Part IThe Story
The Last Copy
In November 2013, a package left a house in Potomac, Maryland, addressed to a game designer named Zack Hiwiller. It contained ZZT, a game drawn entirely from the 255 characters of the IBM PC's character set. Its hero was a white smiley face on a navy blue rectangle. Its lions, tigers, and bears were letters and symbols, its sound effects the clicks, pops, and snaps of a PC speaker. The man who filled the order was Paul Sweeney, who had spent his working life at the Defense Mapping Agency turning satellite imagery into maps, and who had been fulfilling ZZT orders from that address since January 1991. The game had been released as freeware on October 10, 1997. People kept sending money for it anyway, for sixteen more years. That November, Paul Sweeney shipped the last one.
The address had once been the headquarters of a company called Potomac Computer Systems. It was really a bedroom. The company's founder, Paul's youngest son, had been a mechanical engineering student at the University of Maryland who drove home to use his computer and decided, somewhere between a text editor and a business plan, that the editor was the game. By the time the last box went out, Potomac Computer Systems had been renamed twice and moved south to Cary, North Carolina, and it was in the middle of the most dangerous reinvention of its existence.
Consider what Epic Games looked like that month. Seventeen months earlier, in June 2012, Tim Sweeney had sold roughly 40 percent of the company to Tencent, a Chinese internet conglomerate, for $330 million. His design director of twenty years was gone. So were his president, his top Gears of War producer, the founder of the Polish studio he had just bought, and five senior people who left together to start something smaller. In that same November, Epic renamed that Polish studio Epic Games Poland and pointed it at a game called Fortnite, which had been announced two years earlier and was nowhere near finished. Two months later, Epic would sell Gears of War, its signature franchise, to Microsoft.
So here was the arrangement. The company's oldest product was being mailed by hand, one copy at a time, by a man in his seventies or eighties from a suburban address in Maryland. Meanwhile the company itself was being rebuilt with Chinese capital around a business model in which games are never shipped at all. They run. They update. They take a small payment, then another, indefinitely. The last physical artifact of Epic's first life went out the door just as Epic bet everything on a life with no artifacts.
Sweeney has a habit of numbering his company like software. He called the console era "Epic 3.0" and the Tencent era "Epic 4.0." Each version, as it happens, was a negotiation with the same problem: who owns the road between the person who makes a thing and the person who wants it. A father with a mailbox. A shareware catalog. A bulletin board. A publisher. A console maker. A storefront taking 30 percent. A phone with one store on it. Sweeney has spent thirty-five years trying to get rid of the middleman, becoming a very large middleman himself, and then suing the larger ones for doing what middlemen do.
He is also, by most measures, the largest private conservation landowner in North Carolina. He buys forests so that nobody can build roads through them.
By the Numbers
The Sweeney Ledger
~20,000Lines of Pascal in ZZT, written in under 1,000 hours
$100/dayZZT's take by November 1991, enough to choose a career
$330MTencent's 2012 price for roughly 40% of Epic
$32BEpic's valuation after the April 2022 Sony and Kirkbi rounds
41.4%Sweeney's stake, still controlling, after a decade of dilution
12%The Epic Games Store's revenue cut, against an industry-standard 30%
50,000Acres of forest Sweeney has acquired for conservation
The Whole Car
The Potomac of Sweeney's childhood was, in the words of the Academy of Interactive Arts & Sciences citation that inducted him into its Hall of Fame, "an idyllic part of rural Maryland, swallowed all-too-soon by urban sprawl from Washington, D.C." He was born in 1970, the youngest of three brothers. His mother raised the boys and worked in a flower shop. His father made maps for a government agency that did not discuss what it mapped. "My generation had massive individual freedom," Sweeney has said. "The mothers of the neighborhood would give us breakfast and say, 'Be back by dark.' We would go and roam. You might find us 10 miles away on our bikes exploring some woods."
He took apart a lawnmower when he was five or six. Later he built go-karts and drove them around the neighborhood. The decisive influence was his oldest brother, Steve, sixteen years his senior, an electronics enthusiast who went into the hardware business and always had "massive amounts of old and scrapped computing hardware lying around." The two of them went to junk auctions and bought computers from the 1950s, '60s, and '70s, then set up a basement full of these dead machines and tried to build things out of them. It is a strange image: a boy in the Reagan years presiding over a museum of obsolete mainframes, not out of nostalgia but out of something closer to scavenging. He wanted to know how a thing worked, and he did not much care whether anyone still used it.
Steve moved to California to work in the early computer industry. "I saw it as this glamorous lifestyle," Sweeney recalled, "and from that, you can tell what a geek I am." At eleven (most tellings say eleven, one says thirteen, and the age drifts the way founding myths do) he spent a week at his brother's startup, sat down at an early IBM PC, and learned BASIC. "From that point on, I was really addicted to programming." The family had owned a Commodore 64 and an Atari 2600, and the Atari's Adventure was the only console game that held him. When the Sweeneys got an Apple II for Christmas, he began trying to write Adventure 2.
He estimates that between eleven and fifteen he logged more than 10,000 hours teaching himself to program from bulletin boards and trial and error. He finished several games and shared none of them. When he did show his work, it was to neighborhood kids, whom he watched like a researcher. "That really informed me a lot about how to build a game that makes sense." He played other people's games, the Hall of Fame citation says, "as a scientist, rather than a gamer," and he still says so himself, with a kind of cheerful indifference to the medium that made him a billionaire: "I was never really a serious gamer — I just played enough of these different games to understand what they did, reverse-engineer their techniques, and learn from them." He has said he has finished very few games in his life.
The money for equipment came from lawns. He offered to mow the yards of wealthy neighbors for half of what professional services charged. It was a fourteen-year-old's discount strategy, undercutting incumbents on price while doing the labor himself, and it would surface again at much larger scale. "I've always really looked at cash-flow," he said decades later, "as a key driver of being able to develop cool things."
— Tim Sweeney, to Develop magazine, 2013I very quickly realised that a computer could do absolutely anything, as long as I could figure out how to issue the right commands to get it to do what you wanted.
He entered the University of Maryland around 1989 to study mechanical engineering, which he took seriously and then quietly began to resent. The resentment was specific. He realized a working engineer would specialize, and that rather than building an entire car, he would build a tiny part of one. Programming offered the opposite. "I could write a program from beginning to end, and it could be something really complicated." Mechanical things, he told Develop, "were always so hard to fix. It was different with computers."
That preference for authoring the whole car over machining the bracket explains a great deal of what followed. Owning the stack, the engine, the editor, the store, the payment rail, all trace back to it. He did not escape the car, either. In 2011 Stephen Totilo of Kotaku reported that Sweeney had owned several Ferraris and believed he had been born five years too late to become an auto mechanic. When Sweeney retires a cell phone, he takes it apart and looks at the components "and see[s] what they're doing." The engineering degree, in the end, went unfinished. By his own account he left one credit short, though the university's news office now lists him, generously, as a member of the class of '93.
