,20,26,28,29,30,31,35,36,37,40,41,45,47,49,50,52,59,64,66,78,79,80, a figure echoed as of September 24, 2026 52, logged 15 million rides in 2025 15,25,56,83, and has driven over 200 million fully autonomous miles 2,21,36,42,44,56,83. That scale is real, but it is concentrated: approximately 80% of the fleet sits in California and Texas 52,61,66, and the fastest recent growth has come from the Ojai, a modified Zeekr RT minivan built on Waymo's sixth-generation self-driving system 52,66. The strategic question for Alphabet is no longer whether driverless service can be delivered; it is whether a geographically concentrated, capital-intensive operation can turn those miles and rides into a durable business before competitors and regulators compress the window.
Scale is visible, but its foundation is narrow
Waymo operated commercial service in Phoenix, Los Angeles and San Francisco in September 2024 52 and had expanded to 15 U.S. cities by September 24, 2026 52, with most city launches occurring in the year preceding reporting 52. Fleet size is reported above 4,000 vehicles 32,34,35,38,39,44,47,59,64,85, while a separate estimate puts operations at approximately 3,000 robotaxis 10,12,80. In Texas, registered Waymo vehicles rose to 1,102 by September 24, 2026 52,66,80, following a 49% increase in the preceding three weeks 52. The Ojai accounted for about one-third of that Texas fleet 52, and Waymo has said it is scaling production capacity toward tens of thousands of units annually 57. Base vehicles are imported and outfitted in Arizona without Chinese connected-car technology 52, while U.S. tariffs on Chinese-made vehicles are increasing Waymo's costs for each imported Ojai 59,66. A shipping analysis cited by the article projects about 5,100 Ojai imports into the United States by year-end 64. The supply path has a second rail: Hyundai plans to begin mass-producing IONIQ 5 robotaxis for Waymo at its Georgia plant as early as October 69, with commenters estimating Driver compute and sensors at under $20,000 69.
The expansion path therefore depends on a narrow geographic and supply-chain base. The material itself does not assess the consequences of that fleet concentration in California and Texas 64, but the risk is visible in the import dependence and tariff exposure.
Safety is the strongest asset and the least settled question
The most widely reported safety claim is that Waymo's fully autonomous vehicles were involved in 82% fewer injury-causing crashes than human drivers over more than 270 million miles 85, with a separate statement of 841 fewer injury-causing crashes than human drivers would have had over the same distance 85 and a 90% reduction in serious injury-causing crashes 40,83,85. An independent Insurance Institute for Highway Safety analysis found 68% fewer police-reportable crashes 85. The aggregate reduction varied by city, from 66% in Los Angeles to 90% in San Francisco 85, with Phoenix at 71% 85 and Austin at 84% 85.
But the edge cases are accumulating at the same time. Waymo reported 94 new crashes in September 2026 53, with one fatality, three serious injuries and additional moderate and minor injuries associated with those reports 53. It has filed six recalls with NHTSA since February 2024 85, including recalls after an empty robotaxi drove into floodwater 85 and after vehicles drove into active freeway construction zones 85; all six recalls were fixed through software 85. The typical pre-crash speed in the September 2026 reports was 7.56 mph 53, which suggests many events are low-speed interaction failures rather than high-energy collisions. The same corpus includes doorings, barrier-arm contacts and bollard strikes 53, alongside a fatal Dallas incident in which a pedestrian was hit by an SUV and thrown into an empty Waymo, with police not finding Waymo at fault 85. The safety dataset covers Waymo's longest-running markets 85 and does not include Denver, San Diego or Tampa, where paid service started after the period covered by the dataset 85.
The pattern is the point: the aggregate safety lead is genuine, but the residual failure modes are numerous, visible and local, and they will be litigated market by market as much as through headline crash-rate comparisons.
The financial pivot has not yet happened
Waymo's funding and valuation are those of a large option, not a mature business. It raised $16 billion at a $126 billion valuation in February 2026 15,25,40,42,43,47,48,75,76, and its revenue run-rate is reported at $350 million 75, implying a P/S ratio of 360.0 75. Earlier estimates valued Waymo near $200 billion 83, while another excerpt reports an implied valuation of $152.72 billion 75. Sundar Pichai said Waymo could contribute meaningfully to Alphabet's revenue as soon as next year 65,84, and Alphabet is described as transitioning away from heavy reliance on advertising over time as Cloud and Waymo gain share 65.
