Robotaxis show signs of resilience, but now they need a business plan

Despite enormous technological hurdles and economic headwinds, self-driving vehicles with no humans behind the wheel are carrying paying passengers in San Francisco, Phoenix and China. Now comes the hard part.

Companies have invested an estimated $160 billion in automated and self-driving technology over the past dozen years, according to global management consultancy McKinsey & Co. Yet the viability of the fledgling robotaxi industry remains tenuous. Those attempting to scale a technology that once promised to upend transportation still must commit more money and summon more patience.

“Right now, there’s an acknowledgment that it’s going to be a slow-drip adoption,” said Philipp Kampshoff, a senior partner at McKinsey & Co.’s Center for Future Mobility.

That drip may become a steady trickle, should the industry stick to its blueprints for the year ahead.

General Motors’ self-driving car subsidiary Cruise concurrently operates as many as 150 robotaxis in San Francisco. The company anticipates deploying its purpose-built podlike Origin taxi this year. Fleets in Phoenix and Austin, Texas, could reach a size comparable to San Francisco’s by the end of 2023, said Cruise CEO Kyle Vogt.

Waymo established the world’s first self-driving commercial service in metro Phoenix in 2018. The company has expanded that service area, and it now runs a precommercial “Trusted Tester” program for select riders in San Francisco. Waymo announced Los Angeles would be its third city targeted for commercial activity.

Baidu operates robotaxis in Beijing, along with Wuhan and Chongqing in China, and plans to add as many as seven cities in that country this year. Such an increase would mean Baidu operates driverless vehicles across several hundred kilometers, constituting the world’s largest service area for an autonomous fleet, the company said.

Other competitors, including Mobileye, Motional, Zoox, May Mobility, AutoX and WeRide are forging ahead. But a jumbled outlook remains. Just as some are ramping up their robotaxi plans, others have given up their pursuit.

Only six months ago, Ford Motor Co. and Volkswagen shut down Argo AI, a leading self-driving startup.

One aspect in Argo’s demise may be indicative of industrywide sentiment. Car companies might favor driver-assistance systems — not robotaxis — as a faster path to profitability and a less onerous technology challenge. With limited R&D dollars, they’re shifting their bets.

“That’s the one substantial revenue stream where the marketplace is seeing a real, positive return,” said Alain Kornhauser, a Princeton University professor who directs the school’s transportation program. “You have Ford realizing, ‘My goodness, we don’t have to get to the point where we remove the driver.’ ”

Advanced driver-assistance systems provide both convenience and a seminal moment of hands-off-the-wheel operation — though they still require a licensed human driver — that might be valued at $15,000 per vehicle, he said.

Companies exclusively focused on robotaxis do not have advanced driver-assistance as a fallback. Investors and parent companies are squeezing them in the wake of Argo’s closing. TCI Fund Management sent executives at Google, Waymo’s parent, a letter in November urging the company’s board of directors to “dramatically” reduce robotaxi-related losses.

Others underscored the fundamental challenges ahead.

“The prospects of having a profitable business based on this technology keep slipping further into the future, and the costs have increased dramatically,” said Sam Abuelsamid, principal research analyst at Guidehouse Insights.

Cruise, for example, lost $1.9 billion before interest and taxes in 2022 after losing $1.2 billion in 2021, according to GM’s latest financial report. The losses resulted from higher development expenses in commercialization efforts, the company said.

Controlling those costs is essential for Cruise and others as they shift gears from proving technology toward scaling operations. The complexity of that undertaking has long been misunderstood.

Robotaxis operate like ride-hailing services. Both incorporate data analytics to determine where rides will be needed, maximizing utilization of their platforms. Customers hail them via apps. The similarities end there.

“It is an entirely different business than ride-hailing,” said Alisyn Malek, CEO of consulting firm Middle Third and former COO of self-driving tech startup May Mobility.

