What YouTube did for video,
Arcade AI will do for games.
Text → Game.
1 click → Global distribution.
Runs on any computer. Directly on the web.
Creation and distribution just became universal.
Sign up to @joinarcadeai in comments.
Weekly Update #7 is live.
This week on Arcade:
↩️ Restore your game from chat
💾 Automatic saves
🎨 Pick colors directly from objects
⏳ Keep builds progressing while you’re away
🧗 Better climbing
⚡ Major performance improvements
🧱 More reliable saving & reloading
Growth is accelerating too:
+4,000 new creators this weekend
+400 onboarded
+600 Discord members
We’ll onboard more creators every week ahead of the public launch.
Retention keeps improving week after week.
Full update ↓
https://t.co/YDnwIg5wyj
Arcade is heading to Gamescom
We’ll also be joining the Invest Circle this year.
If you’re around, shoot me a DM and let’s find a time to connect!
See you there ✌️
Seven features. Seven days.
And the curve is following. 2,000+ on Discord, 10k+ new users in a few days, acquisition is starting to look genuinely good.
Been heads down. Posting these every week from now on.
@alightinastorm Pointing at personal investment doesn’t answer the argument.
The core claim still stands: demos aren’t games.
And don’t worry, growth is going very well on our side.
This take is popular.
It’s also completely wrong.
No one has shipped a real full-AI game yet.
What you’re seeing are demos: assets assembled, a few basic rules, a short loop that looks good in a 30-second clip. That’s not a game.
A game is systems that hold over time, pacing that works, and something players actually come back to.
Assembling assets has been easy since Unreal. Making something sticky is not.
Vibecoding is fun. It’s also 1% of the problem.
It’s like saying cameras made YouTube irrelevant.
Cool, you can film. Now how do you publish? How do you get distribution? How do you know what people actually want to play? How do you make a good game?
Where you’re right is that most AI engines are taking the wrong direction. There’s no point building interfaces and systems that just look like a slightly better Cursor. The market is completely missing it, too focused on the technical barrier, which is already collapsing with current models.
The real value was never there.
That’s why we still need new workflows and new platforms, not another general-purpose engine, but something that helps people create games worth playing.
General Purpose Game Engines Are Dead -
so stop building a new one
If you haven't noticed yet, it's easier than ever to build complementary editors for your games with AI.
That's why you see most of us vibe game devs so obsessed with building them: Figure out the right format, a good level of abstraction that still has (some) reusability across games in a niche, try to integrate AI vertically into it
So you noticed that building an AI native unity is just 3 months of /looping with claude away? Guess what, we all noticed it a long time ago and no, we're not stupid and did not leave that part of the field open for you to grab it. It's a dead end, it's where most of us started over 1.5 years ago.
"Fine, i'll make it a platform instead, built on top of my engine": Great idea bro! Now you just expanded your problem space by 100x in a developing market with an uncertain future. You're developing an engine which is DoA and also want to play the hardest economical game by building a platform. Notable exceptions here are platforms that try to become novel game like experiences themselves - risky move, worth a shot.
But there is value in having a good pre-packaged engine that helps you ship games faster! Yes, indeed, but what if it doesn't look like another bloated UI where you have to watch a 15 minute Youtube Video to figure out there is a hidden plugin menu in the engine ui? What if that framework is now code-only and you build your own editing experience, tailored to your game?
Go, experiment, build a game, build the engine and use reliable libraries, build libraries for others!
But stop building the next unity, no one wants it
@GamefxAI Agree that starting from scratch is inefficient.
But the deeper issue isn’t assembly or graphics.
Those are mostly solved already.
The real unlock is when AI lets a non-technical person make a game that people keep playing, not just generate a playable prototype.
Generating a demo in 48 hours with Opus for a few hundred bucks is cute.
But that’s not going to revolutionize video games.
Right now we’ve got demos that get a ton of buzz.
Kart racers, FPS, Vampire Survivors clones.
They run. They look clean in screenshots.
People like them.
And then almost nobody actually plays them.
Or they play for 15 minutes to flex, then move on.
Because they’re not games.
They’re technical demos.
Most studios are still anti-AI deep down, or they’re scared of how their audience will react, so they use AI in secret.
The ones who do adopt it use it to speed up tasks, not to generate the entire game from a prompt.
And on the other side, the people running these loops today are rarely actual game designers.
Result: clean code, sometimes even pretty, but it lacks feel, pacing, art direction, that thing that makes a game actually grab you.
The real problem isn’t coding faster.
It’s helping a regular guy who really wants to create, who has ideas, but who’s not technical and not a game developer…
actually ship something good.
Breaking the technical barrier is fine.
But also helping him with creativity, art direction, game design.
And giving him the tools to publish and monetize without it being a nightmare.
The day a random person ships a full AI game,
and thousands of people play it without even knowing it was made with AI,
just because the game is actually good…
That’s the inflection point.
Not before.
Right now we’re still in the “look what I generated” phase.
Not the “holy shit this game is so good” phase.
Excited to be back at Stanford tomorrow for the Stanford × Google DeepMind Hackathon.
@joinarcadeai is supporting the event, and I’ll be joining the judging panel alongside Jessy Tang.
Looking forward to meeting ambitious builders and seeing what they can ship in just a few hours.
Hackathons played a major role in Arcade’s journey, so it feels meaningful to now be on the other side of the table.
See you at Stanford 👋
Join here:
https://t.co/2mZCMq44bK
Remember when creating a game required years of coding, art and design?
Today, I typed an idea and had a playable world in minutes.
Testing out @joinarcadeai and it feels like a glimpse into the future of gaming.
The only thing holding us back now is our imagination.
@MrrRemi 🎮👾🕹️
Arcade AI is becoming one infinite world where you can build, share, and play anything with your friends.
Most AI game tools stop at generating code and assets.
We didn’t.
We built the systems needed to create and play instantly.
No friction.
No waiting.
No broken engine.
1/5
We’re now 12 people working days, nights, and weekends to build a new way to create and play.
Public release announcement coming in July.
Arcade AI is getting real.
4/5