City of Heroes was the first great superhero MMORPG, and on November 30, 2012 NCsoft switched off its servers for good. Built by Cryptic Studios and launched in North America on April 28, 2004, the game let players design an original costumed hero from a deep, joyful character creator and then patrol a sprawling city — Paragon City — fighting crime alongside thousands of others. There was no Spider-Man, no Batman; the appeal was that every hero on screen was someone a player had invented. For eight years it ran a steady, sociable, mid-sized MMO, peaking at roughly 125,000 subscribers in 2008, and it cultivated one of the most tightly knit communities the genre has known.
The end came without warning. On August 31, 2012, NCsoft announced it was closing Paragon Studios — the in-house team it had created in 2009 to run the game — laying off around 80 staff and ceasing all billing the same day, “as part of the publisher’s efforts to realign its focus and publishing support.” The servers, NCsoft said, would go dark before year’s end. They did, at midnight on November 30, 2012. There had been no scandal, no security breach, no slow public decline: the game was simply on the wrong side of a Korean parent company’s portfolio review.
The players did not go quietly. A community-led campaign — “Save Paragon City,” coordinated largely through the Titan Network fan site — petitioned, fundraised, lobbied, and even pitched a buyout. Former insiders argued, with figures, that the game had been profitable, grossing a reported $12 million a year against roughly $4 million in costs; NCsoft disputed the claim and said the studio was unprofitable. None of it changed the verdict. On the final night, thousands of heroes gathered in Atlas Park to watch the lights go out together.
That, for seven years, was the whole story — a beloved world deliberately erased. Then in April 2019 it emerged that a small group of fans had quietly reverse-engineered and preserved the game’s server code, running it in secret as “SCORE” since not long after the shutdown. The leak detonated, fan servers sprang up overnight, and the largest of them, Homecoming, drew over 100,000 returning players. In an outcome almost no shuttered MMO ever gets, NCsoft eventually blessed it: on January 4, 2024 it granted Homecoming an official license. The verdict in 2012 was Shut Down. The afterlife is the rarest thing in this encyclopedia — a resurrection.
Star Wars Galaxies was the most ambitious thing anyone ever tried to do with the Star Wars license in a video game, and on December 15, 2011 Sony Online Entertainment switched it off. Launched in North America on June 26, 2003, developed by SOE and published by LucasArts, it was not a game about being Luke Skywalker. It was a sandbox: a player could be a moisture farmer, a cantina dancer, a doctor, an architect who designed buildings other players lived in, a crafter whose blasters everyone wanted, or a politician who got elected mayor of a player-built town. Combat was almost beside the point. The world was the point, and for a few years it was unlike anything else the genre had.
It also became the genre’s most infamous act of self-sabotage. In November 2005, two years in, SOE deployed the “New Game Enhancements” — the NGE — a top-to-bottom rewrite of the game’s core systems that collapsed a famously deep profession structure into nine classes, bolted on twitch-style shooter combat, and made the once-rare Jedi a starter option. It was pitched as making the game more accessible. What it did was take the game thousands of people had spent two years living in and replace it overnight with a different, shallower one, while those people were still standing in it. The base never forgave it, and the game never recovered.
By any honest accounting Galaxies was already a niche product before the NGE; the boxed-copy peak SOE touted in 2005 — a reported one million units — was a sales figure, not a live population, and unverified reports of the active subscriber base ran into the low tens of thousands. But the NGE turned a stable niche into a cautionary tale, and the eventual shutdown, when it came, was almost gentle by comparison: SOE and LucasArts said the license was expiring anyway, and a new Star Wars MMO — The Old Republic — was about to launch. The decision, SOE president John Smedley said, was “first and foremost a business decision.”
On the final night the holdouts gathered, as MMO communities do, to watch their world end. What they had lost was a place no other game offered — a living economy and a player-run society — that its own makers had partly demolished six years before the servers went dark. Fans have rebuilt it ever since, emulating the pre-NGE game from leaked code, which is its own quiet verdict on which version people actually wanted.
