The metaverse failed because world-building was treated as a content problem, not an infrastructure problem. In the metaverse era, world-building was a manual content pipeline: brittle tools, asset fragmentation, endless exports, and iterations. You can’t bootstrap a new medium when creating a basic artifact is a week-long ordeal. World models are the correction. They treat worldbuilding as infrastructure: a model that can generate a coherent 3D environment, keep it consistent as you move through it, and let you edit and expand it without restarting from scratch. Marble is a clean example of what this should look like. You feed it text, images, video, or rough layouts, and you get a navigable 3D world you can iterate on: expand scenes, refine structure, combine worlds, and export outputs you can actually use downstream. That’s the point: not a metaverse you visit, but a world-building layer you can build on.