Signals Inbox·July 28, 2026·Autonomous Systems

Waymo vs Tesla: who will win autonomous driving?

Waymo is winning autonomous driving today because it already operates the larger, safer and more repeatable driverless service. Tesla has the cheaper architecture and the bigger distribution opportunity, but much of that advantage still depends on autonomy it has not yet proved at scale.

Q1Why does everyone keep comparing Waymo and Tesla?

People compare Waymo and Tesla because both want to own the autonomous ride, but they are taking opposite routes to get there.

Waymo treats driverless transport as a managed city service. It validates a defined operating area, equips vehicles with cameras, radar and lidar, runs fleet depots, works with local authorities and gradually opens the service to riders. In Austin and Atlanta, Uber handles the customer-facing ride-hailing layer. In most other markets, riders book through Waymo’s own app.

Tesla is trying to build a general driving system that can spread through manufacturing and software. Its current Robotaxi fleet uses Model Y vehicles, while Cybercab is meant to become the purpose-built vehicle. The larger Tesla plan goes further: customer-owned cars could eventually join the network and earn money when their owners are away.

The overlap has become concrete. Both companies now offer rides in Austin, Dallas, Houston and Miami. A rider in those cities can compare two different visions of autonomy rather than two distant prototypes. Waymo brings a controlled fleet and a mature driverless service. Tesla brings cheaper hardware, its own manufacturing system and the possibility of much wider distribution.

Q2Why is it so hard to say who is winning autonomous driving?

The argument gets messy because people keep mixing the robotaxi service that exists now with the network Tesla hopes to build later.

Waymo leads the first one. It operates a Level 4 service where the vehicle handles the driving task inside its approved operating area. The passenger can sit in the back and ignore the road.

Tesla has two separate products. Robotaxi is the small autonomous service. FSD (Supervised) is the much larger consumer product, and Tesla’s own wording still requires active driver supervision. The billions of miles collected through FSD can help train future systems, but those miles include a human fallback. They cannot be treated like rider-only miles.

There are three separate races. First comes what is already working: rides, autonomous miles, cities and safety. The second is economics: vehicle cost, fleet utilization, fares and the expense of launching a new market. Third comes future reach: factories, cars already on the road, software distribution and regulatory approval.

Waymo dominates the first race. Tesla has the more powerful setup for the third. The economics remain partly hidden because neither company tells us what an individual robotaxi ride earns or costs.

Q3Who is actually running a real robotaxi business now?

On real robotaxi scale, Waymo is miles ahead. The gap is closer to two orders of magnitude than a close contest.

Alphabet’s latest earnings call said Waymo had passed 500,000 fully autonomous rides per week. That pace equals roughly 26 million rides a year. Waymo completed 15 million rides during all of 2025, so its latest weekly run rate sits about 73% above last year’s average.

Tesla discloses far less. Its first-quarter presentation showed cumulative paid Robotaxi mileage reaching roughly 1.7 million miles by the end of March, after nearly doubling during the quarter. The company withheld weekly rides, active vehicles, paid hours, utilization and revenue. Those missing numbers make it difficult to judge whether Tesla has a dense transportation network or several small fleets generating rapid percentage growth.

The mileage definitions also differ. Waymo reports every rider-only mile, including mileage without a passenger, while Tesla’s chart covers paid Robotaxi mileage. Even with that caveat, Waymo’s disclosed autonomous mileage was about 130 times larger. Waymo was also driving more than four million autonomous miles each week, over twice Tesla’s entire cumulative paid total at the end of March.

Waymo and Tesla robotaxi operations compared

Q4Who is growing faster right now?

Waymo has the better growth story right now because it is adding large volumes from an already substantial base.

Its weekly ride count climbed from roughly 150,000 at the end of 2024 to more than 250,000 in the spring of 2025, then above 400,000 around the end of that year. The newly disclosed run rate is another 25% higher. The curve has repeated across several reporting periods rather than appearing in one large launch followed by silence.

Tesla’s paid Robotaxi mileage nearly doubled quarter over quarter. The pace is fast, and the service has since added new markets. The base was still small enough that a few dozen extra vehicles or higher utilization could create a dramatic percentage increase. Tesla has also avoided publishing the trip count behind the mileage curve, removing the cleanest way to separate genuine rider adoption from longer average journeys.

