The new cloud gaming services (Stadia, xCloud, etc.) demonstrate that we can run computationally demanding graphical applications remotely with acceptable input latency and high visual fidelity. Regular, everyday usage of a PC is much less demanding than running a game, and video of regular desktop usage is much more compressible and less bandwidth hungry than video of a fast-moving game. There's a huge pricing disparity between Stadia (like $5/month with one-time fees for application purchases? Somewhere around there...) and virtual desktop services like amazon workspaces (like $700/mo for a similarly powerful, reserved machine or $22/mo+$2/hr of usage).<p>Do you think we might be moving to a new ultra-thin client structure for personal computing? What are the challenges?
Something clear from Stadia is that the prices are not real.<p>You or I could rent a server for games and make a service like Stadia on top of Gcs/Aws/Azure -- but the pricing would be constrained by the factors you mention.<p>The one time purchase might change things a little -- some of the controller will be bought for Christmas and never played, others will play just a little, so you might be able to pay for 40 hours of gaming or so at market rates.<p>Google doesn't pay market rates, but it has the opportunity cost of not selling servers to customers that will pay more.<p>The business doesn't have to work financially for Google since they have "more money than God" and because they might be happy to kneecap EA, Msft and Sony the same way that Xbox was insurance for a home computing transition away from Windows.<p>Of course there will be a day that the bean counters take a look at how much cash it is bleeding and decide to axe Stadia. Most likely it will be a rounding error for GCS.