Think the Google Stadia is here to stay?

No, it will Die.

 

He who wins the game streaming wars will likely make a pretty penny. Whether Google/alphabet have the funds and will to make it is a different matter. They do have a lot of money to burn though and most of the things they kill aren't exactly money spinners or things they can terribly justify having coders (a harder to come by resource occupy) their time with.

As for the one that buys it.
If they can crack the latency thing (possible but expensive -- speed of light is a problem so having servers in a third world country or even the other side of the country does not work like it does with simple bulk data, you then want something near everybody aka machines in every local area you want things in) then you avoid having to buy consoles, avoid having to upgrade your fancy PC, can have 1000 people playing online multiplayer quite happily with basically zero ping (if you are all on one machine... though might want to be local people), avoid cheats (if you are streaming a video and sending some control inputs in then hard to fiddle with memory which leaves exploits in the game engine that could also be patched in short order, not to mention be detected early if people are getting resources at a rate faster than average/expected), can reasonably pay actors to do things in a game (best AI conversation and quests ever then), don't have to worry about updates (be it doing them or making sure everybody has the same one), and still have the benefit's of the streaming company's basically supercomputer rendering all the graphics, and while a household connection is nice then with some of the takes on mobile phones you can probably do that on the go as well.
Since Steam's rise however many years and the rise of downloadable games on consoles and even handhelds this last decade or so, it has also shown many people are quite prepared to not own games if that is a thing you think will prevent people from going to it. Get the lag, selection and price sorted (not so much of a technical issue as a business one, and other than lag hardly insurmountable) and you will have plenty willing.

 

it was dead on arrival

 

No, Google will come up with a "new better" service and will kill it off just like it has so many other of their services before.

 

Microsoft's Xcloud+ultimate game pass offers a much better price proposition and library, and unless google does something, then it will be left in the dust.

 

If you can stream a video of equivalent resolution (and that was solved years ago) your connection speed is not the issue. Might hit saturation if everybody in an area did it on their phones but that is a different matter.
The ping is however and that gets us to physics. If they managed to crack the speed of light then... making a game streaming service work would be about the least profitable application to do with such toys. Alternatively that is a lot of capital investment to stick a box in every if not cabinet then local exchange.
I can stick a normal web server somewhere with cheap electricity as nobody cares about 600 milliseconds on a website. Rather harder when it is a twitchy game.

Distance there and back, plus controller input (though you could help here), plus actually calculating (though you could get this down by doing the whole supercomputer thing which is the main selling point anyway), plus getting data through (all those nice take the signal and boost it for the next section in optical cables takes time), plus the craptacular screen that the person is using (modern LCDs are not exactly as fun as old school CRT in this regard)... that adds up, and that is before we get to dropped packets and variable latency.

 

Nah, it's gonna take a lot more than what they have on offer, and at a lower price, to finally sell gamers on streaming platforms. Not to mention ISPs will have to step up their availability and speeds. 50+ ms input lag added to whatever you would normally have simply isn't acceptable, unless maybe you're sticking entirely to turn-based games.

 

I give it a year, max.

 

Pre-rendered video is a totally different beast than streaming game assets. With a 400KB download speed you can easily stream HD video, but you can't even stream a PS1 game with that speed. You can easily fit an HD movie in 4GB of space, the size of a PS2 game, which is (on average) 3 times smaller than PS3. With PS5 era we are reaching 100GB games and SSDs that are on the 500MB/s mark on their worst day. How exactly do you expect to stream that to the entire population of gamers?

 

What does one have to do with the other? Indeed the whole point of streaming would presumably be to dodge the need to stream game assets by instead having a little video controls program (albeit in the form of a game controller) and take back video that the fancy super computer cluster in the cabinet down the road/exchange in town renders*. There were some trying hybrid approaches with some local assets, however that was more for MMOs, DRM systems and some aspects of rapid turnaround development. It can yield some interesting results but that is not what most of the streaming companies are aiming for.
Latency concerns with encoding mean you might drop compression ratios a bit compared to prerendered or streams you can happily delay by easily 30 seconds (and in doing so get a bit of overhead for your computer to encode, and video encoders to get a nice thing for their group of pictures and predicted frames, maybe even a segment you can then do a two pass encoding on) but it is still well within common household internet bandwidth ranges.

*and presumably when not playing games because it is 4am gets farmed out much akin to Amazon's EC2 or AWS stuff so as to extract maximum value from it.