The counterweight is just as explicit. Waymo is burning cash while expanding 47,83, there is no evidence in the material that it is currently profitable 66, and the Ojai is intended to reduce costs and help reach profitability rather than a sign that profitability has arrived 52. Vehicle cost, durability, rider experience and converting fleet expansion into efficiency are all identified as core challenges 64. The financing discussion also names lenders including PIMCO, Blackstone and Sixth Street Partners 46,47,75. On the available figures, Alphabet is funding a capital-intensive growth option with credible scale and aspirational multiples 84, not a near-term earnings driver, although the funding capacity itself is real 70.
Competition is shifting from proving autonomy to owning demand
The operational lead is no longer uncontested. Tesla, Uber and Waymo have all received permits to operate commercial robotaxi services in Nevada 33,75, and Tesla's unsupervised Robotaxi service reportedly expanded to Austin, Dallas and Houston in Texas 73,74 and to Miami, Orlando and Tampa in Florida 73,74, while Bay Area operations remain subject to a safety-supervisor requirement 73. Apollo Go operated in 28 cities and had provided 23 million cumulative rides as of June 82, and Zoox, an Amazon subsidiary, opened its first robotaxi serial production facility 5,6,7,72 and began commercial operations in Las Vegas before Waymo's launch 57. Fully driverless robotaxis remained a small but rapidly growing share of ride-hailing in a limited number of cities 73.
The deeper competition is over demand. Uber is positioning its network of human drivers as a strategic core asset 54, schedules human drivers and self-driving vehicles through the same app 54, and could aggregate supply and route demand without choosing a single autonomous-driving provider 55. Waymo has used the same platforms selectively: it began commercial service in Austin through Uber in March 2025 52,66,80, ended its pilot partnership with Uber in Phoenix 24,27,75, and reported a Lyft integration with rides available through the Lyft app in Nashville 75. Grab is doing the same abroad, having invested in or partnered with six autonomous-vehicle businesses over roughly 18 months 71. If vehicle autonomy becomes commoditized, the winner may be the demand aggregator rather than the fleet operator, and Waymo's direct distribution, pricing and rider experience will determine whether it retains customer ownership or becomes capacity inside someone else's network.
International expansion tests transferability, not just capability
Tokyo, Singapore and Munich show Waymo attempting to export a U.S. operating model into different roads, rules and labor systems. Waymo plans to launch a fully autonomous commercial taxi service in Tokyo in 2027 81, working with GO and Nihon Kotsu 81, and the service is explicitly positioned to coexist with human-driven taxis rather than simply replace them 81. Singapore is planned for 2028 51,68,71,76 as an all-electric service 60, with an initial fleet of all-electric Jaguar I-PACE vehicles planned 68. Munich testing is underway 62,75, with commercialization referenced for 2027 75. Before Singapore can accept passengers, Waymo must complete mapping, supervised driving and autonomous validation 76, obtain approval from the Land Transport Authority 67, and adapt to local road geometry and monsoon weather 68,76. A safety incident or poor performance could delay the planned 2028 launch 76. International expansion is therefore a test of whether operational experience transfers 81, and it carries regulatory risk that cannot be resolved by adding more miles in Phoenix.
What this means for Alphabet
For Alphabet, the discipline should be to treat Waymo as an infrastructure business rather than a technology demonstration. The wider physical-AI thesis is that artificial intelligence is moving from digital workflows into physical operations 58, and Waymo is the company's most visible operating test of that transition. The proof is in the performance, not the promise, and performance here means auditable safety, unit economics and integrated demand. The most valuable near-term signals will be whether the Ojai and Hyundai IONIQ 5 supply paths reduce per-vehicle cost, whether city-by-city safety data hold up under NHTSA and local scrutiny, and whether Waymo can lock in transit, taxi-operator and distribution partnerships while competitors build purpose-built fleets and aggregators consolidate demand. Certification, in this context, is not a box to be checked; it is the floor that determines how fast the network can scale without a derailment.