The coursework stuck. "During my studies I really didn't understand why I was learning vector calculus and these other esoteric things," he said. "But when I started to write 3D rendering code, it turned out to be the secret sauce."
A Game Made Out of a Text Editor
ZZT began as a complaint. Sweeney disliked the editors that came with the programming languages available on his PC, a 286 with an IBM Model M keyboard that clacked like a typewriter. In 1989 he set out in Turbo Pascal to rebuild a programming environment like the one he'd made on the Apple II. He did not yet know how to program graphics. What he had was a screen of text characters and a blinking cursor.
He gave the characters collision. He made the cursor something you could steer. He added bullets, then creatures to shoot them at, and noticed that this was more fun than the editor. He began building levels as text files, with boards that connected to other boards in the manner of Adventure. Because everything was text anyway, he could put in a talking tree without breaking the illusion. He took the bomb from a game called Kroz; nearly everything else was his. The whole thing came to roughly 20,000 lines of Pascal, built over six to nine months of nights and weekends, in under a thousand hours, while he attended engineering classes by day.
The defining feature was something most players never needed and some played for years. ZZT shipped with its own editor and a scripting language, ZZT-OOP, so the tool Sweeney had used to make the game was in the box. Anyone could make boards, script objects to talk or shoot or flash messages, and build whole worlds. He marketed it as the first major game with object-oriented programming. He named it ZZT so it would appear last in alphabetical shareware listings, which is either a joke or a very precise piece of distribution thinking, and probably both. A fan later proposed the backronym "Zoo of Zero Tolerance," and Sweeney endorsed it.
He wrote to Scott Miller at Apogee Software, then the dominant shareware publisher, asking how distribution worked. Miller replied with advice and encouragement. ZZT went out on January 15, 1991, using Apogee's episodic model: Town of ZZT and the editor were free to copy and spread across bulletin boards, and the other worlds came by mail on a single floppy with hand-drawn maps. Sweeney revived the name of a computer consulting business he'd tried and abandoned, Potomac Computer Systems, and had orders sent to his parents' address. By November 1991 he had sold about 800 copies, three or four a day, bringing in roughly $100 a day.
A hundred dollars a day was the whole argument. It meant a twenty-year-old could earn more selling something he made than working for a wage, and that settled it. By May 1999 ZZT had earned him about $30,000, most of it in the first year.
The more consequential discovery was what the users did. Shortly after release Sweeney ran a contest for registered players to submit their own worlds, and more than 200 people did. The best were packaged and sold as The Best of ZZT and ZZT's Revenge in 1992, and the six designers behind the winning worlds of ZZT's Revenge got jobs at his company. Sweeney noticed that the community's work was matching or exceeding his own. Most game makers would take that as an insult. He took it as a business model.
The community kept going long after he left it. By 2021 more than 3,000 ZZT worlds existed. Fans built third-party editors and source ports for the Dreamcast and the Nintendo DS, and scholars have written about the game as an outlet for people, transgender players among them, who wanted to make something without knowing how to draw or code. Anna Anthropy wrote a book-length study of it for Boss Fight Books. Sweeney has claimed that tens of thousands of game-industry workers once made ZZT worlds. That's probably unverifiable and probably not wrong.
The company needed a bigger name. Apogee and id Software dominated shareware, and "Potomac Computer Systems" sounded like an office that fixed printers. So, within the year, it became Epic MegaGames. "Epic" and "Mega" were meant to make it sound large. "It was kind of a scam to make it look like we were a big company," Sweeney told Benj Edwards in 2009, over lunch. "Of course, it was just one guy working from his parents' house."
His next game, Jill of the Jungle, needed skills he lacked, so he assembled a team of four and finished it by mid-1992. Then he went looking for a partner who could do what he couldn't.
Mark Rein had just left id Software, the Texas studio of Wolfenstein and soon Doom, and moved to Toronto. He never relocated for Sweeney in the early years. He worked remotely, handling sales, marketing, and the publishing deals that Sweeney credits with much of the company's growth. Where Sweeney was soft-spoken and technical, Rein was the opposite: a frontman, "infectiously positive about games, and not afraid to speak his mind," as Develop put it. He has been vice president for more than three decades and owns about 4 percent of the company. He is the one who says out loud what Sweeney is: "People think of him as a technical genius but he is a business genius as well."
Around the same time, a teenager named Cliff Bleszinski sent Sweeney a game he had made, Dare to Dream. He was seventeen. Sweeney hired him. Twenty years later he would be the most recognizable designer in American console games, the face of Gears of War, demoing it from Grauman's Chinese Theatre at Microsoft's E3 press briefing in 2006. He was a user who became a builder, the ZZT contest in human form.
By late 1992 Epic MegaGames had twenty people: programmers, artists, designers, composers. By 1993 it had more than thirty, and by 1997 about fifty, scattered across the world and connected by modem, publishing Jazz Jackrabbit, Epic Pinball, One Must Fall: 2097, Tyrian. The founder, by his own account, "stepped back for a while." He thought he would never be a competitive programmer again.
The Engine Nobody Ordered
He was wrong about that, and the error built the company.
To ship Unreal, a 3D first-person shooter co-developed with Digital Extremes and released in 1998, someone had to build the technology underneath it. Sweeney did. "I thought I'd never be a competitive programmer again," he told Gamasutra in 2017, "but got back into it to ship Unreal. I had to build the engine for it, which sort of set the stage for everything we're doing now." He wrote roughly a quarter-million lines of the original Unreal Engine, by his own estimate about 80 percent of it, along with its scripting language, UnrealScript. The editor was built into the engine, as in ZZT. Players could reshape the game with the same tools its makers used.
The engine looked so good that a magazine put screenshots of an Unreal-powered game on its cover before release. Then the phone started ringing. Other developers wanted to license what they'd seen. Epic had not set out to sell an engine; customers told it that it had one. The Unreal Engine business, which would win a Guinness World Record in 2014 as the "most successful videogame engine," a BAFTA Special Award in 2019, and an Engineering Excellence Award from the Hollywood Professional Association, began as a side effect of a founder going back to writing code because nobody else could.
In February 1999, Epic MegaGames moved its headquarters to Cary, North Carolina, and became Epic Games. Rein's explanation was operational: "Unreal was first created by developers who were scattered across the world, eventually, the team came together to finish the game and that's when the real magic started. The move to North Carolina centralizes Epic, bringing all of the company's talented developers under one roof." Sweeney's explanation was briefer and more revealing. They dropped "Mega" because they no longer wanted to pretend to be a big company. The scam had become unnecessary, because it was now true.
Unreal Tournament shipped later that year to critical acclaim, a multiplayer game with open modding tools. Rein later called it ZZT's spiritual successor, and Epic increasingly treated mods as a pipeline for talent. In 2004 it launched the Make Something Unreal contest. A small team called Tripwire Interactive won $80,000 in cash and hardware over the first contest and turned a mod into a company. A studio called Psyonix, founded a few miles from Epic's offices, contributed work to Unreal Tournament; fifteen years later, Epic bought it, along with Rocket League. The shareware contest of 1991 had become a farm system.