Tesla may post larger percentage increases for a while. The absolute additions still favor Waymo. A 25% increase on Waymo’s recent base represents millions of extra annualized rides. Tesla has yet to show anything comparable.

Q5Is Waymo expanding faster than Tesla?

City expansion is another clear Waymo win, and the launch pattern increasingly looks repeatable.

Waymo opened Dallas, Houston, San Antonio and Orlando to public riders in one coordinated launch, then added Nashville. Alphabet said that brought the service to 11 major US metropolitan areas and six new cities during the year. Waymo’s latest city update also confirmed that Denver, Las Vegas, San Diego and Tampa are progressing toward public service, although they remain upcoming markets rather than open ones.

Tesla’s public Robotaxi page now lists four active metropolitan areas. The company deserves credit for moving Austin beyond its initial launch and starting unsupervised rides in Dallas and Houston. Miami has also joined the network. Still, Tesla’s geographic rollout remains much narrower, and its Bay Area operation used a safety driver in the latest detailed quarterly disclosure.

Waymo can now open several cities while preparing the next group in parallel. Tesla has moved beyond a single pilot, but it has yet to repeat that pattern.

Q6Which autonomous-driving technology is working better today?

The real-world verdict currently favors Waymo. It has removed the driver across far more miles, rides and cities.

Waymo combines cameras, lidar and radar with mapping, simulation and remote assistance. The extra hardware raises vehicle cost and makes integration harder, but it gives the system several independent ways to understand distance, movement and road geometry. Its sixth-generation Driver is designed to lower that hardware cost while preserving sensor redundancy.

The new Ojai vehicle has already started carrying selected riders with the sixth-generation system. Waymo says its Mesa factory is scaling toward capacity for tens of thousands of vehicles a year, and the new platform is intended to handle snowier markets. The sequence connects cheaper hardware and a purpose-built cabin to an actual manufacturing plan and public deployments.

Tesla’s camera-led approach is more elegant economically. Its FSD page currently shows more than 12 billion supervised miles, and the company says its fleet contributes unusual real-world driving examples at enormous scale. The first-quarter update also described FSD v14.3, a revised vision encoder, reinforcement-learning changes and up to 20% lower inference latency. Tesla has completed the design of its AI5 inference chip and started pilot production of Cybercab.

Those are serious technical assets. But autonomous driving is ultimately judged by how often the human can disappear. Tesla’s huge supervised dataset has produced a much smaller unsupervised service, while Waymo’s heavier sensor stack has already spread to rider-only operations across several cities.

Q7Who has the stronger safety case now?

Waymo leads on safety evidence by a wide margin, even after several recent operational failures that deserve real scrutiny.

Waymo’s newest safety update covers 220.6 million rider-only miles in five operating areas. Against human drivers in comparable places and periods, the company reported 94% fewer crashes causing serious or fatal injuries, 82% fewer airbag-deployment crashes and 82% fewer crashes involving a reported injury. It also reported 93% fewer injury crashes involving pedestrians. The analysis includes collisions regardless of fault and adjusts the human benchmark to the areas where Waymo drives.

The scale now supports conclusions that were impossible a few years ago. Waymo estimates that its current mileage prevents one serious-injury-or-worse crash about every eight days compared with matched human driving. The latest dataset also includes statistically significant results from Atlanta, where the road layout and traffic mix differ from Waymo’s older markets.

Crash safety and day-to-day reliability now need to be judged separately. NHTSA has investigated Waymo’s behavior around stopped school buses and a collision involving a child near an elementary school. A separate recall covered 3,871 fifth-generation systems after vehicles could enter closed freeway construction zones. Another recall addressed flooded-road behavior. More recently, vehicles caused disruption during crowded Fourth of July traffic in San Francisco, and Waymo paused part of its service during a later power outage.

Those incidents weaken the claim that Waymo has mastered every urban condition. The crash-rate evidence still stands. The incidents expose another weakness: a fleet can drive safely on average and still struggle during construction, floods, blackouts, emergency scenes or mass crowd movements. Regulators are increasingly focused on how the whole fleet behaves when a city is under stress.

Tesla publishes a “seven times safer” claim for FSD (Supervised), but the comparison uses driving with an attentive human fallback. Tesla’s unsupervised Robotaxi fleet remains too small and too lightly disclosed for a comparable crash analysis. NHTSA is also examining FSD behavior around traffic signals and degraded road conditions.

At present, Waymo’s evidence is much deeper. Its recent recalls also show where the next safety battle has moved.

