Bandwidth restrictions and general network limitations are a hassle, but not my main concern with streaming games as they’re something that I can predict improvement in. My major concern is business principles being applied to cloud computing.
With a console/PC, you have a guaranteed allotment of processing power which allows for a certain level of performance. With a subscription-based cloud computing model, it would be economically crazy for the provider to have an infrastructure that can provide that to every single customer simultaneously - you’ll work on the basis of average performance requirements across your userbase and provide some statistical leeway. It’s not like any web company has enough server capacity to service everyone on the globe simultaneously.
That’ll work decently at most times, but I have concerns.
Logically you’ll have local processing points, say per country or broader region - otherwise the sheer physical limitations on response times grow too considerable. This means that at certain hours of the day you’re going to have higher draw on the same equipment as in a single region, you’ll have more people playing at certain times. Cost-cutting here could have an impact on the quality of service you receive at the times you want to play most. What about holidays, Christmas etc.? Higher demand, more struggle to cope. Frustrating if your computer works worse exactly when you need it and can use it.
A new, popular game is released, servers can struggle. Imagine if that increased demand meant that EVERY game performed worse, because the hardware itself was being divided too thin across too many people.
Where we’ve tested large networking solutions and are improving them rapidly, cloud computing isn’t something that (so far as I’m aware) has been implemented to the scale that would involve. I’m a little skeptical.