Concord was a $40 PvP hero shooter from Sony’s Firewalk Studios, and it was alive for fourteen days. It launched on PlayStation 5 and Windows on August 23, 2024; on September 3 Firewalk announced it would be taken offline; on September 6, 2024 the servers went dark and every buyer was offered a full refund. It is, by most reckonings, the second-shortest-lived major online game on record, and by budget the most expensive flop of the live-service era — a game that reportedly cost a sum somewhere between very large and unrepeatable, and recouped essentially none of it.
The numbers are the story, and they are brutal. Concord peaked at roughly 700 concurrent players on Steam at launch and was down to about 162 within a week; estimates put total sales across all platforms around 25,000 copies, a figure that would embarrass a small indie release, let alone a first-party Sony tentpole. The development timeline is contested — designer Jon Weisnewski described “around eight years” in the making, while Firewalk later clarified that full production began only in 2022 — but no one disputes that this was a long, expensive bet. A podcast host’s claim of a $400 million budget was loudly contested by industry figures who called the figure impossible, but even the skeptics agreed the true number was enormous. Whatever it was, Sony refunded it.
What killed Concord was not a bug, a scandal, or a server fire. It was indifference. The game launched into a hero-shooter market already saturated and largely free — Overwatch, Valorant, Apex Legends, Marvel Rivals on the horizon — and asked $40 up front for a competent but unremarkable take on a genre players already had, for free, in games they already played. The hero designs drew ridicule; the marketing left no cultural footprint; and at launch the audience simply did not show up. There was no community to mourn, because there was barely a community to begin with — the unusual mercy, and the unusual cruelty, of this particular death.
On October 29, 2024, Sony shut down Firewalk Studios entirely and cancelled any revival. Around 210 jobs were lost across Firewalk and a second studio caught in the same cuts. The hubris is fair game — eight years and a fortune to build a paid entry in a free-to-play genre nobody asked Sony to enter — but the people who lost their jobs are not. Concord is the cleanest cautionary tale the live-service era has produced: proof that scale, money, and a platform-holder’s full backing buy you nothing if the players decline to arrive.
Club Penguin was the virtual world where millions of children spent their childhoods, and on March 29, 2017 Disney switched it off. Launched on October 24, 2005 by a small Canadian studio, New Horizon Interactive, it was a snow-covered island populated by waddling cartoon penguins — each one a child’s avatar — who threw snowballs, decorated igloos, adopted Puffle pets, played minigames for virtual coins, and chatted in a deliberately locked-down environment built so parents could trust it. It was a massively multiplayer game for people too young for most MMOs, and it was enormous: by July 2013 it reported more than 200 million registered accounts.
Disney saw what it had and bought it. In August 2007 the company acquired Club Penguin and its studio for roughly $350 million up front, with up to another $350 million in performance-based earnouts that were never paid because the growth targets were not met. For a while Club Penguin was one of the crown jewels of Disney’s online strategy — a safe, subscription-funded, wildly popular kids’ world. But it was built on Adobe Flash, the web technology the entire industry was abandoning for mobile, and as children migrated to phones and apps, the island’s traffic eroded. Rather than rebuild the beloved thing, Disney decided to replace it.
The replacement was Club Penguin Island, a from-scratch mobile app that launched in March 2017 with none of the original’s items, coins, or memberships carried over. The original Club Penguin shut down on March 29, 2017; the mobile successor it was sacrificed for lasted barely twenty months before Disney closed it too, in December 2018, laying off staff. The franchise that had once cost Disney a third of a billion dollars was, by the end of 2018, entirely gone — the beloved version killed to chase a mobile reboot that itself promptly failed.
This is one of the entries where the encyclopedia’s dry wit has to step aside. Club Penguin was, for a generation now in their twenties, a first online home — the place they made their first internet friends, hosted their first parties in a pixel igloo, learned the etiquette of a shared digital space. When it closed, that place stopped existing. Fans rebuilt it on private servers, and those servers became their own cautionary tale, because some of them stripped out the very safety controls that had made the original worth trusting.
Tabula Rasa was the science-fiction MMO that Richard Garriott — the designer better known by his in-game persona “Lord British,” creator of the Ultima series — built for NCsoft, and on February 28, 2009 NCsoft switched its servers off, barely sixteen months after launch. Developed by Garriott’s studio Destination Games under NCsoft’s Austin operation, the game shipped to early-access pre-order customers on October 30, 2007 and to retail on November 2. It cast players as soldiers in a desperate war against an alien force called the Bane, blending squad-based shooting with the persistence and progression of a massively multiplayer world. On a roughly $25 million budget and a team that reached nearly 150 people, it was a serious, expensive bet by one of the most decorated names in the medium.