Licensing an engine meant living with licensees, and not all of them were grateful. On July 19, 2007, the Canadian studio Silicon Knights sued Epic, claiming it had failed to "provide a working game engine." The suit said Epic had been "sabotaging" Unreal Engine 3 licensees by holding back improvements as game-specific and spending licensing fees on its own titles. Epic countersued. On May 30, 2012, it won $4.45 million for copyright infringement, misappropriation of trade secrets, and breach of contract, and the award later doubled with interest and fees. Judge James C. Dever III found that Silicon Knights had "deliberately and repeatedly copied thousands of lines of Epic Games' copyrighted code," and that it had copied not only functional code but "non-functional, internal comments Epic Games' programmers had left for themselves."
That detail stays with you. Somewhere in a Canadian studio's engine were notes written by Epic programmers to other Epic programmers: reminders, apologies, jokes in the margins. The court ordered Silicon Knights to destroy every line derived from Unreal Engine 3 and to recall and destroy unsold copies of Too Human and X-Men Destiny. The studio filed for bankruptcy on May 16, 2014. Sweeney's company could be generous with its tools. It was not generous with its code.
Bleszinski, years later, summarized Sweeney's contribution in five words: thanks to him, "graphics finally don't suck." He meant something more commercial too. Sweeney, he said, made an engine that let developers "make good-looking things fast," which produced what Bleszinski called "system justifiers," the demo you put on stage when you want people to buy a new phone or console. By the 2010s, if Apple or Sony or anyone else wanted to prove its hardware was powerful, the easiest proof was the latest Unreal Engine running on it, with someone from Epic on stage. That relationship would matter a great deal when Sweeney decided to make enemies of the people whose stages he stood on.
Version Numbers
Around 2006, PC gaming was being hollowed out by piracy, and single-player PC games, the core of Epic's business, were becoming hard to make pay. Sweeney moved the company to consoles and called it "Epic 3.0."
It worked at once. Gears of War, released for the Xbox 360 in 2006, cost about $12 million to make and grossed about $100 million. Gears of War 2 sold more than three million copies in its first month in 2008. Gears of War 3 followed in 2011, and the prequel Judgment in 2013. Unreal Engine 3 sat underneath hundreds of PC, Xbox 360, and PlayStation 3 games, including Mass Effect and BioShock. On September 1, 2010, Epic released Epic Citadel, a tech demo showing its engine running on an iPhone, and on December 9 of that year it shipped Infinity Blade for iOS, a sword-fighting game that showed console-quality graphics in someone's pocket. Apple was delighted. This is worth remembering.
Underneath the success was a trap. A fourth main Gears game, Epic estimated, would cost at least $100 million, a number Polygon later reported could have killed the company. Epic proposed a multiplayer-only Gears, continuously improved from player feedback, the Unreal Tournament idea in console form. Microsoft, which owned the platform and published the series, said no. Epic, as the company's own history puts it, "recognized the troubles of being held to the business objectives of a publisher." Sweeney was blunter about the economics. "We're looking at companies like Activision spending $100 million developing each new version of Call of Duty," he told Game Developer in 2013, "and that's insane! We can't afford that sort of budget."
His diagnosis came in a single sentence: "There was an increasing realization that the old model wasn't working anymore and that the new model was looking increasingly like the way to go." The new model was games as a service: free or cheap to enter, monetized over years, never really finished. The company that knew how to run that model at scale was Tencent, whose portfolio included Riot Games' League of Legends. In June 2012, Tencent paid $330 million for about 48.4 percent of Epic's then-issued share capital, equivalent to 40 percent of the total company including employee options. The deal valued Epic at roughly $825 million. Tencent got the right to nominate directors. Sweeney said it got very little control over what Epic made. He called it "Epic 4.0." The company had about 200 employees.
Over the following months, a good many of the people who had built Epic 3.0 left.
Rod Fergusson had been the producer who got Gears of War out the door. He was a fixer, the person you bring in when a big game is late. He looked at where the Tencent deal pointed and did not want to go there. He wanted, he said, to keep making "a AAA, big-narrative, big-story, big-impact game," and he left in August 2012. He went to Irrational Games to help finish BioShock Infinite, and while there started the conversations among 2K, Epic, and Microsoft that ended with Microsoft buying the Gears of War rights outright on January 27, 2014. Fergusson took over the series at a Microsoft studio renamed The Coalition, which shipped Gears of War 4 in October 2016. The franchise left with the man who wanted to keep finishing things.
Cliff Bleszinski had been at Epic for twenty years, since he was seventeen. He said he had grown "jaded" about the industry in the period before Tencent arrived. He tried to renegotiate his contract after the investment, failed, and started thinking about retirement. Then he simply stopped coming to work and spent his days at his beach house. Eventually Sweeney came down to see him, had what Bleszinski described as a heart-to-heart about the company's new direction, and asked him to decide whether he was in. Bleszinski wrote his resignation letter the next day. The official line, in October 2012, was "It's time for a much-needed break."
Mike Capps, Epic's president, announced his retirement that December to become a stay-at-home father for the son he and his wife were expecting, and in March 2013 he cut his remaining ties. Adrian Chmielarz, who had founded People Can Fly and come to Epic with its acquisition earlier in 2012, left with several colleagues to start The Astronauts, because free-to-play service games didn't fit what they wanted to make. Lee Perry, a lead designer on both Unreal and Gears, concluded that Epic had grown too large to be an eccentric game developer, and that the management a service business required would cramp its creative freedom. He and five other senior people left to found a studio called Bitmonster.
Each departure was its own small essay on the same subject. These were people who loved a certain kind of finished object: the big narrative game, the auteur's sequel, the studio small enough to be strange. Sweeney let them go and kept the engine.
The Tencent money bought something else besides a new business model. In March 2015, Epic made Unreal Engine 4 free for anyone to download and use, source code included, taking a 5 percent royalty only on the games that made money. A tool that had cost licensees serious sums was now open to a teenager in a bedroom. It was ZZT's free first episode applied to the most sophisticated real-time graphics software in the industry. Three months later Epic sold People Can Fly back to its management. The company was shedding the people and the projects that belonged to its previous version, and Fortnite, its supposed testbed for "living games," was still not out.
The Six-Year Overnight
Fortnite was announced at the Spike Video Game Awards in December 2011. It was a cooperative game about building forts and fighting off monsters. The staff departures slowed it down, and so did the move from Unreal Engine 3 to 4. Epic started other projects meanwhile: a free, community-developed Unreal Tournament, a free multiplayer battle arena called Paragon, a remaster of Shadow Complex, a virtual-reality shooter called Robo Recall that it released free with full source code so other VR developers could see how it was made. In July 2017, nearly six years after the announcement, Fortnite went on sale in paid early access.
