Manifold AI (流形空间) Raises $140M Pre-A, Hits Unicorn Status in Under a Year

Manifold AI (流形空间) Raises $140M Pre-A, Hits Unicorn Status in Under a Year

Manifold AI, a Beijing-based startup focused on world models—AI systems that simulate and predict real-world dynamics—has raised approximately 1 billion yuan ($140M) in a Pre-A funding round. The company achieved unicorn valuation within just 12 months of founding, placing it among the fastest-growing AI labs globally.

This funding event updates the player map for the world model substrate. Manifold enters a segment where most capital has historically flowed to US labs like World Labs and autonomous-driving-world-model startups. The $140M pre-A round—nearly unheard-of for a seed-stage company—signals that Chinese capital is aggressively betting on world models as the next frontier, distinct from pure large language models. The velocity from founding to unicorn (one year) matches the fastest-ARR-ramp pattern seen among top-tier LLM companies like Moonshot AI, but applied to a different technical thesis.

What makes this notable is the round's stage and size. A pre-A round at unicorn valuation compresses what would normally be three or four capital cycles into one. This capital-compression arc reflects a market dynamic where investors, wary of missing the next paradigm shift, are pricing world-model tech as a structural bet rather than a vertical application. The key open question: will Manifold's world model approach outperform the simulation-centric routes pursued by Google DeepMind and Open AI's ongoing work, or will it converge on similar outcomes?

#WorldModels #ManifoldAI #ChinaAI #PreAUnicorn #CapitalCompression #SimulationAI