Strong Density
Saw Unity announce their AI Open Beta this morning. I have been out of gaming and the game engine business for a long time, but I still remember meeting their ex-CEO JR in SF more than ten years ago. Good that Unity finally revealed the AI parts of what they have been building. Less good that, like my last post about density, there is no density and no space of imagination in what they showed.
One reply on the thread caught my eye. Studio Zamudio One wrote, “I was interested until I saw the $210 a month! Probably better to vibe code this in three.js.” That one line tells me Unity is not ready yet.
Step back for a second. A game engine traditionally provides two things, a framework which is mostly design thinking (ECS for Unity is a good example), and ease of use, the part that actually saves developers time. That ease of use piece is also what we focused on at Cocos Creator. I personally acquired the team that built it and integrated it into Cocos at the time. The underlying rendering part matters less and less every year. Back in the early mobile era we ported Ogre to iOS just to get 3D rendering running at all, and that kind of work was central back then. Now it sits much closer to the bottom of the stack.
And the rendering story is shifting under everyone’s feet anyway. World models, real-time generative game video, and real-time AI-driven render carry much more meaning right now than another iteration on the legacy pipeline that has been around for decades. Unity should not be sticking to the old legacy at this point. They have had decades to see this coming.
The AI part of building a game requires much more density than what I am looking at here. Treating an amateur fan the same as a pro team is a mistake. Anything that can be done in five minutes with Claude and three.js belongs in the amateur lane. Pro teams need something else entirely, since they already have the tools and the experience, with Claude Code/Codex already part of their daily workflow. A game engine AI built for them has to provide enough density, and not just enough, strong density, the kind that actually creates value at the scale a studio operates.
No content-building platform with AI aimed at the average creator is going to succeed in the long run. Look at what just happened with Sora, OpenAI stopped it. The average creator never pays enough to sustain that kind of platform, and the content that comes out of it rarely sticks. The game industry itself makes a lot of money though, and that money sits with the studios actually shipping the games. Real value lives in the dense end of the market, with the people who already know what they are doing.