Earlier that year, a game called PlayerUnknown's Battlegrounds had become a phenomenon. Its premise was simple: a hundred players dropped onto an island, a shrinking circle, last one standing wins. Epic had an engine, a building system, a cartoon art style, and an online backend that could run on PC, console, and mobile. In September 2017 it released Fortnite Battle Royale, free across all of them.
What followed doesn't need much elaboration, only numbers. More than 125 million players by May 2018. An estimated $1 billion from microtransactions, mostly cosmetic outfits and a "battle pass," by July. Epic, valued at $825 million in 2012, was estimated at $4.5 billion that July, and by October 2018, after a $1.25 billion round from KKR, ICONIQ, Kleiner Perkins, Lightspeed, and others, at nearly $15 billion. Nearly 250 million players by March 2019. Industry trackers put Fortnite's 2019 revenue around $1.8 billion. In 2020 Forbes named Sweeney its Media Person of the Year for turning a shooter into a social network, a place where a Travis Scott concert drew 28 million viewers and Christopher Nolan's Inception was screened on an island called Party Royale.
Some of the most important work was diplomatic, not technical. In September 2018 Epic persuaded Sony to drop its resistance to cross-platform play, so that a child on a PlayStation could play with a friend on an Xbox, a Switch, or a phone. That had been unthinkable for most of console history. Sweeney would later describe social ecosystems split among PlayStation, Xbox, Switch, Steam, and publishers' private systems as "a lot more broken than most people realize." The fix began with one game big enough that no platform could afford to wall it off.
Success let him be ruthless, which he has never been squeamish about. Epic shut down Paragon in January 2018 and stopped development on the new Unreal Tournament that December. Both had been free, both had communities, and both were in the way.
The University of Maryland, describing its famous non-graduate, called Fortnite a game "where players come together as a digital community to battle for survival — and occasionally dance." It's a nice line, and an accurate one. The dances made money, though. They were sold.
Thirty Percent
The money from Fortnite gave Sweeney something he had never had, which was the means to act on an old grievance at full scale.
The grievance had a long paper trail. In March 2016, years before Fortnite was a battle royale, he published an op-ed in The Guardian attacking Microsoft's Universal Windows Platform, under which certain new Windows features were available only to apps sold through Microsoft's own store, where Microsoft took 30 percent. He called it "the most aggressive move Microsoft has ever made" and predicted that "PC UWP can, should, must, and will die as a result of industry backlash." Critics pointed out, fairly, that he seemed hazy on some technical specifics, and on a Polygon podcast he conceded the point that mattered most: "Microsoft is a black box. I know a lot of people there who are really awesome, smart people who want to do the right thing… There is not proof of an evil plan to do this, but just the fear." The fear was the point. He was arguing about what a platform owner could do, not what it had done yet. It's the argument he would make for the next decade.
In August 2017 he said publicly that Valve's Steam, which took 30 percent of every PC game it sold, could run profitably on 8 percent. In July 2018 Epic cut its own cut on the Unreal Engine Marketplace from 30 percent to 12. When Fortnite came to Android that August, Epic skipped the Google Play Store entirely and asked players to install it from Epic's website. Then, on December 4, 2018, Epic announced its own PC storefront. The Epic Games Store opened two days later during The Game Awards. It would take 12 percent, and waive Unreal Engine's 5 percent royalty on games sold through it. By Epic's own math, after delivery and services, the store would keep about 5 percent of gross revenue, perhaps 6 or 7 at scale. It was the lawn-mowing gambit again: do the job at half the price and let the incumbent explain why it cost so much.
Valve had cut its own rate just days earlier, to 25 percent on a game's revenue above $10 million and 20 percent above $50 million. Indie developers noticed that this helped mostly the biggest publishers. The timing made Epic look like the insurgent.
The trouble was that cheaper isn't enough when nobody's in the store. To get customers, Epic gave away games: one every two weeks at first, then one every week, then sometimes two. Documents from the Apple trial later showed Epic paying developers buyouts of typically $100,000 to $1 million per giveaway, working out to roughly 50 cents to $5 per new user. When it gave away Grand Theft Auto V in May 2020, more than seven million new users signed up and crashed its servers. Nineteen million people claimed Star Wars Battlefront II in January 2021. To get supply, Epic paid for exclusives: timed deals, usually six months or a year, with guaranteed minimum revenue, Epic covering any shortfall.
The backlash was ugly and personal. Metro Exodus went Epic-exclusive a few weeks before launch, and Steam users review-bombed the developer's older games. Crowdfunding backers of Phoenix Point felt their money had been used to get the game to a point where it could be sold to Epic. When the developers of Ooblets announced their deal with a jokey line about "immature, toxic gamers," they got thousands of hostile messages, including threats. Sweeney responded by describing "the coordinated and deliberate creation and promotion of false information, including fake screenshots, videos, and technical analysis, accompanied by harassment of partners, promotion of hateful themes, and intimidation of those with opposing views." He did not retreat. He called exclusives the "solution proportionate to the problem" of Steam's 70/30 split and said he would stop seeking them if Valve cut its rate. Valve didn't.
Gabe Newell's response was gracious in a way that conceded nothing. Competition, he said, "is awesome for everybody. It keeps us honest, it keeps everybody else honest," but in the short term, "it's ugly."
The numbers were both impressive and bad. In 2019 the store drew 108 million customers and $680 million in sales, $251 million of it on third-party games, and 90 percent of that third-party money came from exclusives. Epic said sales beat its projections by 60 percent. In its Apple filings, Apple cited deposition testimony from store head Steve Allison: the store lost about $181 million in 2019 and $273 million in 2020, and Epic had committed $444 million to minimum guarantees. A later analysis by Simon Carless found that of the first wave of 2019 exclusives, only Satisfactory had earned back what Epic paid for it. Sweeney's answer, on Twitter, was characteristically cheerful: the store "has proven to be a fantastic success in reaching gamers with great games and a fantastic investment into growing the business." Epic said it expected profitability by 2023. In 2023 the store did $950 million in revenue with 270 million users.
In April 2021 Microsoft matched Epic's 88 percent developer share on its own PC store. Steam held its standard rate. In June 2025 Epic stopped taking any cut on a game's first $1 million in sales. In February 2026, the store's general manager, Steve Allison, said in public what users had said for years: "the launcher sucks."
The store was never really the point. It was a seven-year demonstration, at a cost of hundreds of millions of dollars, that a distribution business could run on 12 percent. Sweeney was preparing evidence for a different argument with a much larger opponent.
Project Liberty
In June 2015, a few weeks after Epic appeared at Apple's Worldwide Developers Conference to praise Apple's Metal graphics API, Sweeney emailed Tim Cook directly. The subject line was "iOS as an open platform." "The App Store has done much good for the industry," he wrote, "but it doesn't seem tenable for Apple to be the sole arbiter of expression and commerce over an app platform approaching a billion users." He proposed separating App Store curation from security review and distribution: keep the safety checks, let other stores exist.
Cook forwarded the email to Phil Schiller with one line.
