The Live Service Era Didn’t End with a Bang. It Ended This Spring.

Four publishers. One quarter. Roughly $1.6 billion in live service bets quietly written off, cancelled, redesigned, or trimmed down to keep breathing. Sega killed an $882 million project that never had a name. Sony wrote down $765 million on a studio it bought in a panic four years ago. Marvel Snap fired its community team to survive. Project Ethos changed genres mid-development. None of them announced it dramatically. They buried it in financial slides, internal memos, and carefully worded statements about “renewed direction.” That’s how an era ends — not with a trailer, but with a budget line.

This isn’t a run of individual failures that happened to cluster in the same season. The live service model didn’t collapse because these specific games were bad, though some of them were. It collapsed because the structural economics that made the model look like a viable bet in 2019 stopped working around 2024, and every studio that started building into that window is now hitting the same wall at the same time. The math failed. The bills are coming in. Spring 2026 is just when the accounting caught up.

· · ·

§01 The Math That Stopped Working

The live service model was always a bet on one thing: that you could build a game, keep players inside it indefinitely, and extract recurring revenue from a playerbase that grew over time. That bet made sense when the market was expanding — when new players were coming online and the competition for attention was still manageable. The problem is that the market stopped expanding in any meaningful way for new live service challengers around 2023, and no studio building a game in 2021 was modeling for that.

Here’s the core number: 60% of gaming playtime in 2023 went to games six years old or older, per Newzoo’s annual industry report. Only 8% went to genuinely new titles. Five games alone — Fortnite, Roblox, League of Legends, Minecraft, and GTA V — accounted for 27% of all playtime that year. These aren’t just popular games. They’re entrenched economies with social graphs, unlockables, rank, history, and friend networks built up over years or decades. The cost to pull someone away isn’t just “make something fun.” It’s “be so obviously superior that players are willing to abandon years of investment in a game they already know.” Nobody launching a live service in 2025 was clearing that bar.

The player data confirms it. More than half of live service games released in 2025 lost over 90% of their players within months of launch, according to an analysis by The Gamer covering 19 major releases. Not 90% eventually — 90% fast. The market wasn’t waiting to be convinced. It simply didn’t show up.

This is the backdrop against which Sega, Sony, Second Dinner, and 2K’s 31st Union all walked into the same wall. Their games failed for different surface reasons. Underneath, it was always the same ceiling.

· · ·

§02 The Scoreboard

Sega’s $882 million ghost. In 2021, Sega announced the “Super Game” — a vague but ambitious live service initiative with a budget of roughly $882 million and a stated goal of going “beyond the traditional framework of games.” It never got a name. It never got a trailer. In May 2026, buried inside Sega’s FY2026 financial results — which also showed a $31.6 million net loss — the project was cancelled. The reason Sega gave: “intensifying market competition.” Over 100 developers have been reassigned from free-to-play development to “Full Game” teams working on Crazy Taxi, Jet Set Radio, Golden Axe, and Streets of Rage. Classic IP. Finished products. The exact opposite of the model they spent five years building toward.

Sony’s $765 million panic-buy hangover. When Microsoft announced the $70 billion Activision Blizzard acquisition in 2022, Sony panicked and bought Bungie for $3.6 billion. The logic was that Bungie knew how to make live service games. Four years later, Sony has written down $765 million on that acquisition — $204 million in Q2 when Destiny 2 missed performance targets after its concluding expansion, and $565 million in Q4, timed nearly exactly with the launch of Marathon. That game — Bungie’s new extraction shooter, built to reclaim the live service ground Destiny 2 was losing — is currently hovering between 10,000 and 15,000 peak concurrent players on Steam. Analyst Jacob Navok has said publicly that more write-downs are likely next quarter.

The Bungie deal was a reactive acquisition made under competitive pressure. Sony confused “studio that once made a successful live service game” with “studio that can make another one in a market that has fundamentally changed.” The $765 million is the cost of that confusion.

Marvel Snap’s survival cuts. Marvel Snap isn’t dead. It’s still a good card game with an active player base. But in late April 2026, developer Second Dinner confirmed layoffs across engineering, community, and QA — 16 staff total, including community manager Griffin Bennett, who came from Destiny 2 and Overwatch. Revenue fell from $175 million in 2024 to $150 million in 2025. Co-founder Ben Brode framed it as a survival measure:

“We’re still here, still building, still committed to this game and to you. This is us making hard decisions to make sure we can keep going, not a sign that we’re winding down.”