Unlike Uber and Lyft, which rely on independent drivers who own and maintain their own vehicles, robotaxi providers must account for vehicle ownership and maintenance costs, whether they absorb those themselves or partner with fleet management companies.

That business is akin to conventional taxi providers, something software-minded investors may underappreciate. Still, robotaxi providers might eventually offer cheaper fares and improve operator margins from the human-driven taxi model.

Autonomous taxis carry a cost between $1.42 and $2.24 per mile for operators, while conventional taxis run at $3.55 per mile, according to a September 2021 Environmental Research Letters report that analyzed both businesses using San Francisco as a case study.

And yet the “pathway to profitability is challenging, at best,” said Ashley Nunes, one of the study’s authors. Autonomous taxi operators need utilization rates as high as 78 percent and must persuade reluctant consumers to pool rides to achieve those margins.

Even then, variables exist. Removing human drivers may result in cost savings, but self-driving fleet operators must hire employees to monitor robotaxis on the road. Companies are tight-lipped on how many vehicles a single operator may monitor, and no regulations exist governing the practice. The ratio of these “teleoperators” to vehicles may have a substantial impact on operating margins, Kampshoff said.

A 2022 McKinsey study suggests the per-mile costs of operating a robotaxi may fall by more than 50 percent between 2025 and 2030 as fleet operators improve back-end operations, utilize purpose-built autonomous vehicles, optimize insurance costs and reduce hardware expenses with the price of high-performance chips declining.

Mobileye envisions an alternate path toward reaching the robotaxi market. Or perhaps two paths.

The Israeli company, which spun off from Intel in October 2022, provides driver-assist systems to the auto industry. It holds a $17.3 billion pipeline of contracts that run through 2030 for its eyes-on-the-road driver-assistance and SuperVision products, which are both camera-based.

Whereas Ford shuttered Argo AI to make these types of systems its focus, Mobileye treats its driver-assistance system revenue as both a financial and technical bridge to a self-driving future. Baidu shares that approach, and it’s one that’s resonating with analysts.

Mobileye ranks first among AV competitors in strategy and execution in a Guidehouse Insights assessment of leading companies conducted earlier this year. Baidu is third, ahead of Cruise.

“Mobileye has a very strong strategy for what they’re doing over the next several years, and they have a business that can fund that strategy,” said Abuelsamid, the report’s author.

The company further differentiates itself by avoiding assembling its own robotaxis. Instead, Mobileye intends to sell its self-driving systems to partner manufacturers — Schaeffler, Holon and an unnamed European automaker — who will then provide them to operators for use in private and public transportation fleets.

Separately, Mobileye showcased a self-driving system intended for consumer vehicles, called Chauffeur, at CES in January. Chauffeur is an eyes-off-the-road system projected to cost $6,000 on the bill of materials presented to automakers, according to Dan Galves, Mobileye’s chief communications officer.

Any approach brings the stubborn challenge of figuring out how to scale without a parallel increase in expenses.

Cruise has cracked that formula, Vogt said. The company developed its self-driving system in San Francisco because the city offered the penultimate technology challenge — dense traffic, pedestrians and cyclists, steep hills and fog. Establishing service in subsequent cities requires far less engineering work, he said.

“You build out that repeatable playbook, so that the effort it takes to launch each new city is less than the one before,” Vogt told Automotive News. We can do it for less money, less time and with a higher initial offering in terms of the service area.

Customers will have high expectations for short wait times. They’re willing to pay more if that means a five-minute wait for a ride instead of 15, yet most companies would need a minimum of 300 cars in a given area to have adequate coverage for that, Kampshoff said.

Saturating an operating area with enough vehicles to make that happen becomes both a service ultimatum and cost-prohibitive.

“It’s a question of your financial stamina,” Kampshoff said. “Even that number is probably not sufficient to be profitable. So even if you unlock this technology in a city, can you make it through this valley of tears, and finance this until it turns into a positive business case?”