Q8Who is closer to making robotaxis economically work?

The economics are closer. Waymo has proved more, while Tesla offers the cheaper ride and the more attractive cost structure on paper.

Neither company publishes contribution margin per ride, fleet utilization, depot cost or autonomous revenue. We therefore have to use indirect evidence: rider prices, wait times, operating mode, vehicle production and the density of the service.

Obi’s large San Francisco Bay Area pricing study initially found an average Tesla fare of $8.17, compared with $19.69 for Waymo. Tesla’s average wait was 15.32 minutes, versus 5.74 minutes for Waymo. The long Tesla wait suggested limited supply, and the rides still carried a human safety monitor in California.

A later Obi study covering early 2026 found that Tesla’s average prices rose 41%, much faster than Waymo’s 18% increase, although Tesla remained the cheapest provider. This looks like a service moving away from launch pricing. It still tells us little about profit because Tesla has never disclosed the full cost of those trips.

Waymo’s fares have been higher, but the earlier Obi work found its premium over Uber had narrowed to 12.7%. For medium-distance journeys, the gap was around 2%. Waymo achieved those prices with the driver already removed and with wait times close to mainstream ride-hailing.

Waymo has shown that customers will repeatedly pay for a driverless ride at prices approaching Uber. Tesla could eventually win on vehicle and sensor cost, but cheap launch fares are not proof of a cheaper complete service.

Robotaxi economic indicators

Q9Who has the stronger distribution?

Distribution is Tesla’s best category. Its potential reach dwarfs Waymo’s, even though Waymo already delivers far more autonomous rides.

Tesla delivered more than 480,000 vehicles in the most recent quarter and had reached 9.2 million cumulative deliveries by the end of the previous one. It also reported 1.28 million active FSD subscriptions, 51% more than a year earlier. Factories, service centres, charging infrastructure, a global app and millions of connected cars give Tesla an unusually direct route to customers.

That installed base could flip the race very quickly. A software system that works safely on existing vehicles can spread much faster than a fleet requiring new sensor integration, local depots and city-by-city vehicle deployment. Tesla also controls the vehicle design, AI computer, manufacturing process and customer relationship.

Waymo’s distribution is smaller but fully active. It reaches riders through its own app in most cities and through Uber in Austin and Atlanta. The Uber partnership gives Waymo access to existing demand without forcing it to win every rider directly. Waymo can also place its Driver on vehicles from different manufacturers, which reduces dependence on one car platform.

Tesla has the larger route to future scale, although those millions of cars still form a supervised fleet. Waymo’s narrower network already sells the autonomous outcome.

Q10Can Tesla really turn customer-owned cars into robotaxis?

Today, Tesla’s customer-owned Robotaxi network is still a promise. The consumer fleet requires supervision, and the operating model remains undefined.

The idea is powerful. Owners could make their cars available when they are at work or asleep, Tesla could expand supply without buying every vehicle, and cities with a large installed base could gain Robotaxi coverage quickly. No rival has an equivalent pool of connected consumer vehicles.

The practical gaps are large. Tesla has yet to specify which hardware generations will qualify, how vehicles of different ages and maintenance histories will be certified, who carries insurance, how owners and Tesla split revenue, or who cleans and charges each car between commercial rides. Camera obstruction, tyre condition, interior damage and inconsistent servicing all become network problems once private cars carry strangers without their owners present.

The software gap is more basic still. FSD (Supervised) tells the driver to remain attentive. Tesla’s current autonomous service uses company-controlled vehicles in restricted areas, giving the company control over hardware, maintenance, software versions and operating boundaries. Moving that capability into ordinary customer cars across broad areas would be a much harder deployment.

For now, the installed fleet is a valuable option. It becomes a winning network only after Tesla shows unsupervised customer vehicles completing paid trips safely and repeatedly.

Q11Which advantage will be hardest for competitors to copy?

Waymo currently owns the harder-to-copy advantage in autonomous ride-hailing: safety evidence, city operations and deployment experience all compound together.

A competitor can buy lidar, cameras and radar. Reproducing more than 200 million rider-only miles, city approvals, remote-assistance workflows, depot operations, crash analysis and thousands of edge cases takes far longer. Each new city creates additional real-world operating data, and every large mileage increase strengthens Waymo’s safety case with regulators.