It did not pay off. Subscriptions came in well below the numbers a game of that cost needed, and NCsoft never had reason to publish figures it would only have wanted to hide. On November 22, 2008 — barely a year in — the company sent players an open letter announcing that the game would end public service on February 28, 2009, citing a lower-than-expected population as the deciding factor. To cushion the closure, NCsoft waived subscription fees from January 10, 2009 to the end and handed subscribers three free months on other NCsoft titles plus beta access to its incoming MMO, Aion. The lights went out on schedule.
What lifted the story above the ordinary live-service failure was the human drama braided through it. Garriott had spent much of October 2008 not at his desk but in orbit, having paid a reported $30 million to fly to the International Space Station as a private astronaut — the second-generation spacefarer following his astronaut father. On November 11, 2008, while he was in post-flight quarantine, an open letter went out announcing he had left NCsoft “to pursue other ventures.” Garriott later said NCsoft wrote that letter to force him out, and he sued.
He won. In July 2010 an Austin jury found NCsoft had breached his stock-option agreement by treating his exit as voluntary — which collapsed his option window from ten years to ninety days — and awarded him $28 million. NCsoft appealed; the Fifth Circuit affirmed in October 2011, and with interest and fees the judgment grew to roughly $32 million. Tabula Rasa, the game, was a write-off. Tabula Rasa, the legal fight, made its creator a fortune.
Glitch was a whimsical, non-combat browser MMO built in Adobe Flash by Tiny Speck, and on December 9, 2012 the studio switched it off — a little over a year after launch. Created under the lead design of Stewart Butterfield, the Flickr co-founder, Glitch was a deliberately strange and gentle thing: a 2D world set inside the imaginations of eleven sleeping cosmic giants, where there was no combat and no killing, only collaborative crafting, gathering, exploring, and a steady drip of surreal, generous humour. It launched on September 27, 2011, reverted to beta two months later to keep iterating, and won a small, fiercely devoted following charmed by a game that asked players to be kind rather than to win.
It could not find a mass audience, and the reasons were structural. Tiny Speck had raised roughly $17 million from blue-chip investors including Accel and Andreessen Horowitz, and the game’s free-to-play economics demanded a large player base — the team has been described as needing around 200,000 players to reach profitability. Glitch never got there. Its non-combat, slow-burning, intentionally peculiar design was a hard sell to the mainstream, and it ran on Adobe Flash precisely as the audience migrated to iPhones and Android devices that Flash served poorly or not at all. A charming game on a dying platform aimed at a niche taste is a difficult business to scale.
On November 14, 2012, Tiny Speck announced Glitch would close, noting it had not found a buyer willing to keep it running, and the world went dark on December 9. By Butterfield’s own account the closure was wrenching — he could not get through sixty seconds of telling his staff the game was over without breaking down. The studio was left with several million dollars of its funding unspent.
What it did with that money is why Glitch is remembered at all. To build a game with a distributed team, Tiny Speck had created an internal communication tool — a searchable, integrated replacement for the IRC chat they used to coordinate. When Glitch died, that tool lived. The team turned full-time to it in early 2013, named it Slack, and built it into one of the defining workplace platforms of the decade — a company Salesforce later acquired for about $27.7 billion. The game was a failure. The byproduct conquered the world.
WildStar was a stylish, deliberately hardcore science-fantasy MMO from Carbine Studios and NCsoft, and on November 28, 2018 NCsoft shut its servers down after four and a half years. It launched on June 3, 2014 with a monthly subscription, a vivid cartoon-Western-meets-space art style, sharp comic writing, and an explicit pitch to lapsed “hardcore” players who missed the punishing rigour of early World of Warcraft. It had real buzz — Metacritic settled around 82, the housing system was widely praised, and the developers, a studio founded by ex-Blizzard WoW veterans, leaned hard into 40-player raids and demanding attunement quests as the marrow of the endgame. For a few weeks, WildStar looked like the genre’s most confident new entrant in years.
