Some times ago I had a very short conversation with Maddingue about what could be the future of video games. Since then I've been thinking a little bit more about it.
There's a possibility that things will change quickly and a lot for the consoles, and I believe that we will move toward streaming platforms quickly. Microsoft, Google, Facebook, Amazon have the resources to make this possible. This also shows why Facebook spent a lot of money on Oculus.
Still, streaming has its own challenges (read dormando's post on what he had to do to get a smooth game experience with DQX on 3DS) : latency, lag, you need a fast and always on internet connection, etc.
We already know that Nintendo is already exploring this market with remakes of some of its games on the DS. Sony tried with Gaikai, unsuccessfully so far. They service is running and available (at least in north America) but I don't know anyone using it.
This generation of console (PS4 and xbox one) is growing faster than any previous generation. Sales for the PS4 are amazing, and even if Microsoft is lagging behind, the xbox one is selling extremely well compared to the xbox 360.
But it might be hard for MS to catch up. Because there's already a lot of PS4 outside, some publishers might decide to focus on this platform instead. Developing a game to support two or three platforms is very expensive for them, and if the majority of players are on one platform, why bother. Also it seems that the PS4 is easier to develop on (it was not the case on the PS3, so Sony has learned from this error), and is more powerful.
Now, MS also have the resources to announce in two years that they have a new and inexpensive console to play games streamed from their platform (Azure). We can even imagine that they offer it for free if you take a 2 or 3 years subscription to the xbox live. In this scenario, Sony would not be able to compete: they don't have the cash, and I really doubt that Gaikai is going to be ready anytime soon.
Let's take a look at the four other giants now: Google, Amazon, Facebook and Apple. They all build phones, tablets, and devices that you plug into your TV. They also have huge cloud infrastructure, and know how to distribute content that way.
The thing that is not clear to me is how publisher would join. Right now, all the companies provide their own SDK and if you want your game to work on different devices, you have to develop different version of the games for each device. It's in the interest of the constructor/cloud providers, but not the publishers, and this can't scale. What they might end up doing is to provide a SDK to create a light client for each platform (android, iOS, windows, Oculus), and the publisher can get his game running on the cloud infrastructure of his choice (e.g: Google Compute or AWS). Cloud providers will compete on the infrastructure (fast platform, a lot of POP, etc). This could also helps issues that we saw this year, where publishers are releasing a game but they don't have enough servers to accept all the players on day one.
The Oculus Rift is going to be a huge game changer, but for now it still requires a PC to play, and a lot of people either don't have one, or don't have one powerful enough to play games. If the game can be rendered in the cloud and just be streamed to the oculus, everybody with an internet connection can play games.
This also “fix” an old an annoying problem for gamers: retrocompatibility. If we stop developing games for specific hardware, but instead just for PCs, either on Linux or Windows, and all we have is a stream, it should be easier to play a game in 10 years that it is today.
I might be totally wrong on this. But my gut tells me that's where we're heading. We've already started to move all the computation in the cloud, video gaming is the next step (and it's one of the last because of it's complexity).
Here's my prediction: the move will come from the new players (Google/Amazon/FB/Apple) and it will be there in less than 5 years.