What will it take for robotaxis to go global?
On the streets of Palo Alto, Uber drivers are facing the reality of an autonomous future. “The Waymos are everywhere now and soon we’re going to be replaced,” says Erik, who started serving passengers around Silicon Valley in his red Tesla Model 3 during the Covid-19 pandemic.
With self-driving vehicles on his turf, the former baker is getting ready to pivot. Because his car’s driver-assist system can navigate the wide suburban streets almost entirely on its own, he intends to take out a loan to finance a small fleet of Teslas after Elon Musk promised that his customers will soon be able to send their cars out to work for them.
Fifteen years after Google first unveiled its self-driving car project and kick-started a race that has already swallowed tens of billions of dollars of investment, so-called robotaxis finally appear ready to be a part of everyday life.
Waymo — owned by Alphabet, Google’s parent company — is already operating more than 250,000 paid rides per week across five US cities, surpassing ride-hailing app Lyft’s market share in parts of San Francisco. Tesla launched a limited robotaxi service in Austin last month, while Amazon-owned Zoox is confident that it will go live in Las Vegas later this year.
While the US is moving ahead more rapidly in the autonomous space than other countries, scaling up those services beyond a handful of cities in the domestic market remains a challenge. Companies must not only prove that robotaxis are safe and reliable, but also need to secure tens of billions of dollars more in capital investment. Margins in the ride-hailing market are also notoriously thin and the costs of operating a robotaxi service at scale are largely untested.
Even the largest American automakers, GM and Ford — not to mention Apple, which killed off its car project last year — are rethinking their plans for self-driving cars. That leaves the future of the US robotaxi market in the hands of just a few of the richest, most determined tech giants — and a handful of start-ups bold enough to challenge them.
The Financial Times has interviewed more than a dozen current and former senior executives and employees at Tesla, Waymo and Zoox, as well as start-up founders and their backers, to explore how these companies are racing to scale operations and win over investors.
Advocates of autonomous vehicles have long argued that they are safer than cars driven by humans: they do not get distracted, text, or drink and drive — and they have a 360-degree field of vision.
But for the benefits to be realised, investors need to believe that the economics of robotaxis can deliver meaningful returns not just in Silicon Valley but for billions of people all over the world.
“There was a long period of time where autonomous vehicle progress was overhyped and the substance didn’t match up to the headlines. You could argue [today] the reverse is true: they are here [and] they are scaling,” says Andrew Macdonald, Uber’s president.
“[The debate] is transitioning from, ‘is this technology solvable?’ to ‘is this commercialisable?’”
Waymo launched its first paid-for public service in August 2023. At the time it was operating roughly 10,000 paid rides per week, according to regulatory filings. This year, it is on course to have carried 20mn customers.
Almost a decade after it was spun out of Google to be a standalone business under the Alphabet umbrella, Waymo’s head of business development and strategic partnerships, Nicole Gavel, says it is now in “a mode of growth”. “We’ve been focused on safety from the get-go and now, we’ve got the technology that scales, [and] we’ve got the business model that scales,” she says.
Waymo has pulled well ahead of its rivals by making its services publicly available to millions of people. However, its vehicles — which use bespoke sensors attached to Jaguar I-Pace SUVs — are costly and the company has never disclosed whether its robotaxi service is profitable. Waymo declined to comment, but many analysts assume that the total cost of its current vehicles is close to $150,000 each.
“In terms of unit economics, we believe the autonomous space in the US will remain largely unprofitable until scale increases and vehicle costs materially decline, likely to under $100k,” says Doug Anmuth, an analyst at JPMorgan.
Waymo is seeking to cut costs by refining its design — its upcoming sixth-generation “driver” uses half as many cameras — and through deals with South Korea’s Hyundai and Zeekr, which is owned by China’s Geely, for less expensive vehicles.
By the numbers: Waymo’s fleet
250,000
Waymo’s paid rides per week across five US cities
1,500
Number of Waymo vehicles currently on the road
$11bn
Amount of funding raised by Waymo
Gavel says that Waymo’s initial focus was on ensuring its technology worked safely. “Now we are investing in driving costs down as well,” she adds.