The buzz did not survive contact with the launch. The early game was rough — a busy, clunky interface, performance problems, lingering bugs, and signature systems like the “Path” feature that weren’t fully realized at release. More fundamentally, the endgame that was the whole selling proposition was punishingly difficult: 40-person raids with long, gated attunement requirements and unforgiving mechanics that demanded enormous coordination and time. That was catnip to a small hardcore minority and a wall for everyone else, and the everyone-else left fast. Reports point to a peak in the low hundreds of thousands of players that drained steeply within months; first-year revenue has been reported around $33 million, collapsing toward roughly $5 million by 2016.
NCsoft and Carbine did what struggling subscription MMOs of the era did: on September 29, 2015 WildStar went free-to-play, hoping a lower barrier would refill the world. It bought time, not a future. The studio had already cut staff in 2016, server merges signalled a thinning population, and the underlying problem — a subscription-first, brutally hardcore game launched into a free-to-play, mass-market era — never resolved.
The end came abruptly. On September 6, 2018, NCsoft announced it was closing Carbine Studios immediately and winding WildStar down, after rejecting two new project pitches from the team; around 50 employees lost their jobs. A shutdown date was set for November 28, 2018, with refunds promised on purchases made after July 1. Carbine’s farewell thanked the players “who made planet Nexus such a special place of the last four and a half years,” and then planet Nexus, like the studio that built it, was gone.
LEGO Universe was the official LEGO massively multiplayer online game, and on January 31, 2012 — barely fifteen months after launch — The LEGO Group switched it off. Built by the Denver studio NetDevil and published by LEGO, with Warner Bros. Interactive handling global distribution, it launched on October 26, 2010 (after an early-access “Founders” window beginning October 8). It did the obvious, lovely thing: it let players gather virtual bricks, build models on their own properties, and explore a colorful online world together. For a brand whose entire promise is creative freedom, an online sandbox of unlimited bricks looked less like a product than a destiny.
The game ran on a subscription — 9.99 US dollars a month, with discounts for six- and twelve-month commitments — and the subscriptions never came in the numbers LEGO needed. In the summer of 2011 the company added a free-to-play zone covering the first two worlds, hoping a no-cost on-ramp would convert players to paying members. It did not move the needle far enough. On November 4, 2011, LEGO announced the closure; the VP of LEGO Universe stated plainly, “Unfortunately, we have not been able to build a satisfactory revenue model in our target group, and therefore, have decided to close the game.”
That is the official cause, and it is true. But underneath the revenue line sat a structural problem that has since become one of the most-cited cautionary tales in online-game design: the cost of moderation. Because LEGO Universe let children build anything from bricks, children built exactly what one might fear, and LEGO — a brand parents trust absolutely — could not let those creations appear in public unscreened. The result was a moderation regime so expensive that, by the account of a former developer, human moderators were “the single biggest cost center” of the entire operation, or close to it. A creative MMO drowned in the cost of policing its own creativity.
When the servers went dark on January 31, 2012, players lost their properties, their builds, and the worlds they had explored, with nothing to carry away. Years later, a fan project called Darkflame Universe reverse-engineered the server and brought the game back unofficially — a small, characteristic afterlife for a deliberately closed online world.
Google Stadia was Google’s cloud-gaming platform — play console-quality games streamed over the internet, with no console to buy — and on January 18, 2023 Google turned it off. It launched publicly on November 19, 2019, on the promise that the most expensive part of gaming, the hardware, could simply disappear: the data center would run the game, your screen would show it, and a 129-dollar Founder’s Edition with a controller and a Chromecast was all you needed to get started. The technology mostly worked. Almost nothing else did.
Stadia arrived with a thin launch library of 22 games, a confusing structure in which a Stadia Pro subscription gave you some titles while you still bought most others at full price, and a market that had heard Google make promises before. The trust problem was not incidental — it was the headline risk, and Google validated it almost immediately. In February 2021, barely fifteen months after launch, Google shut down Stadia Games and Entertainment, the in-house studios meant to produce the exclusives that would give the platform a reason to exist, laying off roughly 150 people including its high-profile head, Jade Raymond, before either studio had shipped a game. A platform sold on long-term commitment had visibly stopped committing to itself.
