Waymo Robotaxi Expansion Hits 15 Cities as Tesla, Aurora Race to Scale

The autonomy industry moves from pilots to production

For most of the past decade, self-driving vehicles stayed in a holding pattern of small pilots, safety drivers and limited service areas. Coverage from the past month suggests that period is ending. Waymo is adding cities and vehicles fast. Tesla says one software release stands between its robotaxi fleet and round-the-clock service. Aurora is running driverless semitrucks commercially in Texas. Federal regulators, meanwhile, are drafting the first performance rules written for cars that have no human driver.

These developments are moving at different speeds and rest on very different levels of evidence. Taken together, the reporting shows that scale has replaced technical feasibility as the main question, and the regulatory system is still trying to catch up.

Waymo: filling in the map

Waymo is clearly in the lead. The Alphabet unit now runs robotaxi service in 15 U.S. cities. It reported more than 500,000 paid trips a week across 14 cities in early September, and its stated goal is 1 million weekly paid rides across 20 cities worldwide by the end of 2026.2 Two years ago it served only Phoenix, Los Angeles and San Francisco.3

The pace has picked up over the past several weeks. Denver, San Diego and Tampa got paid rides on September 1, which brought Waymo to 14 cities running fully driverless commercial service.7 Las Vegas followed on September 14.2 On October 14, Waymo dropped its invitation-only phase in San Antonio, San Diego and Tampa together, and it now offers driverless service at San Antonio International Airport.6

The fleet numbers matter more than the city count. State registration data show Waymo's Texas fleet growing 49% in roughly three weeks, to 1,102 vehicles as of September 24.13 About 80% of the roughly 4,000 Waymo robotaxis are in California and Texas.3 Much of the new capacity is the Ojai, a purpose-built minivan on the Zeekr RT platform that runs Waymo's sixth-generation driving system.8 MoffettNathanson tracks shipping records and estimates Waymo will import about 5,100 Ojai vehicles by year-end.1 Because the Ojai is built in China, tariffs are an added cost, and the reporting indicates Waymo is willing to absorb it.13

The financing has grown to match. Waymo has taken a $5 billion term loan from lenders led by PIMCO, Blackstone and Sixth Street. It is the company's first debt deal and follows a $16 billion equity round.2 Waymo is also preparing to go abroad. It plans a Tokyo launch with Nihon Kotsu and GO in 2027, a Munich service toward the end of 2027 and bookings in Singapore from 2028.9

The 1 million-ride target looks ambitious. Waymo has not published a much higher weekly figure since spring, so it is hard to tell how much the new cities are adding.2 Hitting the goal would mean roughly doubling April's volume within about eight months.2 The fleet buildup makes that plausible but not certain. Each new market needs months of public adoption before it carries real volume.

The reports also disagree on some details. One account says Waymo serves 15 cities and 500,000 weekly rides at once.3 Another puts the 500,000 figure across 14 cities, with the fifteenth city arriving in September.2 A secondhand summary of CNBC's reporting says the names of three new cities were not disclosed.4 Waymo's own announcements named Denver, San Diego and Tampa.7 The overall picture is the same in every account, but readers should treat precise city counts with some caution.

Zoox and the competitive crowding

Amazon's Zoox is moving at the same time. It is adding Houston, returning to San Diego and growing to 12 locations, but it still lacks regulatory clearance to charge fares in most of them.7 Its vehicle was designed from scratch with no steering wheel and seats facing each other, which puts it in direct conflict with federal standards written for conventional cars.4 One analysis argues that both companies are expanding faster than individual cities or states can write rules, and that they treat that speed as an advantage.7 That reading is speculative, but the timing supports it.

Tesla: everything rides on FSD v15

Tesla's autonomy story is mostly about software timing. Tesla AI vice president Ashok Elluswamy has said FSD v15 will reach the robotaxi fleet in October and allow 24/7 commercial operation.31 The model reportedly has about 10 billion parameters, ten times the current version, and Tesla describes it as the key to expanding Robotaxi into many more markets.34 The company has been preparing launches in Las Vegas and Phoenix.34

Public signals point to a slower pace. The latest change to Robotaxi hours moved closing time from 10 p.m. to 11 p.m.31 Elon Musk said the main remaining nighttime problem is spotting pets that are hard to see in the dark, "grey kittens on grey tarmac."33 He has also rejected adding thermal cameras.33 One outlet pointed out that a one-hour extension does not obviously lead to 24/7 service by the end of the month.31 That skepticism looks justified.