Waymo, which already has more than 1,500 vehicles on the road, plans to equip at least 2,000 cars before the end of next year. A new factory in Phoenix, which it operates in conjunction with automotive supplier Magna, will eventually be capable of equipping “tens of thousands” of vehicles a year.
While Waymo adapts existing vehicles, Amazon’s Zoox is producing its own bespoke robotaxis: toaster-shaped glass boxes with no steering wheel. The company opened a 220,000 square foot final assembly site in Hayward, California, last month. By the end of next year, Zoox hopes to produce roughly 5,000 robotaxis per annum, with the aim of 10,000 vehicles a year in the future. It has not disclosed the cost of its vehicles, but Zoox’s custom cars are more expensive than Waymo’s, according to a person familiar with the matter.
Tesla is taking a radically different approach, using its existing Model Y SUV for its limited robotaxi service in Texas. The company only uses cameras, eschewing the use of lidar and radar, significantly lowering costs. Its sensor suite is estimated at roughly $400, with plans to retail its so-called Cybercab — which Musk says will go into mass production at the end of next year — for less than $30,000.
Although Tesla is operating a fleet of its own vehicles to start, it plans for customer vehicles to be deployed on the service in an Airbnb-style sharing model. This could improve the economics of its robotaxi business, but first it must overcome scepticism about the safety and reliability of its self-driving system after its Austin debut included several hazardous incidents. This prompted questions from the National Highway Transit Safety Administration (NHTSA), the regulator that has also investigated Waymo and Zoox for incidents involving other road users.
Tesla’s so-called “Full Self Driving” is a level 2 system on the Society of Automotive Engineers’s (SAE) six-point scale for autonomous vehicles, three levels below full autonomy. It requires the driver to take over duties as requested and to intervene in the event that it cannot function properly in a certain environment. Zoox and Waymo deploy level 4 systems that can operate without intervention.
Musk told investors in April that the service would start to positively affect the electric-car maker’s bottom line around the second half of next year. “Once it does start to move the financial needle in a significant way, it will really go exponentially from there,” he said, predicting Tesla will have “a 99 per cent market share”.
But one former Tesla executive says that the company’s technology is still not safe enough to deploy on public roads without a safety driver or remote operator. They add that the technology, which relies solely on cameras, means that the vehicle is “driving through a dense fog” and this provides false assurances around safety. “You can’t outrun human complacency,” they add.
Wayve, a SoftBank-backed developer of autonomous driving systems based in London, is working closely with Uber to bring more Tesla-like robotaxis online in the coming years, including a UK trial starting in 2026. The start-up, which relies primarily on cameras but plans to use radar and lidar for its level-4 system, is working with automakers including Nissan to bring its technology to market.
Wayve is betting that a couple of thousand dollars’ worth of sensors integrated into regular, mass-produced vehicles will soon be capable of full autonomy, at a fraction of the cost of Waymo’s or Zoox’s custom cars.
Suranga Chandratillake, general partner at Balderton Capital, one of Wayve’s financial backers, argues that its “generalisable” self-driving technology reduces mapping and testing time needed before launching in a new city — a process he claims costs their rivals hundreds of millions of dollars.
“No one in the world has demonstrated a scalable robotaxi solution yet,” says Alex Kendall, Wayve’s co-founder and chief executive. “But it’s coming.”
The upfront cost of robotaxis is just one of several overheads for their operators, alongside maintenance, charging and fleet management. “We’re selling rides, not vehicles,” says Aicha Evans, Zoox’s chief executive.
Neel Mehta, former head of strategy at Zoox and a partner at venture capital firm G2, says that the focus of robotaxi operators is on driving down the marginal cost of operating a ride. “Naturally, what will win long term is the cost structure,” he adds.
Robotaxi operators are relying on their vehicles operating longer shifts than human drivers. Zoox is aiming for 16 hours on a single charge, whereas Uber caps human drivers at 12 hours per shift.