— Tim Cook, forwarding Sweeney's June 2015 email to Phil Schiller (trial exhibit, Epic v. Apple)Is this the guy that was at one of our rehearsals?
It probably was. Epic had been in the WWDC rehearsals, though Sweeney hadn't gone on stage. For Apple, he was the man who made its hardware look good, and the idea that this person might be an adversary seemed not to have occurred to anyone.
He kept at it. In January 2018 he asked Mark Rein to push for a meeting with Apple's marketing chief, Greg Joswiak. "If the App Store [were] merely the premier way for consumers to install software and not the sole way," Sweeney argued, "then Apple could curate higher quality software overall, without acting as a censor on free expression." Rein got a hearing through an Apple contact named Tim Kirby and reported back with an optimism so hedged it reads as despair: Kirby would "line up people who will listen and not shoot it down like Phil Schiller would.... so maybe there's a smallest little crack on the very outer surface of the many feet of ice covering the frozen lake that is the Apple store."
The lake did not thaw. Epic spent months on a plan it called Project Liberty, which included a sixty-page complaint and a short film. On June 30, 2020, at 4:00:09 p.m. Pacific, Sweeney emailed Cook, Schiller, Craig Federighi, and Matt Fischer under the subject "Consumer Choice & Competition." He asked for two things, competing payment processing in Epic's iOS apps without Apple's fees, and a competing Epic Games Store app on iOS with "equal access to underlying operating system features for software installation and update as the iOS App Store itself has." He noted that Epic had been "required to accept your standard, non-negotiable contracts." He asked for an answer in principle within two weeks.
On August 13, 2020, Epic updated Fortnite on every platform with a permanent 20 percent discount on V-Bucks, the in-game currency, for anyone who bought directly from Epic. On iPhones and Android phones, players who paid through Apple or Google didn't get the discount, and the game said why: the 30 percent. Apple removed Fortnite from the App Store within hours. Google followed. That same day Epic filed antitrust suits against both in the Northern District of California, sought no money damages, only injunctions, and released a parody of Apple's "1984" Super Bowl ad in which Apple was Big Brother. Fortnite players were the people in the gray uniforms.
Apple said Epic's problem was "entirely self-inflicted and is in their power to resolve." The next day Sweeney wrote that Epic was "fighting for the freedom of people who bought smartphones to install apps from sources of their choosing, the freedom for creators of apps to distribute them as they choose, and the freedom of both groups to do business directly." The opposing argument, he wrote, amounted to "Smartphone markers [sic] can do whatever they want. This is an awful notion. We all have rights, and we need to fight to defend our rights against whoever would deny them." When Apple suggested Epic wanted a special deal, he answered: "We're fighting for open platforms and policy changes equally benefiting all developers. And it'll be a hell of a fight!"
It was. Apple moved to terminate Epic's developer accounts, which would have cut Epic off from updating Unreal Engine for iOS and macOS and stranded every developer who relied on it. Judge Yvonne Gonzalez Rogers protected the engine, citing "potential significant damage to both the Unreal Engine platform itself, and to the gaming industry generally," but would not restore Fortnite: "The current predicament appears of [Epic's] own making." That sentence is about as accurate a description of Sweeney's method as exists. The predicament was of his own making, deliberately, timed to the second, built so that the other side had to act first and in public.
On September 10, 2021, Rogers ruled for Apple on nine of ten counts, including the 30 percent commission and the ban on rival app stores. She ruled for Epic on the tenth: Apple could no longer forbid developers from steering users to cheaper purchases elsewhere. The Ninth Circuit upheld the decision, and on January 16, 2024, the Supreme Court declined to hear the case. Apple complied in a way that kept most of its fees. Ten days later Sweeney tweeted: "Under what possible theory of antitrust regulation is it acceptable for a monopoly to decide what companies are allowed to compete with it, and on what terms they can compete? Apple makes a mockery of free market competition." On April 30, 2025, Rogers found that Apple had willfully violated her injunction, and that an Apple executive had lied under oath. She barred Apple from charging fees on outside purchases or discouraging users from them. Within a week, Epic announced it would bring the Epic Games Store to iPhones in the United States. Fortnite returned to the App Store that May.
The Google case went to a jury in November and December 2023. Before trial, Sweeney testified that Fortnite distributed through the Samsung Galaxy Store as an alternative was converting at "a much lower rate than we'd estimate if we were on Google Play." His summary: "It's not doing well." On December 11, 2023, the jury found unanimously for Epic on every claim. The Ninth Circuit upheld the verdict in July 2025. In March 2026 the two companies settled: lower Play Store fees, more room for alternative billing and third-party stores, and Google's withdrawal of its Supreme Court petition. The binding term sheet also contained something else. Sweeney, who had called Android a "fake open platform" and Google's developer-retention program "astonishingly corrupt," agreed not to publicly disparage Google on Play Store distribution and fees until 2032.
He had won against Google, and in exchange he would stop talking about it.
— Tim Sweeney, on the Channels podcast, 2025I started programming back on an Apple II when I was 13: You turn the computer on, you get a BASIC programming prompt. Anybody can write code, anybody can save it to a floppy disk, you can share it with a friend, you can sell it.
He has said the fight with Apple cost Epic more than $1 billion, and that it was worth it. For all the language of rights and freedoms, the moral argument is really a memory. The Apple II he started on came with a programming prompt and no store. You could make something and mail it to a stranger, and a retired cartographer could fill the order.
The Other Ledger
The man who spent a decade accusing platform owners of exploiting their users ran a platform too, and in December 2022 the Federal Trade Commission described how.
Epic agreed to pay $520 million, a record, to settle two sets of charges about Fortnite: that it had collected personal data from children without parental consent, in violation of the Children's Online Privacy Protection Act, and exposed children and teens to potential harassment through default voice and text chat; and that it had used design tricks to push players into purchases they didn't intend. The FTC called these "dark patterns." Of the total, $245 million went to the purchase charges, and by December 2024 the agency was sending $72 million in refunds to more than 600,000 claimants. Epic's statement was unusually soft: "No developer creates a game with the intention of ending up here. The laws have not changed, but their application has evolved and long-standing industry practices are no longer enough." The company had built child accounts and parental controls before the settlement. The ledger was still the ledger. The V-Bucks that Epic discounted to provoke Apple were the same V-Bucks the FTC said Epic had nudged children into buying.
Then came the overreach. With Fortnite money and repeated funding rounds, $1.78 billion in mid-2020 at $17.3 billion, $1 billion in April 2021 at $28.7 billion for "long-term vision for the metaverse," and $2 billion from Sony and Kirkbi, the Lego family's investment company, in April 2022 at $32 billion, Epic bought broadly. It bought the video-chat app Houseparty in June 2019 and shut it down in October 2021. It bought the indie music platform Bandcamp in March 2022 and sold it to Songtradr in September 2023. It bought ArtStation and Sketchfab in 2021, cut their fees to 12 percent, and sold both to KitBash in August 2026. Sweeney once said "metaverse" was "a word that kind of has a stock price. When we release something cool, it goes up and when we release something that's not cool, it goes down." Epic, as a private company, had no stock price, but it had a valuation, and when Disney invested $1.5 billion for a 9 percent stake in February 2024, reporting on the terms showed that valuation had been cut sharply.
