VAST Raises Nearly $200M and Reveals Its Project Eden World Model Roadmap

QbitAI reports that general-artificial-intelligence company VAST recently completed a combined nearly US$200 million in A+ and A++ funding rounds, with lead investors including Yingce Capital and the China Life Yangtze River Delta Sci-Tech Innovation Fund, and participating investors spanning market-oriented funds, state-owned platforms, industrial capital, and existing shareholders. This is another capital injection following VAST's US$50 million Series A completed in March 2026. Beyond the funding, the company also officially disclosed its world-model roadmap, Project Eden. VAST's core proposition is not simply to do "action-conditioned video generation" or "static 3D scene generation," but to natively decouple the underlying world state evolution from frame rendering: the lower layer maintains a global world state that is cross-temporal and independent of camera viewpoint; the middle layer converts the 3D state into semantic and geometric constraints under a specific viewpoint; and the upper generative rendering layer focuses on texture, lighting, materials, and local dynamic details. VAST believes this architecture can improve on the limitations of single monolithic video models in long-range consistency, environment persistence, and multi-person interaction, so that an object still exists in the underlying state even after leaving the frame, and when a user or AI Agent re-enters, they see a consistent world result. On data strategy, VAST mentions two layers of sources: first, leveraging Tripo's 3D foundation model capabilities to reverse-extract depth, camera pose, and geometric trajectories from large volumes of internet 2D video; and second, synthesizing data through game engines, letting an Agent continuously explore within the engine environment and record training data with precise 3D state annotations. The article also reviews VAST's accumulation across the Tripo series, Tripo Studio, and several open-source 3D projects, framing Project Eden as an extension from "creating all things" to "creating worlds." However, VAST also acknowledges remaining challenges, including the physical evolution of more complex scenes and autonomous state maintenance that does not rely on external annotations or engine assistance. Overall, this is an industry story combining a large funding round, accumulated AI 3D technology, and the unveiling of a world-model roadmap.