From there the outcome was a formality that took two more years to arrive. On September 29, 2022, Google announced Stadia would close. Phil Harrison, the executive who had championed it, wrote that while the underlying technology was sound, the service “hasn’t gained the traction with users that we expected.” The servers went off at 11:59 PM Pacific on January 18, 2023.
What distinguished Stadia’s death was the exit. Google refunded essentially everything — all the hardware bought from the Google Store, and all the games and add-ons bought from the Stadia store (Stadia Pro subscription fees excepted), automatically, to the original payment method. It was, by the standards of dead game platforms, an unusually clean and generous funeral. It was also, unmistakably, the Killed-by-Google pattern playing out exactly as the skeptics had warned, and the refunds were less a surprise than a confirmation: the product was so plainly Google’s to switch off whenever it lost interest that giving the money back was the only graceful move left.
The Sims Online was Electronic Arts and Maxis’s attempt to turn the best-selling PC game of its era into a persistent online world, and on August 1, 2008 EA switched it off. It launched on December 17, 2002, riding the staggering success of the original The Sims, with the reasonable-sounding theory that if millions of people loved managing a virtual household alone, they would love doing it together. The pre-launch hype was enormous — a Newsweek cover, breathless press about a new kind of online society, and TIME’s Lev Grossman calling it a “daring collective social experiment.” Will Wright, the original game’s celebrated designer, spoke of building a sprawling real-time social drama.
The drama did not materialize. The Sims Online reached roughly 105,000 subscribers in 2003 — a respectable MMO number, but a catastrophic one against the expectations EA had set, which reportedly ran to 200,000 subscribers by spring 2003 and 400,000 by year’s end. The trouble was that the specific magic of The Sims did not survive the trip online. The single-player game’s pleasures — directing a dollhouse, fast-forwarding through chores, telling your own little stories with total control — depend on the world bending to one player. Online, time runs for everyone at once, other people don’t behave like cooperative furniture, and the celebrated “life simulation” collapsed into grinding repetitive jobs to earn currency. The thing players loved about The Sims was precisely the thing an MMO could not provide.
EA tried to rescue it. In early 2008 it relaunched the game under a new name, EA-Land, with an expanded world and new features, hoping a reset would breathe life back in. The reset failed almost immediately. On April 30, 2008, barely two months after the EA-Land relaunch, EA announced the whole thing would close, and on August 1, 2008 the servers went dark.
The players who remained — a small, loyal community that had built years of in-game homes, relationships, and economies — lost all of it; EA offered modest compensation in the form of store credit and a Pogo membership. The episode left a clear lesson that the industry kept relearning: a beloved single-player experience is not a multiplayer experience minus the loneliness, and porting one to the other can subtract the very magic it was meant to scale.
Warhammer Online: Age of Reckoning was Mythic Entertainment and EA’s big realm-versus-realm MMO, the game meant to take a serious bite out of World of Warcraft, and on December 18, 2013 its servers were switched off after the Games Workshop license expired. It launched on Windows on September 18, 2008, built around a single, genuinely distinctive idea: the whole game was a war. Two factions — Order and Destruction, split across three racial pairings of Dwarfs against Greenskins, Empire against Chaos, and High Elves against Dark Elves — fought a persistent, structured conflict from the lowest levels up to citywide sieges. Realm-versus-realm combat was the point, not an endgame afterthought, and for a moment it looked like the formula that might finally crack WoW’s dominance.
The launch was, by the numbers, a hit. The game sold over a million copies, and by October 10, 2008 EA could report that 750,000 people were playing; the subscriber base is generally cited as peaking around 800,000. Then the floor gave way. Within a few months the population had collapsed to roughly 300,000, and it kept sliding. The game that was supposed to be a WoW-killer instead became a textbook study in how fast a strong launch can bleed out when players sample it, drift back to the incumbent, and never return.
Mythic spent years trying to arrest the decline — consolidating servers, reworking systems, and at one point planning a free-to-play conversion to widen the funnel. None of it reversed the slide, and the free-to-play plan was ultimately cancelled. The decisive constraint, in the end, was contractual rather than purely commercial: Mythic’s right to operate a Warhammer game came from a license with Games Workshop, and that license was coming to an end. Rather than renew for a game whose population had long since thinned, EA let it lapse.