In September 2023, Epic laid off 870 people, about 16 percent of staff. Sweeney said it was necessary to rein in spending and that he did not expect more layoffs. In March 2026, citing a decline in Fortnite engagement since 2025, Epic cut more than 1,000 more. Sweeney said the company was spending significantly more than it was earning and expected the cuts, along with reduced contracting and marketing, to save more than $500 million. Epic shut down Fortnite modes nobody was playing: Rocket Racing, Fortnite Ballistic, the Fortnite Festival Battle Stage. Bloomberg reported that some had been released before they were ready. Then Sweeney, meaning to be kind, tweeted that "employers will see a stream of resumes of once-in-a-lifetime quality folks," that Epic "never lowered our hiring standards as we grew," and that "it's a safe bet that anyone with Epic Games on their resume is in the top few percent of their field."
The industry did not take it as kindness. "All of this is irrelevant unfortunately," wrote Robert Morrison, a senior animator at Void Interactive. "As there are very very few job openings. The majority will be out of work for 6 months to a year or more." Michael Douse, Larian Studios' director of publishing, called it "absolute LinkedIn brain rot." A GDC survey that year found that 48 percent of game workers laid off the previous year were still unemployed. The engineer who reasons from first principles about fees and freedoms had, for an afternoon, reasoned from first principles about other people's grief, with predictable results.
There were smaller embarrassments, the kind that come with scale. In May 2025, with permission from James Earl Jones's estate, Fortnite added an AI-voiced Darth Vader that players could talk to. Within two days players had made it say racial and homophobic slurs, and Epic pushed a hotfix. SAG-AFTRA filed an unfair labor practice charge on May 19. Tencent's two board members stepped down after a Justice Department inquiry into their simultaneous seats at Riot Games. Visitors to Epic's campus in Cary, the News & Observer reported, were met at the front desk with a stack of nondisclosure agreements. The champion of open platforms runs a very closed building.
One property tells the whole story. In January 2021, Epic bought the Cary Towne Center, a dying regional shopping mall, to convert into its headquarters by 2024. It demolished everything except one section occupied by a Belk department store, which kept operating. Then nothing happened. On December 6, 2024, the Town of Cary revoked Epic's rezoning application as "inactive." The site is a field of cleared ground with one department store standing in it.
Territory
The land is easier to see in North Carolina than the code.
Since the real-estate collapse of 2008, Sweeney has used his fortune to buy forest, roughly 50,000 acres by most counts, making him one of the largest private landowners in the state. He has said little about it in public, which is part of how it works. Conservationists describe a man who buys land when it comes up and holds it until an institution can take it.
The best-documented case is Box Creek Wilderness, about 7,000 acres in the western Piedmont, home to more than 130 rare and threatened plant and animal species. He paid $15 million for it. Then a power company, Rutherford Electric, filed a condemnation suit to run a transmission line through it. The man who had spent years arguing that infrastructure owners should not be able to route through whatever they liked was now defending a tract of forest from an actual utility line. The suit was settled after the U.S. Fish and Wildlife Service and Senator Richard Burr got involved. In 2016 Sweeney donated a conservation easement on the land to the Fish and Wildlife Service, which keeps it from being divided. His statement was courtly and plain: "I'm grateful for the efforts of Senator Burr to help protect Box Creek Wilderness, and for the whole Fish and Wildlife Service team's tireless efforts to preserve vital North Carolina natural areas in partnership with conservation-minded landowners like me."
He donated 1,500 acres to expand Mount Mitchell State Park, around the highest peak in the eastern United States. In April 2021 he announced he would give 7,500 acres in the Roan Highlands to the Southern Appalachian Highlands Conservancy, which would manage it as a preserve, run scientific studies with him, and offer guided hikes. It was the largest private conservation land donation in the state's history. In early 2020 he bought Falkland Farms, a 7,326-acre peninsula in Halifax County, Virginia, twelve square miles where the Dan, Banister, and Staunton rivers meet, for $11.5 million. A University of North Carolina botanist named Alan Weakley had suggested it because the tract held a rare oak and pine savannah and the state couldn't move fast enough to buy it before it sold. Sweeney explained the terms in a direct message on Twitter: "When it goes to conservation, I will sell it for the price I paid."
The man who charges 12 percent on games will hold a forest at cost.
He has offered a reason in arithmetic: "Land conservation is the one unquestionably practical and cost-effective thing we can do to protect ecology and the future habitability of the planet." He has also offered reasons that aren't arithmetic. "I'm single. Unmarried, no kids. So I lead a fairly simple life," he told The Wall Street Journal in 2019, in one of his rare interviews. He wears T-shirts and cargo pants. His favorite food is Bojangles' fried chicken. When he isn't working, he is in the woods. "I love going off the trail and finding things that I think nobody's ever seen before." A boy who rode ten miles on his bike to explore woods, back by dark, is now a man who owns the woods so they will stay unexplored.
The interviewers who follow the metaverse sometimes make the connection explicitly: he thinks about ecosystems, physical and virtual, as one problem. Perhaps. The rhyme I keep coming back to is his father's. Paul Sweeney spent his career at the Defense Mapping Agency making maps from satellite imagery, long before anyone could pull up the planet on a phone. In March 2021 his son's company announced a partnership with a geospatial firm called Cesium to offer Unreal developers a free plug-in that would let them recreate any part of the mapped surface of the Earth. The father made the map, and the son made the map you can walk around in. Then the son bought large pieces of the territory and stepped back from them.
The small shelf of books that has gathered around him (there is more than one) tends to cast him as a visionary, which is true, or a contrarian, which is also true. Neither word quite fits a man who wanted to build the whole car, then the whole road, and then a forest with no road in it.
One more artifact. The source code for ZZT, the 20,000 lines of Pascal written at night on a 286, was gone. "If I had it, I'd release it," Sweeney said, "but I lost it in a crash a long time ago." A community programmer named Adrian Siekierka spent years reconstructing it from the compiled program, byte for byte, and released the reconstruction in 2020 with Sweeney's blessing. Then, as Siekierka put it, borrowing from Tolkien, "they were all of them deceived, for another backup was made." On January 28, 2023, the original source code for ZZT 3.0 was posted to GitHub with Sweeney's permission, under the MIT License.
The last copy had left the house in Potomac ten years earlier, in a package addressed to one person. The license now attached to the game is a few short paragraphs, and the first of them opens: Permission is hereby granted, free of charge, to any person obtaining a copy of this software.
