Microsoft admits its $80 billion Game Pass bet missed by half — and 3,200 people pay for the gap

Gaming 4 min 80 sources

Xbox cut 3,200 jobs and spun off four studios as its new CEO conceded the Game Pass subscription bet fell far short of the 77 million subscribers it was built for.

Key takeaways

Microsoft cut 3,200 Xbox jobs and released four studios after admitting its Game Pass subscription bet reached about 30 million subscribers against a plan for 77 million.
The company sized its whole games business — studios, salaries, an $80 billion spend — to a subscriber forecast that came in less than half as big, and the gap fell on workers.
The same week, Steam had a record year and Ubisoft's remake sold 2 million on day one, showing the subscription-everything model failed on its own terms, not the market's.

Microsoft’s Xbox division cut about 3,200 jobs this week and let go of four game studios, in what its own leadership called the most significant restructure in Xbox history [1][3]. New Xbox CEO Asha Sharma said the “business today is not healthy” and conceded the plan the company spent years and roughly $80 billion building — a games subscription called Game Pass — has not worked [5][46]. This is the fresh part: not just another layoff wave, but a public admission that the model at the centre of Xbox’s strategy fell short of its own target by more than half.

The bet, and the miss

Game Pass is an all-you-can-play subscription — one monthly fee, a rotating library of hundreds of games, modelled on Netflix. The theory was scale: get tens of millions paying every month and the numbers dwarf one-off game sales. Microsoft reportedly planned for around 77 million subscribers by now [47]. It has roughly 30 million — less than half [8][23].

The cost base was built for the bigger number. Microsoft bought studios, funded expensive games to feed the library, and paid about $80 billion — most of it the Activision Blizzard purchase — on the bet [24]. When the subscriber count came in at 30 million against a plan for 77 million, the structure was sized for a future that didn’t arrive. “We simply spread ourselves too thin,” Sharma said [17]. Being spread too thin, in the end, means people: about 3,200 of them.

Who left, and what “spun off” means

Four studios are leaving Microsoft’s ownership: Compulsion Games, Double Fine, Undead Labs, and Ninja Theory [9][28]. “Spun off” here means Microsoft is releasing them to operate independently rather than shutting them outright — the two indie-bound studios say they keep the games they were making [19]. One of the four, Undead Labs, had released no games in eight years of Microsoft ownership [11].

The deeper cuts landed inside studios Microsoft is keeping. Bethesda and id Software — the maker of Doom — were reportedly hit hard, with id losing around half its staff even as new Doom content shipped [4][6]. Obsidian, the studio behind Fallout: New Vegas, reportedly lost about a quarter of its people [41]. John Carmack, id’s co-founder, said his earlier hope that Microsoft would be “a good steward of the brand” wasn’t aging well [66].

The same week, a different result

The contrast this week was sharp. Valve’s Steam — the PC storefront Microsoft competes with — is reportedly having a record year; one analyst said its “long arc is relentlessly up” while Xbox and PlayStation flounder [20]. Steam sells games one at a time and takes a cut; it never bet the house on a subscription.

Ubisoft, too, showed the split. Its remake Assassin’s Creed Black Flag Resynced sold 2 million copies on day one, the series’ best launch in years [16][56]. Days later, Ubisoft laid off staff at its Barcelona studio that had worked on the game [14]. A strong launch and a layoff round, in the same building, in the same week — a reminder that a hit doesn’t mean a healthy company.

The model, shutting down elsewhere too

Subscription and live-service models — games built to keep selling to the same players for years — are visibly straining across the industry. Nintendo is closing Mario Kart Tour in September after years of running it [42]. Square Enix is shutting down Final Fantasy 7 Ever Crisis, a mobile spin-off [45]. Each shutdown means players lose access to a game they invested in, because the ongoing revenue no longer justified the ongoing cost.

Not everything is contraction. Palworld, the independent hit, passed 40 million players two and a half years after launch [34]. A small hide-and-seek game, Meccha Chameleon, has sold 15 million copies [49]. The lesson those numbers keep teaching: the industry’s health isn’t tied to the size of the studio, and the biggest bets are not the safest ones.

