A Decent Week at Microsoft Gaming, With Caveats

Microsoft canceled Copilot for Xbox consoles this week. Six months ago I would not have believed that sentence existed. Six months ago, Microsoft was wedging Copilot into Notepad.

Last week I published a post arguing the honest answer to “should I switch to Linux for gaming” is yes. I still think it is yes. But the news cycle since I hit publish is the first one in a while that did not actively make the argument stronger.

Xbox Mode shipped to Windows 11 on April 30. It is a controller-first, full-screen interface that minimizes the desktop, surfaces your library across Game Pass and your other PC storefronts, and finally answers the SteamOS pitch instead of deflecting it. The actual product is good. It was also called “Xbox Full Screen Experience” three weeks before it shipped, which is not the timeline of a confident brand. And the layout pushes Game Pass harder than the marketing makes obvious. The first tab after Home is Game Pass. The next is Cloud Gaming, which is the same subscription wearing a different hat. Two of the six top-level tabs in a controller-first PC gaming UI point straight at Microsoft’s subscription service.

On May 6, Asha Sharma confirmed Microsoft was halting Gaming Copilot development for consoles and winding it down on mobile. The console version had been on the roadmap since GDC 2025 with the standard 2025 AI sidekick story, and now its entire lifespan on console is the beta. Killing it cost Microsoft very little, since nobody was using it on hardware they have been quietly de-prioritizing. But it also breaks a two-year pattern of wedging Copilot into every product whether the product wanted one or not, and that is the first time in memory Redmond has looked at a planned rollout and said, not here. Either way, it beats the version of this story where they shipped it.

Sharma replaced Phil Spencer in February, after his 38-year run, coming from product leadership inside Microsoft’s CoreAI org. Three months in, she has killed the console Copilot, shipped Xbox Mode, and this week brought in a wave of new leadership pulled almost entirely from CoreAI: Jared Palmer on engineering and developer tools, Tim Allen on design, Evan Chaki from CoreAI’s general management, David Schloss on subscriptions and cloud, and Jonathan McKay, formerly head of growth at OpenAI’s ChatGPT, on expansion. Kevin Gammill, the VP of Xbox UX and game development, is on his way out.

It looks like a contradiction. Sharma killed the most visible AI-in-gaming initiative on a Tuesday and shipped an org chart full of CoreAI alumni on a Wednesday. The optimistic version is that the wrong kind of AI in gaming is dead and the right kind, the kind that lives in the rendering pipeline rather than as a chat overlay watching the player, is what this team is going to build. The pessimistic version is that you do not staff Xbox with the people who ran ChatGPT growth unless you intend to ship something that looks a lot like what you just killed, with better marketing. I do not know which is right yet. Microsoft has the engineering chops to make the optimistic version happen. They also have a recent track record of optimizing for the wrong thing in this exact space and shipping it anyway. Both can be true.

For now I am still on Linux, and still recommending it for the people who can make the switch. But this is the first stretch of Microsoft Gaming news in a long time that I read without immediately bracing. That is not a recommendation, only a low bar that Microsoft was not clearing six months ago.