Why Uber and Lyft's Best Move Against Robotaxis Is Cooperating With Them

Supply is the whole business, and that's the vulnerability

A ride-hailing marketplace has almost no leverage once it's dependent on a supplier for the thing its entire business runs on. No supply, no company. If an autonomous-fleet operator decides to go around a marketplace and build direct-to-consumer demand instead of supplying it, that's a real structural risk — the marketplace can't simply replace that supply with more human drivers overnight if the fleet operator's cars are meaningfully cheaper.

But the fleet operator has the mirror-image problem: owning the cars doesn't solve the harder half of the business, which is acquiring riders directly. That's a genuine chicken-and-egg problem, and it's the reason the obvious "just go around Uber" move is harder to execute than it looks.

Why the obvious move — build a competing app — keeps not happening

Whether it's actually worth it for an autonomous-fleet operator to go fully around an existing marketplace and compete for riders directly is a genuinely open question. Companies with deep resources have tried building consumer-facing transportation businesses before and struggled with it — running a fast-moving, operationally messy, hand-to-hand marketplace business is a completely different organizational culture from running a mature, slower-paced, largely monopolistic core business. People who've spent years inside that kind of company and then land inside a marketplace operator tend to visibly struggle with the pace change — the day-to-day intensity of the two businesses isn't close.

In the near term, that gap in operating DNA is exactly why the more efficient path for a new autonomous-fleet entrant is simply putting cars onto an existing marketplace's demand — not building a competing app from scratch.

Multi-apping, one level up

The driver multi-apping dynamic already covered elsewhere on this site may simply move up a level. Picture an individually owned autonomous vehicle that drives its owner to work, then spends the day running trips across several ride-hailing marketplaces at once — whichever pays best in the moment — before returning to pick its owner up. It's the same logic that already makes a human driver's loyalty structurally weak: the asset isn't captive to any one platform, so it goes wherever the money is. That's likely a decade-plus-out picture, not a five-year one — full self-driving has been described as "almost there" for close to ten years already, and a fully autonomous, self-managing personal vehicle is a harder problem still.

The scenario that should actually worry incumbents

The more dangerous near-term play isn't a slow multi-year rollout — it's an autonomous-fleet operator flooding a city with thousands of free or near-free rides for a year, purely as a customer-acquisition move, specifically to pull riders off the incumbent marketplace before it ever turns on real pricing. A large existing marketplace's cost structure — significant corporate overhead relative to a hardware-subsidized entrant — makes that kind of price war genuinely hard to match.

The mitigating factor: autonomous-fleet operators aren't likely to hand any single marketplace that much leverage either. The more probable outcome is cooperation with many platforms at once — a dozen or so different fleet operators, each supplying multiple marketplaces rather than picking one exclusive partner — with the fleets themselves effectively multi-apping the same way human drivers already do. If that's how it plays out, the whole category likely ends up looking like a lower-margin, more commoditized business than the current narrative around autonomous fleets suggests — the same downward pressure driver multi-apping already puts on take rate, just moved up to the supply layer itself.