Before Google started working on its self-driving car project in 2009, co-founder Sebastian Thrun had a difficult time convincing automakers of the merit of the technology.
Years later, when the project — now an Alphabet subsidiary called Waymo — unveiled its car prototype in 2014, it became increasingly likely that the company might itself become a car manufacturer. Rumors that other companies like Apple were beginning their own forays into building an autonomous car began popping up around the same time.
But it turns out building a car is hard. And that fact has changed how tech companies are approaching the industry. They’re not trying to build cars. They’re trying to build — and own — the operating system that will drive the cars.
Over the last few months, as Google’s self-driving unit Waymo prepared to spin out of Alphabet’s moonshot lab, the company made it clear (time and time again) that it’s not building its own car. The company is building the driver.
Apple, too, has reportedly backed off its plans to build a car and instead may be considering licensing its software out to automakers.
Waymo, for its part, is reportedly considering expanding its partnership with manufacturers like Fiat Chrysler to include building a fleet of self-driving Google cars to operate in a ride-hail network. Ostensibly, as part of the partnership, Waymo will continue to build upon its existing self-driving software while also creating the on-demand logistics platform.
So now that the tech industry got smart to the idea that building a car is easier said than done, it is betting that it can now build a ride-hail network that is as efficient as Uber’s.
The challenge here is the ride-hail industry is incredibly competitive simply because drivers and riders have little to no brand loyalty: It’s all about pricing and convenience. That competition for supply and demand is made more difficult — and expensive — by how easily the companies can game marketshare by rolling out subsidies and discounts.
Uber has also spent the entirety of its existence making the platform more efficient, and it still has much more room to grow.
Tech companies may be underestimating the difficulty of building a ride-hail network in the same way it may have underestimated the difficulty of building a car.
It’s certainly why automakers like General Motors and even Ford have decided to partner with or acquire companies with an existing network like Lyft and Chariot.
(I made a similar argument about Elon Musk’s “master plan” for an on-demand network of self-driving Tesla cars.)
Sources previously told Recode that Google missed its chance to partner with Uber when it could — a relationship that has since soured.
Alphabet still has a leg up on its traditional automaker counterparts when it comes to building a ride-hail network. For one, it has already launched a ride-hail service through its mapping platform Waze; it also has considerable mapping, AI and other user behavior data and expertise that would certainly help it in its effort to catch up to and ultimately compete with Uber.
If anything, it’s clear that the tech companies that were once dipping their toes into the auto industry’s waters are now wading in Uber’s pool.
This article originally appeared on Recode.net.