Code is free. Fun isn’t.

a few thoughts on what’s happening in game dev right now, because i can’t stop thinking about it.

this week someone put GTA V in a chrome tab. 700mb after compression, running as a website. it got taken down for copyright of course, but damn, that’s impressive. people have been mixing and matching games through reverse engineering, which used to be one of the hardest jobs in games. opus 5.5 is scarily good at it.

even funnier: a guy who had never made a game asked claude for “a temu league of legends, browser only, multiplayer.” it worked first try. a day later it had 18 champions, bots and a jungle.

then there’s the one that surprised me most, because it’s ours.

we shipped a game recently. two people did all the code and all the art. it’s doing $15k a day in gross revenue right now. D1R is 61%. in our genre, only the top 1% of games get there. a two-person team got next to the top 1% in a matter of days.

i’ve shipped hundreds of games and never seen anything like this. it feels like the ground moved and i don’t know where to stand yet.

here’s what i think, for now.

building is basically free, and it was never the hard part. players don’t pay for code. they pay for the feeling of wanting to come back tomorrow, and AI can’t give you that. we barely spent time writing code. we spent it deciding what the first 30 seconds feel like, what a gem is worth, and what to cut. so the 61% isn’t an AI number. it’s probably a taste number.

the tools got much better through 2025, and mobile D1R went down that year. if building were the bottleneck, better tools would have lifted the average game. they didn’t. more production gets you more games, not more fun.

cloning might be worthless soon unless you have distribution. if anyone can rebuild anything in a weekend, the only thing left with value is what someone owns. reskinning proven games used to be a business model (a big one in turkiye, where i’m from). now it’s a prompt.

there are about to be more games than anyone could play, and it started before AI got good. steam had 20k+ releases in 2025, about 55 a day. only 608 reached 1k reviews, roughly 3%. the most prolific publisher shipped 138 games that year and none got more than one review. now multiply supply by ten. there are still 24 hours in a day. the hard part moves from making to finding.

the browser might be back. GTA in a chrome tab reads like AI twitter clickbait, but it’s a distribution story. app stores take 30% and own your player. a URL takes nothing. the walls are already coming down: in april last year a US court killed apple’s anti-steering rules, so devs can link players straight to their own web stores. the best midcore games were already doing 40–50% of revenue on web before that ruling. a game that lives at a URL, takes its own payments and owns its players is a very different business from one that rents them from apple.

maybe the big studio is the new SaaS company. AAA budgets went from $50-150m to $200m+, and around 45k jobs disappeared between 2022 and mid 2025. people blame AI. i don’t. the layoffs started before the tools were good, because the cost structure was already breaking. now that cost structure is up against one person with taste and a $200 claude max subscription. that’s not a fair fight, and the small team won’t be the one losing.

so i find it a bit sad that most game devs now think AI is bad for the industry. i get the fear. but the people most scared of the tool are the ones who’d be most dangerous with it.

i don’t know exactly where this goes. anyone who says they do is probably bullshitting.

we do have a preview though. roblox creators earned $1.5b in 2025, mostly kids and tiny teams no studio would have hired. that’s what happens when millions of people can make games: a power law, a lot of noise and a few weird hits.

the next step is stranger. take google’s genie 3 direction: what if the engine dissolves and the world gets generated as you walk through it? i don’t buy it yet, because design needs the taste and juice i talked about above. but what if AI keeps getting better at that too?

if that happens, what’s left is what was always the actual game: rules, economies, status, other people. a world generated just for you is a toy. a world you share with a million people who care who’s top of the leaderboard is a game. AI can make infinite content. it can’t make shared meaning.

so the number of people who can make a game goes from thousands to millions. most of it will be bad. some of it will beat anything a studio would greenlight: one person’s weird idea, straight to players, nobody sanding the edges off.

that part makes me really happy. for most of history, a game got made only if a publisher believed it would sell. now the only filter is the player.

if you run a studio, using AI is a given. the real question is what you own that AI can’t make. your taste? a brand with a real community behind it? distribution? i don’t know.

if you’re a dev, stop optimising how fast you build. everyone wins that race now. optimise what you build and who it’s for.

if you invest in games, stop funding headcount. fund taste, IP and the direct line to players. everything else is turning into a prompt.

code is free. fun isn’t. it never was.