On December 18, 2013, after a little over five years, Warhammer Online went dark. As with several worlds in this archive, the players did not entirely accept the verdict: a fan-run private server, Return of Reckoning, reconstructed the game and has kept the war going for over a decade since. The official servers are gone; the realm war, improbably, is not.
The Matrix Online was the MMO that officially continued the story of the Matrix films, and on July 31, 2009 Sony Online Entertainment switched off its servers, ending a game whose lore mattered far more than its population ever did. It launched in the United States on March 22, 2005 (Europe followed on April 15), developed by Monolith Productions and built on a remarkable premise blessed by the Wachowskis: the events of the trilogy were over, and players would inherit and carry forward the canonical storyline. What happened in the game was, the creators said, real Matrix canon — a level of narrative authority almost no licensed game has been granted before or since.
It used that authority boldly. Live, GM-driven events advanced an ongoing plot, and in one of the most striking storytelling decisions in MMO history, the game canonically killed off Morpheus — Laurence Fishburne’s character — in an in-game event, after he began detonating “code bombs” to expose the Matrix and was hunted down and assassinated. This was not a side-quest; it was the franchise’s official next chapter, unfolding live on Monolith’s and later Sony’s servers. The Matrix Online was, for the small number of people inside it, the most genuinely consequential MMO around.
The trouble was the size of that number. The game’s commercial life was rocky from the start. Within months of launch the publishing arrangement was overhauled: Warner Bros. and Sega, the original publishers, handed the game to Sony Online Entertainment, with operations transferring to SOE on August 15, 2005 — barely five months after release. Under SOE the game was consolidated and kept on life support, but it never attracted a mass audience. By the time the shutdown was announced in 2009, fewer than 500 active players remained.
On July 31, 2009 SOE closed it for low subscription numbers, and with the servers went the only place that official Matrix canon lived. The films’ story had a sequel, and almost nobody saw it. The Matrix Online is filed here as a particular kind of death — not a beloved giant cut down, but a richly imagined world that was canonically important and commercially negligible at the same time.
Battleborn was Gearbox’s ambitious attempt to fuse the hero shooter with the MOBA, and on January 31, 2021 its servers were switched off, taking the single-player campaign down with them. Developed by Gearbox Software — the studio behind Borderlands — and published by 2K, it launched worldwide on May 3, 2016 for PlayStation 4, Windows, and Xbox One. It offered 25 wildly distinct playable heroes, a story campaign that could be played cooperatively, and competitive multiplayer modes that blended lane-pushing strategy with first-person gunplay. It was loud, colorful, mechanically dense, and full of personality. It was also, with hindsight, about three weeks too early to a party that someone else was about to own.
The game opened reasonably: over 12,000 concurrent players on PC at launch, fourth best-selling title in the United States for May 2016 per the NPD Group, and an estimated $18 million in digital revenue that month according to analytics firm SuperData. Then, on May 24, 2016 — twenty-one days after Battleborn shipped — Blizzard released Overwatch. Overwatch was cleaner, more accessible, more beautiful, and backed by Blizzard’s marketing might, and it found over seven million players in its first week. SuperData pegged its May digital take at roughly $269 million against Battleborn’s $18 million. The hero-shooter conversation, such as it was, had a winner, and Battleborn was not it.
The collapse was swift. By July 2016 — barely two months after launch — concurrent PC players had fallen below 1,000, a more-than-tenfold drop. Gearbox kept updating the game, added free content, and in June 2017 introduced a “Free Trial” mode that effectively turned Battleborn into a free-to-play title, rotating heroes available to anyone. None of it reversed the tide. In September 2017 Gearbox announced that the final update, released October 23, 2017, would mark the move into maintenance mode: no more new content. The wind-down followed a long, telegraphed schedule — pulled from sale in November 2019, in-game purchases disabled by February 2020, servers off January 31, 2021.
What players lost was real but modest: an inventive, overstuffed game that never got a fair hearing, and a single-player campaign that — because the game was always-online — became unplayable the moment the servers died. Battleborn is the rare entry here that failed not on its merits but on its calendar: a decent game with one fatal flaw it could do nothing about. It was not Overwatch, and it arrived first.