Waymo has also learned the unglamorous work around autonomy: recovering stalled vehicles, managing charging, choosing pickup points, handling airport access and coordinating with emergency services. Recent failures show that this capability is incomplete, but they also expose the work that a new entrant must eventually face.

Tesla’s moat comes from a different place. It owns high-volume vehicle manufacturing, a large paid software base, custom inference chips, training infrastructure and a global stream of camera data. A standalone robotaxi startup would struggle to reproduce that stack. If Tesla reaches broad unsupervised autonomy, its moat could become larger than Waymo’s because software and manufacturing would reinforce each other.

Waymo’s advantage already protects a functioning Level 4 service. Tesla’s could eventually support a cheaper and wider system. Today, Waymo owns the stronger moat.

Q12Who can keep funding the autonomous-driving race?

Funding looks abundant on both sides. Tesla has the stronger ability to fund itself, while Waymo has more money set aside specifically for autonomy.

Tesla ended the first quarter with $44.7 billion in cash, cash equivalents and short-term investments. It generated $3.9 billion in operating cash flow and $1.4 billion in free cash flow during that quarter. Robotaxi can also reuse Tesla factories, service locations, vehicles and charging infrastructure.

Waymo raised $16 billion at a $126 billion post-money valuation earlier in the year. Alphabet remained the majority investor, while the round brought in Sequoia, DST Global, Dragoneer, Andreessen Horowitz, Mubadala and other large institutions. That gives Waymo a sizeable pool of capital for vehicles, factories, depots and market launches.

The money comes with different headaches. Waymo must spend heavily before each city reaches density, and Alphabet still withholds Waymo’s standalone revenue and losses. Tesla can finance autonomy through a broader profitable company, but Robotaxi competes internally with vehicle factories, batteries, AI compute, Cybercab, Semi and Optimus.

A cash shortage looks unlikely for either side. Execution will decide this race well before access to capital does.

Q13Is Waymo versus Tesla even the right global comparison?

Apollo Go is Waymo’s closest operating rival globally, while Tesla is chasing the largest future upside.

Baidu reported 3.2 million fully driverless Apollo Go rides in the first quarter. Weekly volume peaked above 350,000 in March, total rides grew more than 120% year over year, and cumulative public rides passed 22 million in April. Waymo’s latest weekly rate was roughly 1.4 times Apollo Go’s peak, which makes Baidu a much closer volume rival than Tesla.

Apollo Go is also expanding outside China through testing and partnerships in the Middle East and Europe. Its rise suggests that autonomous ride-hailing could develop through several regional leaders rather than one global winner.

Tesla remains central to the debate because it could alter the market’s cost and distribution structure. If Cybercab enters high-volume production and the same autonomy reaches customer vehicles, Tesla could leap over years of fleet-by-fleet expansion. Until that happens, its public profile is far larger than its disclosed driverless service.

Q14Who is winning Waymo versus Tesla right now?

Waymo is winning autonomous driving by a clear margin. The lead comes from real service rather than branding or theoretical scale.

Three things settle the verdict. Waymo carries far more rider-only passengers, its safety evidence covers a vastly larger unsupervised sample, and it has opened enough cities to show that its deployment process can repeat. The service already handles around half a million fully autonomous trips each week, while Tesla still withholds basic operating figures such as weekly rides and active fleet size.

Tesla leads in manufacturing reach, installed vehicles and likely hardware cost. Those strengths could become decisive later. They carry less weight today because the central technical question remains open: can Tesla remove supervision across a large fleet while matching the safety and reliability expected from a paid transport service?

Waymo’s next challenge is commercial. It must turn technical leadership into an attractive business. The company is targeting more than one million weekly rides by year-end, bringing its sixth-generation vehicles into production and entering many more cities. Recent recalls, freeway restrictions and disruption during blackouts or major events show where that scaling plan can break.

Tesla can change the answer by publishing transparent unsupervised safety and usage data, expanding driverless service well beyond its four current metros, moving Cybercab from pilot to volume production and allowing ordinary customer vehicles to complete paid rides without supervision. Waymo can strengthen its lead by reaching its ride target, lowering vehicle and operating costs, and proving that its network remains dependable during construction, floods, outages and emergencies.

Our final call is firm: Waymo should win the first large phase of autonomous ride-hailing. Tesla still owns the more explosive path to a much bigger network, but that path depends on a breakthrough it has spent years promising and only recently begun to demonstrate in a small commercial fleet.

Waymo versus Tesla: leader by category