Part IIThe Playbook
What follows is not a set of rules Sweeney has published; he hasn't. It is a set of patterns drawn from thirty-five years of decisions, some of which worked spectacularly and some of which cost hundreds of millions of dollars. The principles favor the structural over the inspirational. They are about where to stand in a value chain, when to burn the version of your company that's working, and how to pick a fight so that the other side has to throw the first punch. Several come with warnings attached, because the same instincts that built Epic also produced its worst mistakes.
The Sweeney Principles
- 1.Ship the editor, not just the game.
- 2.Borrow the costume of scale, then take it off in public.
- 3.Version the company before the market versions it for you.
- 4.Sell equity to buy freedom, never the reverse.
- 5.Give away the tool; tax the outcome.
- 6.Recruit from your users.
- 7.Keep a founder's hand on the code.
- 8.Let an opening find the machine you already built.
- 9.Use price as proof, and pay for the proof.
- 10.Engineer the provocation so the incumbent moves first.
- 11.Sue for the rule, not the exception.
- 12.Expect to be held to your own standard.
Principle 1
Ship the editor, not just the game
ZZT existed because Sweeney disliked his text editor, and the thing that made it last was that the editor came in the box. More than 200 users submitted worlds within months, 3,000 exist today, and Sweeney believes tens of thousands of industry professionals got their start inside it. He drew the lesson directly: Unreal was designed around "clean code and editing tools, so that others can build their own games," in the account he and Rein have given, and Unreal Tournament became, in Rein's words, ZZT's spiritual successor. Fortnite Creative and the Unreal Editor for Fortnite are the same move at a scale of hundreds of millions.
The strategic logic is that a product people consume has a ceiling set by your output, while a product people build in has a ceiling set by theirs. "Future [game] worlds are going to be so large," Sweeney said in 2017, "that we can't possibly expect any game developer to create everything." The editor turns customers into suppliers, and suppliers rarely leave.
Tactic: Whatever internal tool you use to make your product, ask what would happen if you shipped a version of it to customers, then do the smallest version of that experiment this quarter.
Principle 2
Borrow the costume of scale, then take it off in public
"Epic MegaGames" was, by Sweeney's cheerful admission, "kind of a scam to make it look like we were a big company." In a shareware market dominated by Apogee and id, a one-man operation in a parents' house needed to look like a peer to get catalog placement, distribution deals, and a partner like Mark Rein. The name did that.
The more interesting move came in 1999, when Epic dropped "Mega" because it "no longer wanted to pretend to be a big company." Taking the costume off at the right moment is a signal of its own: the size was real now, and the company could afford to understate. Founders who never stop performing scale end up managing a reputation instead of a business.
Tactic: Present yourself at the scale of your ambition when you are small enough that nobody checks, and schedule the moment you'll stop performing it once you no longer need to.
Principle 3
Version the company before the market versions it for you
Sweeney numbers Epic's eras like software releases, and the habit is more than a joke. It treats the company's current business model as a build that will be deprecated. Every major transition came while the existing version still worked: consoles while PC games still sold, services while Gears was a hit, a store while Fortnite was printing money. The pattern is to move when the next version of the old model becomes unaffordable, not when the current one fails. Gears 4 at $100 million was the signal.
E
The Versions of Epic
Sweeney named 'Epic 3.0' and 'Epic 4.0'; the earlier and later eras are implied by the pattern.
1991
Potomac Computer Systems ships ZZT via shareware; Paul Sweeney fills the orders.
1998
Unreal ships; unsolicited licensing calls reveal an engine business.
2006
"Epic 3.0": piracy pushes Epic to consoles; Gears of War grosses ~$100M on $12M.
2012
"Epic 4.0": Tencent pays $330M for ~40%; key creative leaders depart.
2015
Unreal Engine 4 goes free with a 5% royalty.
2017
Fortnite Battle Royale launches free across PC, console, and mobile.
2018
Epic Games Store opens at a 12% cut.
2020
Project Liberty: Fortnite removed from iOS and Android; lawsuits filed same day.
2026
1,000+ layoffs and $500M in savings as Fortnite engagement falls.
The cost of versioning is people. The Tencent transition lost Bleszinski, Capps, Fergusson, Chmielarz, Perry, and the Gears franchise itself. Sweeney accepted that. His explanation, years later: "There really are some things that only a founder of a company can do. Sacred cows that need to be addressed."
Tactic: Write down what your company's next version would be if your current model's cost doubled tomorrow, and start funding it before the doubling happens.
Principle 4
Sell equity to buy freedom, never the reverse
The Tencent deal looks at first like a founder giving away 40 percent of his company. Look at what he got: $330 million, operational expertise in service games from the operator of League of Legends, and, by Sweeney's account, very little creative interference. What he kept was control. Through a decade of further rounds from KKR, Sony, Kirkbi, and Disney, he still holds about 41.4 percent and remains the controlling shareholder.
The deal's purpose was to escape a different kind of dependence, on publishers, whose "business objectives" had just vetoed Epic's multiplayer Gears. Sweeney swapped a counterparty who controlled his product decisions for one who owned a share of his upside. That swap paid for free Unreal Engine 4, for Fortnite's long gestation, and eventually for the lawsuits.
Tactic: Before any financing, list the decisions the new capital could block, and refuse any money whose terms touch the decisions that define your company, however generous the valuation.
Principle 5
Give away the tool; tax the outcome
In March 2015 Epic made Unreal Engine 4 free, source code included, and took 5 percent only of revenue from successful games. In 2020 it waived royalties on a game's first $1 million. The Epic Games Store now takes nothing on a game's first $1 million. Twinmotion was bought and given away. MegaGrants, $100 million in total, funded projects up to $500,000 each, including $1.2 million to the open-source Blender Foundation, a tool that doesn't directly benefit Epic at all.
The logic inverts traditional licensing. Upfront fees filter out exactly the small, unproven developers who might become the next Tripwire or Psyonix. A royalty on outcomes aligns Epic's revenue with its customers' success and makes the engine the default choice for anyone starting out. The cost is borne by the hits, which can afford it.
Tactic: Find the fee that keeps your smallest potential customers out, and test replacing it with a share of their eventual success.
Principle 6
Recruit from your users
The six winners of the ZZT's Revenge contest got jobs. A seventeen-year-old who mailed in Dare to Dream became the lead designer of Gears of War. Make Something Unreal turned a mod team into Tripwire Interactive. Psyonix contributed to Unreal Tournament and was later acquired. The pattern holds for thirty years: Epic's tools work as a hiring funnel, filtering for people who have already shown they can build in Epic's medium without being asked.
This is cheaper and more reliable than conventional recruiting because the work sample comes first. A user who has spent hundreds of hours in your editor has already passed the most important interview.
Tactic: Run a public building contest in your product's tools and treat the top entries as your next hiring class, not just your next marketing campaign.
Principle 7
Keep a founder's hand on the code
Sweeney "stepped back" from programming in the early 1990s and thought he was done. Then Unreal needed an engine and he wrote a quarter-million lines of it. In 2013, as CEO, he was described as still coding every day. In 2017 his CTO teased him for having code open on screen during an interview. He has co-authored academic papers on programming languages, including a 2006 POPL talk and the 2022 Verse Calculus with Simon Peyton Jones, the theory behind the language Epic now ships for Fortnite creators.