The sources also disagree about where v15 stands today. One reference says Tesla Robotaxi vehicles were already running FSD v15 as of August.36 Other coverage describes a fleet-wide v15 rollout as still upcoming in October, with only some vehicles testing early builds.31 The most likely explanation is a partial deployment, but this is exactly the kind of gap investors should watch.

For owners, the newest release is FSD (Supervised) v14.3.12. It improves recovery from temporary system degradations, tightens driver-monitoring gaze tracking and uses a single model for Actually Smart Summon, FSD and Robotaxi.37 Older HW3 cars are receiving "Lite" versions with robotaxi-style arrival options.32 Robotaxi service areas remain small next to Waymo's. Austin covers 246 square miles, and Tampa covers about 23.6.34

Tesla also faces regulators. Two days after Waymo's September 1 launches, NHTSA opened an audit query, AQ26002, into the Cybercab during the week it started commercial passenger rides in Austin.7 Earlier in 2026, NHTSA said FSD fails to detect hazards or warn drivers about degraded camera performance in low visibility until just before a crash.14 Nighttime and low-visibility performance is the same weakness Musk is now publicly working on, and that overlap is significant.

Autonomous trucking: Aurora's commercial claim

Aurora is putting driverless trucks into daily service. It has 20 trucks with no one in the cab running Texas routes such as I-45 between Houston and Dallas and I-20 toward El Paso. It plans to reach 200 by the end of 2026.21 At its investor day, the company projected more than 1,000 trucks by the end of 2027 and more than 30,000 by 2030, with revenue above $5 billion and gross margins above 60%.23 Starting in 2027 it plans to shift to a "Driver-as-a-Service" model in which carriers own the trucks. It would keep its own fleet at about 500 vehicles.23

The financial case is strong on paper. Bank of America estimates Aurora's cost at about $0.85 per mile, compared with about $1.30 per mile for driver wages and benefits.21 One customer, Detmar Logistics, says its Permian Basin trucks went from two to 2.5 loads a day to five or six.25 Volvo expects to begin driverless operations with the Aurora Driver in the first quarter of 2027.25

The gap between plans and current results is very large. Aurora reported only $5 million in revenue over the trailing twelve months.28 It lost more than $900 million last year and does not expect positive free cash flow until 2028.30 Its driverless mileage figures also vary by source, from more than 500,000 miles22 to more than 250,000 miles on Texas interstates30. The difference probably reflects different reporting dates. Aurora's hardware schedule depends on partners, and third-generation kits built with AUMOVIO are not due for full production until the end of 2027.27 Aurora is clearly ahead in trucking. The question is whether it can multiply its fleet a thousandfold in four years. Kodiak Robotics, which went public this month and runs driverless trucks in the Permian Basin, is the nearest competitor.30

Regulation: a federal framework built on exemptions

In Washington, NHTSA is moving toward a single national rulebook. In July it proposed a temporary exemption that would let Zoox deploy up to 2,500 vehicles a year for two years. It also announced work with SAE on the first AV performance standards and changed its Part 555 exemption rules.11 The SAE consortium, called ASCEND, was publicly launched in September. Its first priority is setting a competency baseline for automated driving systems that could later become a federal motor vehicle safety standard.13 The agency says current standards do not fit vehicles without steering wheels or pedals, such as the Cybercab and Zoox's robotaxi.13

The opposition is organized. The Transportation Trades Department, a federation of transportation unions, has asked NHTSA to withdraw its interim final rule on exemptions. The group argues that the rule could allow exemptions without proper review and that it skipped the notice-and-comment process. It wants enforceable standards in place before deployment, not regulation by exemption.17

The current system remains patchy. No comprehensive federal law authorizes or bans driverless cars. At least 26 states authorize driverless operation under certain conditions, and at least 13 have no law on it.12 Texas preempts city-level AV rules, which helps explain why Waymo, Tesla and Aurora are all concentrating there.16

The takeaway

The three main players are at very different stages. Waymo is a funded, fast-growing commercial service. Its main risk is whether riders keep pace with the vehicles, not whether the technology works. Tesla has bold plans that rest on a software release still being tuned for night driving. Aurora has a credible lead in trucking but almost no revenue so far. Regulators are trying to set standards while deployment is already underway, which means the rules will be shaped by an industry that is already operating at scale. If ASCEND turns real-world driving data into binding standards before a serious incident occurs, the industry will have its national framework. If an incident comes first, the expansion seen this fall could slow.