The idea has been around for even longer. The first service I remember (and afaik they were actually the first) was onlive. Launched in 2010.
They had already nailed the tech then. Latency was acceptable and so were compression artifacts.
But they eventually failed and so did every other subsequent service that tried to do the same thing.
Not only does it solve a problem that barely anyone has, it’s also insanely expensive to run. I don’t think any of these services ever had a clear path to profitability.
Yep, it all comes down to have enough gamers to afford expensive cloud infrastructure. And a regular geo-distribution of computing resources is not working well in this case.
Let's assume the usage peaks on Friday eve in LA area. The LA resource pool must have enough capacity for the peak. And on Saturday early morning that usage drops almost to zero. Now we might want to reuse freed resources to serve another currently active user pool, maybe in Tokyo. But we are dealing with highly specialized cloud resources not that easily reusable. You can't use LA servers to serve users in Tokyo - the latency will be too high.
Now we need to deploy our resources near all big urban population centres for all target markets. And these resources would stay mostly unused because of the latency and usage patterns (unless we're going to mine crypto with unused GPUs). And once we move out of densely populated areas, the problem becomes bigger.
So it is really hard to see how that can be economically feasible.
They had already nailed the tech then. Latency was acceptable and so were compression artifacts. But they eventually failed and so did every other subsequent service that tried to do the same thing.
Not only does it solve a problem that barely anyone has, it’s also insanely expensive to run. I don’t think any of these services ever had a clear path to profitability.