Xbox admits the price hike cost it "millions" of Game Pass subscribers

Gaming 4 min 80 sources

At Summer Game Fest, a Microsoft executive said a 50% Game Pass price rise shed millions of subscribers in months — a rare confession of how soft a big number can be. Plus Sony's exclusives keep selling fewer copies, and a record 212 million Americans now play.

What the big numbers were really worth

This week the games business said the quiet part out loud. A Microsoft executive admitted, on a stage, that a price hike cost Xbox Game Pass “millions of subscribers” in a matter of months [37]. New data showed Sony selling roughly half as many copies of its own games as it did five years ago [12]. The thread running through both: a headline number is only as solid as the conditions it was counted under — and when those conditions change, the number can move fast.

Xbox admits the Game Pass hit

At Summer Game Fest’s “The Game Business Live” event on Monday, Xbox chief strategy officer Matthew Ball said the company “shed millions of subscribers over the span of a few months” after raising prices in late 2025 [37]. Game Pass is Microsoft’s subscription — a flat monthly fee for a rotating library of games instead of buying each one. Last October, the top Ultimate tier jumped 50%, from $19.99 to $29.99 a month, or $120 a year [18]. People left in droves [35].

The mechanism here is simple and easy to miss. A subscriber count isn’t a measure of how many people love the service. It’s a measure of how many people decided it was worth that price, this month — a decision they remake every billing cycle. A chunk of any subscriber base is sitting near the edge: paying out of habit, half-meaning to cancel, staying because the friction of leaving is higher than the cost of staying. Move the price up 50% and you find out, all at once, exactly how many of those people were ever really yours.

Microsoft does not publish Game Pass numbers, so we don’t know the exact toll [37]. The last confirmed figure was around 34 million members in early 2024 [35]. What we do know is what the company did next: it walked the price back. By April 2026, under new Xbox CEO Asha Sharma, Ultimate dropped to $23 a month — still above last year’s $20, but a clear retreat [37]. Microsoft also pulled new Call of Duty releases out of the service to lower the cost [37]. Ball’s framing was that “the value has changed” [37]. The plainer reading: they pushed too hard, watched the number crater, and reversed.

For players, this is the rare case where leaving worked. Enough people cancelled that the company felt it. The lesson cuts the other way too, though — the cheaper price came with less in the box (no day-one Call of Duty), which is how subscriptions usually “lower” a price: by quietly shrinking what you get [37].

Sony’s exclusives keep selling fewer copies

A second number deflated this week. Data collated by Game File from Sony’s own filings shows first-party PlayStation sales — copies of the games Sony makes itself — peaked at 58.4 million in the 2020 financial year, then fell every year after, bottoming at 28.9 million in 2024 [12]. The 2025 figure recovered slightly to 32.1 million, but that’s still roughly half the peak [12].

The peak had a one-time cause: pandemic lockdowns and a fresh console launch sent gaming demand through the roof, and big PS4 hits like The Last of Us Part 2 were still selling into it [12]. What came after was the real picture — a slow drought of must-have exclusive releases. Part of that was a bet that didn’t pay off. Sony chased “live-service” games — titles built to keep selling to the same players for years rather than once at launch — and planned a dozen of them by 2025 [12]. Most never shipped. The online shooter Concord launched and was pulled two weeks later [12]. Cancellations and studio closures followed [12].

The connective tissue with the Game Pass story: in both cases, a company optimised for a recurring-revenue future — subscriptions, live service — and the durable demand underneath turned out to be thinner than the model assumed. “You need games to sell games,” as one outlet put it [12]. The recurring money still has to rest on something people actually want to keep paying for.

The market kept growing anyway

Here’s the number that didn’t deflate. The Entertainment Software Association’s annual report, out this week, found 212.3 million Americans — 67% of people aged 5 to 90 — now play video games at least an hour a week, up 7.2 million from last year [23]. Eighty-three percent of households played on some device in the past year [23]. Players average 12 hours a week [23].

