Waymo Robotaxi Expansion Eyes 20,000 Vehicles by End of 2027

A fivefold fleet plan, reported but not confirmed

Waymo is reportedly planning to grow its robotaxi fleet about fivefold in roughly 14 months. Semafor first reported the plan on October 19, citing people familiar with it. According to that report, the Alphabet-owned company wants to go from about 4,000 vehicles today to 8,000 by the end of 2026 and 20,000 by the end of 202795. If the interim target holds, Waymo would double its fleet in less than three months21.

Outlets do not all describe the plan the same way, and the difference matters. Most coverage says Waymo declined to comment to Semafor and has taken no official position on fleet targets21327. One financial outlet nonetheless wrote that Waymo "announced" the expansion6, and another said Waymo itself told Semafor about the plan25. The more careful accounts are the ones to trust here. The 8,000 and 20,000 figures are targets from anonymous sourcing, not commitments the company has made publicly. One analysis also notes that the second step is the harder one: going from 8,000 to 20,000 means adding about 12,000 vehicles in a single year, a 150% increase.

The financing, by contrast, is on the record. Earlier in October, Waymo closed a $5 billion loan, its first debt deal, from lenders including PIMCO, Blackstone and Sixth Street2127. That comes on top of a $16 billion equity round earlier this year that valued the company at about $126 billion9. One account says Alphabet did not guarantee the loan and that its terms allow Waymo to later bundle fleet leases into securities25. If that is right, it is a quiet but important change: lenders would be valuing robotaxi fleets on their own cash flows instead of leaning on Google's parent.

The groundwork is already visible

The 20,000 figure looks less like a moonshot when set against Waymo's past two years. In September 2024 it ran paid service in only three cities: Phoenix, Los Angeles and San Francisco. It now operates in 15 U.S. metros, and most of those launches came in the past year1. Denver, San Diego and Tampa opened to paying riders on September 1, and Las Vegas followed on September 144. Weekly paid rides are above 500,000, against a company goal of 1 million by the end of 20264. In May 2024 the figure was 50,0007.

The fleet itself is very concentrated. About 80% of Waymo's roughly 4,000 vehicles are in California and Texas1. Texas is growing fastest: registrations there rose 49% in three weeks, to 1,102 vehicles as of September 2417. Registration counts do overstate how many cars are carrying riders at any moment, since some vehicles are being prepared or serviced7.

Most of the growth comes from one vehicle, the Ojai. It is a Zeekr minivan, from a brand owned by China's Geely, that Waymo fits with its own self-driving system at a plant in Mesa, Arizona21. It replaces the Jaguar I-Pace, which Jaguar stopped building in 202421. The research firm MoffettNathanson, which tracks shipping records, estimates that Waymo will import about 5,100 Ojai vans by the end of this year1. That supply alone would account for most of the jump to 8,000. Waymo describes the Mesa plant as able to support tens of thousands of vehicles a year7, and Hyundai is expected to build Ioniq 5 robotaxis for Waymo in the U.S. as well23. In other words, the factory capacity and import pipeline that a 20,000-vehicle fleet would need are already being put in place.

Regulation: Waymo's own rulebook is less settled than its rivals'

On regulation, Waymo's position is unusual. Most of the federal attention to robotaxis this autumn has gone to vehicles built without steering wheels or pedals. Zoox received NHTSA's first commercial exemption for a purpose-built robotaxi in July. It allows up to 2,500 vehicles a year for two years15. Tesla took a different route and self-certified its Cybercab as compliant with federal safety standards. NHTSA opened an audit query the day the Cybercab entered paid service in Austin, then issued a Special Order demanding sworn answers1511. Tesla now also faces a defect petition alleging that about 1,000 of its robotaxis speed through school zones16.

Waymo's Jaguars and Ojai vans are adapted production vehicles, so it has largely avoided this fight over whether driverless vehicles meet design standards16. That is a real structural advantage for growing quickly. An exemption-based path like Zoox's comes with a hard cap. Waymo's fleet size is limited mainly by its own capital and operations.

That does not mean Waymo is free of regulators. In July, NHTSA Administrator Jonathan Morrison ordered autonomous-vehicle developers to submit plans by the end of that month for fixing how robotaxis behave around emergency scenes. He called the problem a "functional insufficiency." The directive was widely read as aimed at Waymo, after several of its cars got stranded in San Francisco on July 415. No submissions or results have been made public15. A House bill introduced in August would let officials geofence autonomous vehicles during emergencies, though it has not moved since15. Flooding has been a recurring problem too. Waymo paused service in five cities in May after its cars kept driving into flooded roads21, and San Antonio service resumed only on September 17 after an April pause4.

State and local rules are the other limit. Nevada's August approval allowed Waymo up to 1,000 vehicles in the Las Vegas area, while Tesla received a ceiling of 5,0004. California regulators approved an updated safety plan in August covering a larger territory and the Ojai platform4. One law firm calls the U.S. system a fragmented patchwork with no comprehensive federal oversight17. The SELF DRIVE Act, which would create a national framework, cleared a subcommittee in February on a 12–11 vote and is still pending15. A fleet of 20,000 vehicles will have to be assembled one jurisdiction at a time.

The China problem sits under the whole plan

The biggest regulatory risk to the target may come from trade and security policy, not traffic safety. The Commerce Department bars Chinese-made connected-vehicle and self-driving software starting with model year 2027. Waymo says it installs its own sensors and software in Arizona and shares no rider data with carmakers27. One report notes the Ojai is imported without Chinese connected-car software because of federal restrictions. Still, at least one assessment points to a bill in Congress that could force the Zeekr-based vans off U.S. roads by 203027. Tariffs are already raising Waymo's costs, and the company appears willing to absorb them1.

My reading is that Waymo has built its growth plan on a supply chain Washington may decide to cut. If the 2027 target is reached mostly with Chinese-built vans, the company will have more political exposure than any rival. That makes the Hyundai supply line more than a backup.

Safety data will face its toughest test

Waymo's argument for scaling up rests on its safety record. Over 271.3 million rider-only miles, the company reports 0.01 serious-injury crashes per million miles, compared with 0.21 for its human-driver benchmark. For injury crashes of any kind, the rates are 0.67 and 3.7727. An IIHS review of federal crash reports reached a similar conclusion. Its president said the vehicles are safer than human drivers "on a limited scale"27. That qualifier is the key point. A fivefold larger fleet, new vehicles and new cities will be tracked through a reporting system that IIHS itself calls inadequate at that scale27. NHTSA warns that its crash data are not adjusted for fleet size or miles driven14.

What the gap with rivals really looks like

The competitive gap is large, though reports measure it differently. One account cites Texas state records showing only 42 Tesla robotaxis against Waymo's hundreds of vehicles there21. CNBC reported that Tesla's Texas robotaxi fleet includes 420 Model Y vehicles plus Cybercabs in Austin20. The difference probably comes down to what each source counts, such as driverless registrations versus all vehicles in the service. Either way, Waymo has the larger driverless operation. Tesla also still lacks California permits for driverless service20.

Overseas, Waymo will begin testing around Madrid with the aim of carrying riders there in 2027. WeRide and Uber already hold Spain's first driverless permit21. Waymo has also set targets of 2027 for Tokyo and 2028 for Singapore4.

The bottom line

The 20,000 figure is not officially confirmed, but the money, factory capacity and import pipeline behind it are real. Raising capital and building cars no longer looks like Waymo's hardest problem. The harder parts are proving its safety record holds at five times the scale, dealing with dozens of local regulators, and keeping a China-dependent vehicle supply intact. The year-end count against the 8,000 target will be the first real test.