Consumers don’t usually think of companies like Uber, BMW, and Apple as competitors, but such disparate companies are up against each other in a contested field: self-driving cars. Big tech companies, automakers, and transportation startups are all investing heavily in self-driving technologies, and it’s kicked off a heated race between them.

The stakes of that race are high. It's possible that billions, even trillions of dollars of theoretical value are going to waste each year in the form of cars sitting idle in driveways and garages or of people sitting unproductively behind the wheel. And that’s not to mention personnel costs in the trucking and taxi industries, which stand to fall dramatically when the practice of paying human drivers becomes obsolete. Whichever firm can bring autonomous cars to the mass market the soonest, while delivering the a high-quality product, gets to stake out a tremendously valuable claim. “I tend to think there won’t be a lot of unique solutions out there,” says Richard Wallace, the director of transportation systems at the Center for Automotive Research, an auto-industry research firm. “We’ll dovetail towards the best one.”

On Sunday, The New York Times reported that Waymo, the autonomous-driving-tech arm of Google’s parent company, Alphabet, signed a deal to partner with Lyft, the ride-hailing company that perennially comes in second place to Uber (at least in purely financial terms). Lyft itself has a close relationship with GM, which invested half a billion dollars in the company last year. (And GM also bought the self-driving-tech startup Cruise outright in the spring of 2016.) This new partnership may represent the first of the contours of the automotive marketplace of the future come into view. “It looks like a joint front against Uber. I don’t know if that’s truly their motivation, but it certainly wouldn’t surprise me,” says Wallace.

In talking through the Waymo-Lyft partnership, Wallace used an analogy to air travel to sketch out what the automotive industry of tomorrow may be like. “At the end of the day, you may have three or four, possibly even five very strong ‘alliances,’ let’s call them, for lack of a better word,” he says. “It would almost be like airlines. Take Delta, KLM, Air France and their other partners. [As a consumer] it almost doesn’t matter which airline you end up on if it’s one of them. They’re all operated under the same agreement, and they’re focused on fighting against the United alliance and against the American alliance and against the Southwest alliance, if you will.”

For the time being, the most immediate hurdles between the human-driving present and the autonomous future are still technological. Nobody has perfected programming a car to use an array of sensors to drive safely in all conditions, for example handling blizzards and other rough weather, reacting to the unpredictable human drivers sharing the streets, or knowing how to proceed when police are directing traffic with hand gestures that are intuitive to humans but meaningless to a computer looking out at the world through a LIDAR array. But as machines learn and software engineers program solutions, the important question isn’t just about the development of the technology; it’s over how it will be deployed, and by whom. “The more interesting question than who’s ahead right now might be who has the best business model to field it,” Wallace says.

Companies are having just this argument now, and it's a blood sport. First, there are the tech firms, which say they’ll win because they're closer to solving those persistent technical problems. Arun Sundararajan, a professor at New York University and the author of The Sharing Economy, notes that these companies might have an edge over automakers when it comes to winning over consumers’ trust. “Wouldn’t you be more inclined to rely on the software and cybersecurity ingenuity of Google, Uber, Didi, Lyft, Amazon, Apple or Tesla (the exception), rather than trusting the digital capabilities of Ford, Toyota, Daimler or BMW?" he asked rhetorically in a comment shared with reporters.

Detroit automakers, meanwhile, are arguing that it'll be the people who actually know how to build and deliver a vehicle who will capture the market, and no amount of AI wizardry can change that. Sundararajan sees a weakness in this reasoning, though. “Automakers have invested heavily into making themselves sellers of lifestyle products rather than machines,” he says, noting the trend away from marketing cars with signature flourishes and strong brand identities.

And lastly, companies like Uber insist that, particularly when cars can drive themselves, the ownership model of car use will go defunct, and ride-hailing companies’ existing consumer-facing model will be able to simply swap out cars driven by humans to cars that drive themselves—in other words, that the people selling transport as a service, rather than software or hardware as products, will be the winners. And, Sundararajan points out, these are the companies with the most experience navigating (or sometimes subverting) any regulatory hurdles that stand in their way.

There are certainly merits to each view—that selling self-driving tech itself, or selling cars, or selling transportation is the way to come out on top. It’s not clear yet which business model (or combination of business models) will win out, but that’s what makes the Lyft-Waymo (and effectively GM) partnership so interesting, and inauspicious for rival firms from Tesla to Uber to the established major automakers investing in self-driving tech of their own: These three are effectively taking all sides of the argument, hedging their bets by forming a team with a member from each camp.