For cloud gaming, what's past is prologue. When OnLive launched in 2010, it was heralded as an example of " ." It died five years later. Fast-forward another three years, and we're still waiting for that future. But there are signs of progress.
At, Nvidia boasted the same core capability as OnLive for its PC and Mac versions (now in beta) for its GeForce Now service -- you can like it's a $2,000 gaming PC! It's less ambitious in scope than OnLive, mostly because there are already a ton of gaming social networks, but its philosophy is the same: cloud-based play from a select catalog of games. And its official launch is as eagerly awaited. (More in my hands-on with GFN, coming soon.)
On the other hand, at CES this year, and subsequently, Blade announced rollout plans beyond France to the US and the UK for its virtual machines), which shows promise for advancing the state of the art.service, rentable desktop systems running on shared server hardware (i.e.
Cloud VMs aren't new. You've been able to rent virtual machines with high-powered processors and graphics engines for gaming and to run workstation-class applications for a while, from services such as Parsec and LiquidSky.
But Blade raises the bar by pledging up to 16:9 4K/60fps or 1080/144fps -- all other services currently limit you to 1080/60p -- and the ability to almost seamlessly handoff to other devices running MacOS, iOS, Linux and Android as well as smart TVs and monitors via a small, relatively inexpensive Box accessory. In other words, you can start a game on one device and pick it up on another. (Blade is still in preorder and hasn't rolled out on the east coast yet, so I haven't been able to use it.)
The network problem remains
A powerful VM can be a relatively complicated gaming setup compared with GFN's restricted-selection, console-like experience, but the tradeoff is the ability to install and play any game you own or that you license from a DRM-management service like Steam, as well as the ability to run all your other applications. The pricing can get pretty complex, too, though Blade simplifies it by taking storage out of the equation; your only current option is a relatively small 256GB SSD.
OnLive could do this eight years ago because cloud gaming systems offload the processing burden to the remote computer, and the only local processing that needs to be done, at a minimum, is essentially decoding the compressed stream being sent from the server and sending back keyboard, mouse and controller input -- a job even cheap modern processors and internal GPUs can do quickly.
But that means these services' dependence on stable network bandwidth with low latency (how fast a data packet gets from the server to your device) remains as important now as it was then. And they still recommend wired controllers and wired Ethernet connections for PC and Mac rather than wireless, though I find 5GHz wireless can work well if you play at off hours.
The problem of scalability remains, too -- will they be able to ramp up hardware fast enough to maintain quality as demand rises? Nvidia probably can because its approach requires vastly less storage per person, as it doesn't store games after you log out and you can only play one at a time. But others, like Blade's, which use collocated data centers, may face hardware resource constraints; that's why they roll out so slowly.
Taking the internet out of the equation
It doesn't have the low-hardware-cost appeal of a service, but Parsec offers has a free client application that anyone can use to stream games on a local network from a gaming-powered machine to a cheap system, somewhat similar to Steam's In-Home Streaming but for any installable game, not just those available on Steam.
The company licenses a developer's kit, and at CES 2018, HP announced its custom version, branded, which is integrated into the company's gaming control center software.
To a larger extent you can control the stability and latency on your local network, so this setup removes the problem of network congestion. Plus, it can turn single player into multiplayer (two players, I think), as long as you're on the same network subnet.
A new wrinkle for the US
These services still live or die by bandwidth and network congestion, in addition to price and hardware scalability issues. An inconsistent connection can cause short freezes, laggy response to input, audio dropout and popping, pixelation and display quality so bad that you can't tell a rock from a demon. I hate wasting ammo on rocks.
That means, on top of all the existing challenges, our gaming future can also be killed by a lack of neutrality in the pricing of network bandwidth. This isn't a mature market with multiple big players. What if the companies with deep pockets could pay to deliver higher, more stable bandwidth for their less demanding, game-limited, DRM-heavy ecosystems so they could deliver a relatively better experience plus have bandwidth headroom for improvement? Or alternatively, if the small competing services had to offer more expensive, guaranteed-stable tiers to compete with big companies that could absorb the cost of a better connection? In my opinion, we'd end up with the worst of both worlds: the most limited versions of the services and no competition with large players gobbling up the smaller ones.
Plus, with carriers downgrading speed when you hit your secret cap on "unlimited" mobile data, playing away from Wi-Fi on your phone or tablet will be difficult to impossible. Unless a company partners with a selected carrier to offer "free," truly unlimited high-speed mobile data for gaming through its service, which usually requires some market clout to do. And the mobile aspect of cloud gaming is something we really want to encourage, don't we?