Carmakers Seek a New Revenue Stream in Hot Technologies
Consumers are putting a strain on automakers almost everywhere by buying fewer new cars, instead fixing up and holding on to the wheels they have or buying used vehicles. Now, to make up for the shortfall in demand, carmakers are falling back on a traditional strength that has turned out to be in rare global supply—the ability to manufacture complex things at industrial scale.
In particular, automakers are pushing into hot technology industries that require both elite manufacturing expertise and skills they have newly acquired in AI.
The industry’s diversification drive is not new—it started with Tesla CEO Elon Musk, who as far back as 2021 declared that the company was no longer a carmaker at all, but an AI developer that would operate robotaxis and manufacture the Optimus humanoid robot. The same year, South Korea’s Hyundai bought humanoid robot maker Boston Dynamics.
But the trend has picked up in recent months. Last month, General Motors said it would manufacture large batteries for the grid and AI data centers, aiming to deliver its first systems in 2028. Meanwhile, Renault has begun to make drones for France’s military, and China’s BYD says it will manufacture humanoid robots.
There is no proof that any of these business lines will be lucrative. But carmakers are loath to keep doing the same thing while the car industry almost everywhere shrinks.
S&P Global Mobility predicts that U.S. car sales will fall to 15.7 million units this year, down 3% from last year and 10% from their peak in 2016. While EV sales have recovered in parts of the world for Tesla and some Chinese exporters, part of this drop is a plunge in EV sales among the traditional carmakers: Ford’s EV sales in the second quarter fell 41% and GM’s 33%. In China, too, S&P expects car sales to contract 7% this year.
“Everyone wants a piece of physical AI and the AI build-out,” said Neel Mehta, an investor with G2 Venture Partners.
To a degree, history validates the diversification moves. In the 1970s, Ford made tons of money as a megadeveloper of planned communities, and in the 1990s, GM got into satellite television, launching DirecTV. In 2003, GM sold DirecTV’s parent company, Hughes Electronics, to Rupert Murdoch for $6.6 billion in much-needed cash, though it would have done far better had it sold the franchise a couple of years earlier, during the dot-com bubble (Murdoch later sold the satellite TV firm, and it has had a number of owners since then).
Ford is among the few carmakers to already see a payoff from the plunge into another industry, in its case batteries. It’s building factories in Michigan and Kentucky to make gigantic stationary storage systems for the grid and AI data centers. In May, investors—seeing Ford’s new direction as a play on the AI boom—bid up its share price 45% in a two-week surge after it announced its new Ford Energy division. The shares have since dropped considerably, but they are still up 11% since the announcement.
GM’s move into batteries follows the massive success Tesla has enjoyed in the stationary storage business. Last week, Tesla reported a 41% year-on-year jump in its deployment of stationary batteries in the second quarter.
Chinese EV startups are also following Tesla into humanoid robots. Last month, BYD said it had begun developing one. Earlier in the year, Li Auto said it planned to make humanoids as well.
XPeng, which announced its own foray into humanoids in 2023, plans to begin commercial sales of its robot, Iron, next year. And last month, CEO He Xiaopeng said he was taking direct charge of the company’s humanoid unit while Iron moves toward the commercial release.
Recent prototypes eerily resembled a woman, with nimble movements and humanlike skin and musculature.
In April, XPeng took a small group of journalists on a tour of the plant in the Chinese city of Guangzhou, where the company will begin to manufacture Iron at large scale later this year. The plant is near the factory where XPeng makes its EVs, and a company official said it was making a natural progression from using large AI models in autonomous car systems to employing the same intelligence in robots.
In a meeting with reporters, He said XPeng would begin by deploying Iron—which can converse in multiple languages—to greet visitors in stores and factories.
G2’s Mehta predicted that while the move into batteries was bearing fruit now, it would take much longer to deploy robots at large scale. Still, he said it was possible that investors would sharply bid up some automakers’ value based on their expansion into these other technologies.
“Outside of Tesla, I think it’ll happen at scale in China first,” he said.