Ben Brode, Second Dinner — Kotaku, May 2026

Read that carefully. The community team is the operational tissue that connects a live service game’s roadmap to its actual player base. It’s how you keep people showing up. When you cut it to survive, you’re not preserving the game. You’re preserving a smaller version of it that can keep the lights on a little longer.

Project Ethos rewrites its own pitch. 2K’s studio 31st Union launched Project Ethos in 2024 as a free-to-play extraction hero shooter — the same lane that Concord tried to enter and died in 14 days after launch, and Highguard tried and shut down in 6 weeks in early 2026. In May 2026, with layoffs confirmed and development continuing on a smaller team, studio head Ben Brinkman reintroduced the game as “a bold new game with a renewed direction and vision — a skill-based PVP roguelike experience,” per PC Gamer. He added: “It’s become clear that changes need to be made.” Changing genres mid-development isn’t a creative pivot. It’s an admission that the original market thesis is no longer viable.

· · ·

§03 The Counterargument — and Why It Proves the Point

The obvious objection: what about the games that are actually working? Fortnite is still a cultural institution. GTA Online has been printing money for over a decade. Roblox reported 132 million daily active users in Q1 2026. EA Sports FC is a FIFA-branded licensing machine. League of Legends, Apex Legends, Minecraft — none of them are struggling.

That’s true. And it’s exactly the problem. Notice what those games have in common: the youngest of them, Apex Legends, launched in February 2019. Fortnite Battle Royale launched in September 2017. GTA Online has been running since 2013. All of them had years — in some cases over a decade — to build player bases, social graphs, and sunk cost before the current wave of challengers started shipping in 2022–2024. The live service model didn’t collapse for the incumbents. It collapsed for everyone trying to enter after the incumbents locked up the market.

This is the winner-take-all dynamic the industry kept refusing to model for. Live service games don’t just compete for market share — they compete for player identity. The more years someone has inside Fortnite, the harder it is to start over somewhere else. New entrants aren’t competing on quality; they’re competing against sunk cost. And sunk cost always wins.

The counterargument doesn’t rebut the thesis. It confirms it. The model’s apparent health is entirely held up by games established before 2018. That’s not a robust market — that’s a moat. And moats, by definition, are designed to keep new entrants out. Every studio that launched a live service in 2020–2022 was implicitly betting it could cross one. The data says that bet was nearly always wrong.

· · ·

§04 This Has Happened Before

The closest historical analogue is the 1983 North American video game crash. The Atari market had been growing fast, and publishers flooded it with low-quality titles — E.T. the Extra-Terrestrial being the famous example — that exploited a growing audience without delivering on the basic promise of a good game. Consumer trust collapsed. Retailers returned unsold inventory by the truckload. The entire category nearly didn’t recover.

What saved it was Nintendo’s licensing system — the Quality Seal of Approval — which forced a contraction to a smaller number of better products and killed the speculative low-quality tier. The result was the NES era, arguably the strongest sustained period of game quality the medium had seen up to that point. The crash didn’t kill gaming. It killed the part of gaming that was extracting value without creating it.

The current reckoning follows a similar shape. The speculative tier — the extraction shooters, the hero shooters, the free-to-play games chasing the Fortnite ceiling — is collapsing under its own weight. Sega is already moving 100+ developers back to “full game” development. Sony is leaning back into single-player exclusives like Saros. The GDC survey that found 1/3 of all AAA developers working on live service games now reads like a pre-crash inventory count.

What comes after a crash isn’t the death of the model. It’s a contraction to a smaller number of sustainable titles and a broad return to finished, premium games. That’s already happening. The question is just how many more $882 million slides get buried in financial results before the industry officially says it out loud.

· · ·

The $1.6 billion being written off, cancelled, and quietly reassigned isn’t really about bad games or bad luck. It’s about an industry that looked at the top of the live service market — at Fortnite and Roblox and GTA Online printing money — and spent the better part of a decade trying to build the next one, without ever seriously reckoning with whether there was room. There wasn’t. There probably hasn’t been since 2018.

What’s happening now isn’t a surprise correction. It’s the math finally being acknowledged out loud. The studios that built toward a market that doesn’t exist are now either pivoting, shrinking, or folding up their vague project slides and reassigning people to work on Streets of Rage 5. Which, if we’re being honest, is probably a better game anyway.

Sources: Kotaku — Sega Super Game cancellation · Push Square — Sony Bungie write-down · Insider Gaming — Marathon player data · Kotaku — Marvel Snap layoffs · PC Gamer — Project Ethos redesign · The Gamer — 2025 live service player retention · Gameranx — Jacob Navok on further write-downs