This is not nostalgia. It lets the CEO judge technical bets directly: the engine that made Epic, the decision to make it free, the cost of a $100 million sequel. A founder who can read the code knows where the company's real leverage is and which promises engineering can keep.
Tactic: Whatever your company's core craft is, keep a standing, protected block of your week in which you personally practice it, not just review it.
Principle 8
Let an opening find the machine you already built
Fortnite was a six-year slog until PlayerUnknown's Battlegrounds showed that a hundred-player survival format could become a phenomenon. Epic didn't invent the battle royale. What it had was everything needed to exploit one: an engine it owned, a building mechanic nobody else had, a cartoon style that ran on weak hardware, and an online backend for every platform. Fortnite Battle Royale shipped in September 2017, free, everywhere, and had 125 million players by May 2018.
The lesson is not "copy fast." It is that infrastructure built for one purpose gives you the option to act on opportunities you can't predict. The years that looked wasted were spent building the machine.
Tactic: When a competitor proves a new format, ask what you already own that would let you do it better, and move only if the answer is something they can't easily copy.
Principle 9
Use price as proof, and pay for the proof
When Sweeney claimed in 2017 that a store could profit at 8 percent, it was an opinion. The Epic Games Store at 12 percent, keeping about 5 percent of gross after costs, was a working demonstration, and it cost a great deal: roughly $181 million in losses in 2019 and $273 million in 2020, $444 million in minimum guarantees, free-game buyouts of up to $1 million each. Most exclusives never earned back what they cost.
%
Price as Argument
How Epic used its own rates to challenge industry norms.
| Industry norm | Epic's move | Outcome |
|---|---|---|
| 30% marketplace cut | Unreal Marketplace to 12% (2018) | Adopted |
| 30% store cut | Epic Games Store at 12%, no Unreal royalty | Microsoft matched in 2021 |
| Paid exclusives as the exception | Timed exclusives with revenue guarantees | Backlash; most lost money |
| 30% mobile cut, single store | Direct payment with 20% V-Bucks discount | Litigation; partial wins |
The proof had value beyond the store's own P&L. It gave regulators, courts, and other developers a real-world example that distribution could be cheaper, which mattered when Epic stood in court and argued that 30 percent was a monopoly rent rather than a cost.
Tactic: If you claim an incumbent's price is inflated, budget to operate at your claimed price long enough to produce data a skeptic would accept.
Principle 10
Engineer the provocation so the incumbent moves first
Project Liberty was planned for months: a sixty-page complaint, a parody of "1984," a two-week ultimatum emailed on June 30, 2020. On August 13 Epic shipped a discount that broke Apple's and Google's rules on purpose, waited for the removals that came within hours, and filed suit the same day. The incumbents had to act first and in public, and the story became "Apple bans Fortnite" rather than "Epic sues Apple."
The method has a cost, and the court named it: "The current predicament appears of [Epic's] own making." Judge Rogers protected Unreal Engine but left Fortnite off the App Store for years. A staged provocation wins the narrative and gives up the moral high ground of the injured party.
Tactic: Before you force a confrontation, decide in advance which of your assets you are willing to lose in it, and ring-fence the ones you are not.
Principle 11
Sue for the rule, not the exception
Epic sought no money damages from Apple or Google, only injunctive relief "to allow fair competition." When Apple said Epic wanted a special deal, Sweeney replied that Epic was fighting for "policy changes equally benefiting all developers." Asking for a rule change instead of an exemption turned a commercial dispute into a public-interest case, recruited allies through the Coalition for App Fairness, and made a quiet settlement for Epic alone much harder for either side.
The results were uneven: nine of ten counts lost against Apple in 2021, a sweeping willful-violation ruling in 2025, a unanimous jury against Google in 2023. The Google settlement also shows the price of eventual peace. Sweeney agreed not to publicly disparage Google's Play Store practices until 2032. Even a rule-change fight ends in a contract.
Tactic: When you challenge a powerful counterparty, frame your demand so that winning it would help your competitors too, because that is what makes it hard to buy you off.
Principle 12
Expect to be held to your own standard
The FTC's $520 million settlement in December 2022, for children's privacy violations and design that pushed players into purchases, landed on a company that had spent years accusing platforms of exploiting their users. The 2026 layoffs came three years after Sweeney said he didn't expect more. His tweet that laid-off staff were "once-in-a-lifetime quality folks" was received as "LinkedIn brain rot" by peers. The Cary Towne Center was razed except for a Belk and then left dormant. Each episode cost Epic credibility on precisely the terrain where Sweeney argues for a living: fairness to users and creators.
A founder who builds his public identity on a principle gives critics the yardstick. That is not a reason to abandon the principle. It is a reason to audit your own platform against it before anyone else does.
Tactic: Take the strongest criticism you make of your biggest competitor and commission an internal review that applies it, word for word, to your own product.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Tim Sweeney, email to Tim Cook, June 2015The App Store has done much good for the industry, but it doesn't seem tenable for Apple to be the sole arbiter of expression and commerce over an app platform approaching a billion users.
— Tim Sweeney, to GamesIndustry.biz, GDC 2017Minecraft brought about 100m people in, initially as gamers but all of them became digital content creators designing and building 3D scenes. I think that's the future of the industry: make everyone a creator.
— Tim Sweeney, to Gamasutra, 2017The story of Epic is just a bunch of smart, hardworking people starting a company together and learning as we went. And just making some cool things. There weren't investment bankers driving it, or people in suits. It was all completely organic.
— Tim Sweeney, to the University of Maryland Clark School of Engineering, 2019Land conservation is the one unquestionably practical and cost-effective thing we can do to protect ecology and the future habitability of the planet.
— Tim Sweeney, to PC Gamer, 2026It's now clear that nobody's going to end up with an absolute monopoly over gaming.
Maxims
- Put the editor in the box. The tool you used to build the product is often worth more in your customers' hands than in yours.
- Undercut the incumbent, then do the work. A fourteen-year-old mowing lawns at half price and a store at 12 percent run on the same logic.
- Number your company's versions. Treating today's business model as a release scheduled for deprecation makes the next one easier to fund.
- Trade equity for freedom, never control for money. Sell a share of the upside; keep the decisions that define you.
- Tax outcomes, not attempts. Free tools and royalties on success build a pipeline that upfront fees would choke.
- Hire the people already building in your world. A user's finished work is a better interview than any résumé.
- Stay close enough to the code to read it. A founder who still practices the craft knows which bets the company can actually make.
- Make the incumbent move first. A well-built provocation wins the story, but count what it will cost before you trigger it.
- Fight for the rule everyone gets. A demand that helps your rivals too is the one no one can quietly buy off.
- Hold your own platform to your own speech. Every principle you preach publicly becomes the measure your critics will use on you.