Microsoft says it may now make more of its first-party games console-exclusive again — a partial reversal of the everything-everywhere logic Game Pass was built on [22]. What it hasn’t said is where the subscriber number goes from here, or which of the surviving studios still have projects that will ship.

02 · Lesson · why it matters

You can copy the model, not the world that made it work

A business model isn't a machine you can lift and drop somewhere new — it's a set of moves that only pay off under conditions you might not be carrying with you.

The idea looked obvious

Netflix took a business built on selling one thing at a time — a DVD, a cinema ticket — and replaced it with a single monthly fee for everything. It worked. So Microsoft did the arithmetic for games. One subscription, hundreds of titles, tens of millions of people paying every month. On paper the numbers dwarf selling games one by one. The plan was for around 77 million subscribers. The company built for that number: bought studios, funded expensive games, spent roughly $80 billion.

It reached about 30 million. This week its own CEO said the business isn’t healthy, and 3,200 people lost their jobs. The model that transformed one industry didn’t transfer to the one next door.

A model carries its world in the fine print

Here is the thing worth carrying: Netflix’s subscription doesn’t work because subscriptions are magic. It works because of the world it sits in. Shows are relatively cheap to license and there are endless of them, so the library never empties. People watch television most evenings, forever, so almost nobody finishes and cancels. New things arrive faster than you can watch them, so there’s always a reason to keep paying.

None of those conditions are written on the model. They’re the ground it stands on. Lift the model off that ground and the conditions don’t come with it.

Games stand on different ground

Games break every one of those conditions. A big game costs hundreds of millions and years to build, so the library can’t stay endlessly full of them. A player finishes the blockbuster in a month or two and has no reason to keep paying until the next one — and the next one is two years away. And there’s a trap Netflix never faces: when Microsoft puts its own big game into the subscription on day one, that game stops selling copies. The move that grows the subscriber count quietly destroys the sales that used to fund the game.

So the same lever pulls two ways at once. Every push toward the subscriber number hollows out the thing paying for it. The model isn’t broken. It’s just standing on ground that won’t hold it.

The number was a forecast wearing the costume of a fact

77 million wasn’t a measurement. It was a projection — a hope with a spreadsheet around it. But Microsoft built as if it were solid: hired for it, bought for it, sized the whole company to it. When the real number came in at less than half, the building was too big for the ground under it, and “too big” gets counted in people.

This is the quiet part of any borrowed plan. The forecast becomes the foundation, and everyone downstream lives on top of it — the studios bought to feed it, the developers hired to build for it, the veteran let go after 37 years because the number he was hired to serve turned out to be air.

You are standing on this ground too

If you play, you were inside this bet, not watching it. All-you-can-play for one fee is a genuinely good deal for you — and it was also built to keep you inside one company’s world, buying its console, not the rival’s. Both of those are true at once; the convenience served you and served its maker. Now the correction reaches you as well: fewer studios making games, more titles pulled back behind a full price, big games held as exclusives again to give you a reason to buy the box. The bet was partly yours, and so is the bill.

And notice how little any single seat could see it coming. The player saw a great deal. The executive saw a chart climbing toward 77 million. The developer saw a green-lit project. Each was reading a real thing in front of them, and none of them was holding the whole — a model doing well by the one number it was judged on while quietly starving the sales beneath it. The gap between “our subscribers are growing” and “our business isn’t healthy” is exactly the space where a borrowed model hides its missing conditions.

That’s the humbling part. It isn’t that someone was foolish. Smart people copied a model that plainly worked, and did the arithmetic carefully. What they couldn’t do from inside their own success was see the ground they’d stopped standing on. Anyone lifting a plan that worked somewhere else is doing the same thing — carrying the moves, and trusting that the world came along for free.

03 · Lab · your turn

Run the Subscription

Rehearse chasing a borrowed subscription model and feel it fight itself on games' ground — every lever that grows subscribers hollows out the business.

04 · Hope · carry this

The same week a giant conceded its billion-dollar bet had missed, small studios quietly sold tens of millions of copies of games nobody spent a fortune on. Making something people love was never the exclusive property of the biggest spender, and it still isn't.