The companies are seeking to find efficiencies by increasing the number of vehicles per remote operator — a person stationed in a control centre who can help guide a vehicle when it encounters difficult scenarios such as a roadblock, or respond to queries from passengers in their cars — as they grow more confident in the technology’s ability to navigate complex environments and situations.
Uber’s Macdonald says that launching a robotaxi service at commercial scale in a city the size of London or New York would require “billions of dollars in capex”, plus operational costs.
But Walt Piecyk, an analyst at New York technology research firm LightShed Partners, estimates that a fleet of robotaxis could handle the same demand as Uber with as little as a third of the vehicles.
If hardware costs can be brought down to $50,000, he pegs the total cost of a US-wide fleet at $25bn. “These aren’t insurmountable numbers,” he says. Autonomous driving executives hope that such a business plan could lure major pension funds and real estate investors such as Blackstone to buy whole fleets of autonomous vehicles.
For now, the likes of Waymo have focused on launching services in densely populated urban centres, such as San Francisco, where short journey times in the city lead to a higher turnover rate among riders and reduces the number of vehicles that need to operate to achieve profitability.
“When you have a lower population density, there’s less demand and you have to spread out vehicles. This leads to worse economies of scale,” G2’s Mehta says.
Expensive equipment and fleet management costs have translated into a more expensive experience for consumers who are paying roughly $11 per kilometre in San Francisco using Waymo, but around $8 with Uber for the same distance, according to Obi, an app that gathers pricing data across multiple ride-hailing services.
The companies say that the experience justifies the price. Wayve’s Kendall argues that his technology will offer a safer service than human drivers and at a lower cost.
However, some experts of driverless cars doubt that predominantly camera-based systems can pass the safety test.
“Regardless of whether it might be theoretically possible to have a camera system that is sufficient there’s simply no evidence that Tesla has that,” says Bryant Walker Smith, an associate professor at the University of South Carolina who helped write the SAE standard.
Since taking office this year, US President Donald Trump has loosened rules surrounding crash reporting to exempt operators from certain duties so long as their vehicles are being used for non-commercial purposes. These changes have ultimately helped smooth the path for robotaxi deployment.
Musk, who was Trump’s largest financial backer during last year’s US election, oversaw federal lay-offs that included the dismissal of staff within the “office of vehicle automation safety” at the NHTSA, the FT has previously reported. However, after the two men clashed over Trump’s tax and spending legislation, the regulatory road ahead for Tesla is less clear.
Even then, Waymo’s Gavel says that state-by-state restrictions on autonomous vehicles are the industry’s biggest brake on growth: “We hope for acceleration around that as more states feel the benefits.”
With the economics and regulation of robotaxis still unclear, the financial endurance of even the world’s most valuable tech companies is being tested.
Waymo, which has raised more than $11bn including from outside investors such as Andreessen Horowitz and Fidelity, is experimenting with different business models to defray the costs. While it operates its Waymo One ride-hailing service alone in San Francisco and Los Angeles, it partners with Uber in Austin, Atlanta and Phoenix. As well as supplying passengers, Uber is helping pick up the tab in these cities for the cost of the vehicles and their management including cleaning, repair and depot operations.
Zoox is planning to go direct to consumers, though it has held discussions with Uber and Lyft, the FT previously reported. Evans says that she expects its app will coexist with ride-hailing platforms for the foreseeable future. Musk, however, has dismissed the idea of teaming up with anyone else.
Mehta from G2 says that he anticipates most big cities will eventually settle with one or two major players, similar to Uber and Lyft competing in most US markets. “If there’s a third player, and you’re sub-scale, you won’t achieve the necessary unit economics,” he says. “It’s going to be a winner-takes-most market . . . It’s just not obvious who wins.”
If advocates of autonomy are to be believed, road safety may be the ultimate winner. But there will be a human cost too, with robotaxis presenting one of the most visible examples of how artificial intelligence is disrupting the workforce.
Uber executives insist its service will still need more drivers, not fewer, in 10 years’ time, as autonomy accelerates the shift away from car ownership.
Erik the Uber driver is philosophical. “Humans adapt, we always have,” he says. “This is just another opportunity.”
Additional reporting by Stephen Morris in San Francisco
Comments