That’s the quiet backdrop to every gloomy industry headline. The audience is bigger than it has ever been, even as studios cut staff, prices rise, and subscriber counts wobble. The business is straining not because people stopped playing, but because the ways companies tried to make money — ever-higher subscriptions, live-service bets, fewer-but-bigger releases — kept running ahead of what that growing audience would actually pay for. The demand is real. The question every studio is now re-asking is which of its numbers were demand, and which were just a price nobody had tested yet.

02 · Lesson · why it matters

The number was a price you never tested

A count of who's "with you" isn't a fact about loyalty — it's a measure taken at a price nobody has raised yet.

A confession on a stage

A Microsoft executive said something most companies hide. After Xbox raised the price of its Game Pass subscription by half, “millions of subscribers” walked out the door in a few months. Then Microsoft lowered the price again.

The interesting part isn’t that people left. It’s what their leaving revealed. Before the hike, that subscriber number sat in slide decks and earnings calls as a sign of how much people loved the service. After the hike, it turned out a large slice of that number was something else entirely: people who were barely there, paying out of habit, one small nudge away from gone.

The number hadn’t lied, exactly. It had just never been tested.

What a count actually measures

Here’s the trap, and it’s everywhere. A count of people feels like a fact about those people. Thirty-four million subscribers feels like thirty-four million fans.

But a subscriber number isn’t a measure of devotion. It’s a measure of how many people, at this price, in this month, decided staying was easier than leaving. That’s a much smaller, more fragile thing than love. It includes the loyal — and it includes everyone who simply hadn’t been given a reason to do the math yet.

Raise the price, and you force the math. Suddenly every half-hearted member has to actually decide. The ones who valued the thing stay. The ones who were coasting leave. The number drops — not because anything real was lost, but because a number that was never real got corrected.

The same pattern, twice this week

The games business showed this twice. While Xbox lost subscribers, Sony quietly revealed it was selling roughly half as many copies of its own games as it did five years ago. Its peak had a one-time cause — a pandemic and a new console launching into a captive, locked-down audience. That moment passed, and the bigger number went with it.

Both companies had a figure they’d treated as a baseline. Both found out the baseline was a condition, not a floor. The lockdown ends. The price goes up. And the number you’d quietly started counting on turns out to have been holding its breath the whole time.

You are counting the same way

This is easy to enjoy from a distance — clumsy corporations, soft numbers, serves them right. But the habit is yours too, and it runs deeper than spreadsheets.

You carry counts that feel like facts. The friends who’d show up if you needed them. The job that values you. The relationship that’s solid. Each of these feels like a fixed quantity — a number you can lean on. But none of them has been tested at a higher price lately. You haven’t asked the hard favour. Your employer hasn’t been offered a cheaper version of you. The thing felt durable because the cost of finding out stayed low.

That’s not cynicism. Plenty of what you count on is real, and will hold. The point is narrower and quieter: you can’t tell the real from the merely-untested by looking at the count. The number is the same either way. Devotion and habit wear the same face right up until the price changes — and then, all at once, you learn which was which.

What the figure can’t see from its seat

The executive on that stage could read the subscriber number to the decimal and still not know which members were loyal and which were drifting. The number doesn’t carry that information. You’d have to raise the price to find out — and by then it’s too late to keep them.

So the count sits there, precise and confident, hiding exactly the thing you most want to know. The companies got humbled this week not because they were foolish, but because every count works this way, and the only way to read it truly is to risk losing it.

The wiser move isn’t to distrust every number you have. It’s to hold the ones that matter a little more loosely — to remember that the headcount of who’s with you was taken at a price you’ve been lucky enough not to test, and to let that luck make you gentler with it, not more sure.

03 · Lab · your turn

The Price Test

Raise the price on your own subscriber base and watch which members were durable demand and which were only ever coasting habit.