tl;DR - Google didn’t understand their customers and made bad choices with Stadia.
Braindump time.
When Blizzard was getting ready to launch World of Warcraft the high water mark was 1 million subscribers. With EverQuest in the low hundreds of thousands - I forget the exact number - a million seemed astronomical. And really the limiting factor was the backend infrastructure - or more accurately the price tag. There was a real possibility Blizzard would effectively go broke if WoW failed. Blizzard’s parent company at the time was having cashflow issues and we ended up leasing the first two datacenters to afford the hardware. We limited the number of copies by only printing so many boxes, estimating 20% concurrency based on Diablo II and other games. That is to say, for every 5 copies sold only 1 would be in use at any given time - like a gym membership. We also expected some number of boxes to sit on people’s shelves unopened as collector’s items.
The game launched with 35% concurrency and creeping even higher on weekends. All those “collector’s copies” were sold on eBay for $1,000 and up. We found that the limiting factor was the character databases. Fortunately we had dark database servers and storage in place - enough that we were able to add 80% more capacity within a week or two.
Ultimately WoW was a hot item for quite a while. We spend years deploying more and more datacenters to keep up with demand. The game eventually peaked at about 12 million subscribers, but no one knew where the end was going to be until it happened.
That’s an example of how wrong a company could be about the market for their game in the positive direction. But Blizzard looked at the prior attempts to do this kind of game and gave it an educated guess. We were off by a lot.
In contrast every MMO that came out for several years after WoW launched as a purchase plus subscription model like WoW. Every new MMO was a “WoW-killer”. It became a joke, because none of those games ever came close to WoW. Several fortunes were lost chasing that dragon.
So these are examples on the other extreme. And maybe it took a dozen tries to beat WoW before people wised up because WoW was so hugely successful - it could be done! But when Blizzard launched World of Warcraft it wasn’t at all apparent an MMO could reach one million subscribers.
None of the game streaming services before Stadia were a clear success. OnLive went broke and sold out to Sony. Gaikai never produced much but technology, and also sold out to Sony. Sony built PS Now - originally offering PS3 games in the cloud. Sony shipped PS Now clients for PC, Mac, PS4, PS Vita, and in their 2017 TVs. Eventually they dropped all but the PC and PS4. The price of the service dropped significantly as well. It’s still around, but pivoted to offering PS4 games for download too.
Someone convinced Google that Stadia could be the next big phenomenon. Tying in to Youtube makes a lot of sense! Google has datacenters they rent out to others, so they have inherent economies of scale. Internet in the US is - slowly - getting better thanks to Google Fiber and others. 5G promises low latency mobile Internet. I can see it - if you overlook the fact that a platform entirely based on game streaming is a solution looking for a big problem. The one use case which makes sense to me - lowering the cost of entry for play games - was undermined from the start. But I can see how they got to some big expected numbers.
The problem was Stadia tried to sell a discount game console while charging full price for games, turning off existing gamers who could get the games on familiar platforms - often for less money. So, on one side - the hardcore gamers who are already familiar with these games and have ways to play them. The other - uninitiated masses who are content with mobile games and just don’t care. The Stadia customer Venn diagram was people interested in modern video games but too apathetic to buy a console. Not a huge customer base. All Google’s inherent advantages couldn’t make up for the small target market.
Had more folks at Google involved in the decision making process understood the potential customers better they could have saved a lot of money and goodwill. They didn’t understand though, and made some bad choices as a result.