Is that surprising? I never assumed that Google was planning to build actual cars with this tech. I suppose I never really thought that deeply about how it would roll out, but if I had, I would have guessed partnerships with others in the industry.
If they can nail "self-driving tech" as well as they've nailed search, they'll potentially earn licensing fees from all vehicles, rather than from the slice of vehicles they'd manufacture themselves (with a long, expensive learning curve ahead of producing quality vehicles).
At first glance, Google becoming a car industry supplier like Delphi or Bosch seems just as weird. I kind of was expecting them to eventually go for the robocab business, with the steely bits manufactured by some Chinese OEM or underutilized plants in Detroit, Wolfsburg or wherever. It could be nicely phased in in limited access, limited range, low speed settings like e.g. island resorts.
But repeating the Android model, generously handing over software to whoever wants it so that they get a huge market share seems so much more consistent. This is Google, after all. If people can "ok Google" their car to actually take them to search results, Google's stranglehold on local business discovery will be even stronger than it is today. And on the reverse side, when it came to the point where people would "Hi Audi, take me to...", they could get quickly pushed of large parts of the search/decision market that originally emerged from web search and that they then so carefully extended to Maps, Android and voice assist. If cars become "Siri-smart", drivers won't use Google maps anymore. Seen like this, Google's self-driving endeavors suddenly appear like a forced defensive move, very much unlike the quirky billionaire's moonshot hobby vibe surrounding other parts of X.
> I kind of was expecting them to eventually go for the robocab business
Google's already using Maps to commoditize ride services (which will transfer to robocabs once they are on offer), so I can see them having the front and back end of that business while all the boring bits in the middle are commodities provided by others.
Exactly. Search, Maps and now Android (with and without voice assistant) are basically methods to sell influence on our decisionmaking. When cars become the day to day search interface for drivers/riders, Google will be in a much weaker position for that.
Except if they somehow achieve a relationship with hardware companies as symbiotic as in Android smartphones. Just being a backend search provider would be a step back for Google, because that is pretty much a commodity transaction where gatekeepers can negotiate for a sizable slice of the pie (see Firefox).
This. The point is to sell a few hundred million licenses for the software, as software companies like to do.
Car companies likely don't want to further increase already large R&D budgets to build their own self-driving tech, so the "synergies" of doing this are excellent.
I think the insane part of Google's original plan was the "no steering wheel" part, not the fact that they wanted to manufacture the cars themselves. There are so many corner cases. I can't imagine that they would get the tech working 100% out of the gate without needing some sort of manual override.
For a little 25MPH mini-car, full automation was and is quite practical. Google could have manufactured those things and put them on campuses, senior communities, and such. About five companies are already making slow-speed autonomous shuttle vehicles, most of which have no manual controls.
>I never assumed that Google was planning to build actual cars with this tech
They did seem to hint that they were going to. They did invest in their own, in-house built prototypes. If you start here: https://www.google.com/selfdrivingcar/faq/#q4 , you'll see the types of partners they were talking about working with. These aren't automakers.
If they can nail "self-driving tech" as well as they've nailed search, they'll potentially earn licensing fees from all vehicles, rather than from the slice of vehicles they'd manufacture themselves (with a long, expensive learning curve ahead of producing quality vehicles).