1. Steam Machine 2: VR
Pairing a Steam Machine with an "it just works" concept of the Index that can play all of your old PC games (even ones that don't run on Windows 10) and all of your VR games at a decent resolution.
2. Oh shit... Xbox Game Pass is actually a great deal....
They may want to piggy back off of Microsoft's UWP concept to make a game-subscription service for the entire (or most) of the Steam catalog. Valve knows that games industry is going trough a large industry shift. Most consumers don't really care that they're selling out all of their rights for convenience (hell that was Steam's main realization).
3. lol Google thinks people trust them with games haha... but Stadia isn't too bad of an idea...
Containerizing games and packaging up their streaming software (Steam Remote Play) and maybe some edge hardware from #1 would be a great value for a PC gamer. The industry has been attempting to push itself to thin client fat server models for quite some time and I wouldn't be surprised if Valve wanted in on the action with some of their existing investments.
4. They may actually think they're a compelling offering to the couch gamer.
Steam Machines, Steam Links, and the Steam Controllers were neat but, maybe, the execs at Valve think there's enough traction to beat out the existing competition (Nintendo Switch/WII/Gamecube/..., Xbox Series ..., Sony PS..).
I used to think that, but I think computers will get cheaper faster than latencies will get better.
Look at Xbox. Anyone who can afford stadia, would much rather just buy a cheap loss-leader Xbox. Unlike industry compute (where the perks of AWS/Azure/GCP far outweigh having on-premises hardware), gaming is incredibly well suited to having on-prem hardware.
Make them super cost effective (single rendering team/networking team/physics team) and have them churn content into Game Pass and have an impossible to resist offer.
By making Linux more compatible with the games that Valve sells through their store it puts Valve in a better situation to survive this possible course of action by Microsoft because they can switch to selling hardware preloaded with their SteamOS.
They can iterate round after round on this even if it doesn't become a billion dollar revenue stream in 6 months.
Happily :-)
We haven't conquered the desktop yet, but I see more and more people - including even in sales - voluntarily using Linux at work.
And while we haven't conquered the desktop, Linux is huge on mobile - where Microsoft has given up.
Even (parts of) Microsoft realizes where this is likely to go as shown by the fact that they are busily positioning themselves for it.
If you mean Android, the Linux kernel is an implementation detail, only exposed to device OEMs and Google itself.
Userspace is Java based runtime, and in what concerns the NDK, there is nothing Linux specific about the public APIs,
https://developer.android.com/ndk/guides/stable_apis
Since Android 7, Google has increasingly introduced measures to kill process that try to get dirty with private stuff, like directly calling int Linux stuff.
On of the main reasons why termux might not have a bright future, specially since its developers refuse to contaminiate their workarounds with Java/Kotlin code.
If you mean ChromeOS tablets, well even less given that Web platform doesn't care what OS it runs on, and Crostini actually runs on top of a virtualized container, WSL v1 style.
As for Microsoft, they have been one of the first UNIX vendors for PC with Xenix, had an initial POSIX support on NT due to goverment contracts, and have just reacted to the market of Linux guys that give Apple money instead of sponsoring Linux OEMs, so that they give money to Microsoft instead of Apple. The Linux OEMs still don't see any money from those "Linux" lovers.
Succinctly, it's easier and cheaper to port games wholesale than it is to: build out yet another platform for developers to target, on board them, and on board consumers to this new console.
What steam has that consoles don't, are is massive preexisting library. The problem is it mainly targets windows. Their options for bringing this to the console market are either pay for windows licenses (which on top of hardware, would make them more expensive than the alternatives), or port in wholesale. Most console games get a pc (windows) port eventually.
So if they can get proton to a mature point, they'll be the defacto winner of the console wars.
I know the issue isn't settled yet, but does anybody expect the SCOTUS to do anything besides rule for Oracle?
That said it's an attempt to break the chicken/egg cycle. The crowd is small because the crowd is small, if you offer a way to grow it artificially than you can gain critical mass.
The big wrench is DRM/anti cheat though. That's the major thing holding me back from gaming with proton exclusively.
Aside of political motives regarding Microsoft, the technological problem that Valve is solving with using the flatpak technology is creating a single Linux build target for all Linux game releases.
A recurring complaint from gamedevs is that it's difficult to support many different Linux distributions/versions/runtimes. Steam already does a decent job providing its own runtime to link against, but this goes several steps further in abstracting away from the distribution that a user may be running.
We are talking an incredible amount of effort for something that benefits a pretty small crowd.
The only real money making avenue would be that Valve launches its own mature Linux based OS for gamers. Low key tho, they would be trying to be the first real alternative to windows/macs.