China-based generative-AI start-up VAST has completed Series B and Series B+ financings totalling approximately $446 million, or roughly RMB 3 billion, led by Matrix Partners China, in one of the largest AI application funding rounds to close anywhere in the world so far this year.
The financing brings VAST's cumulative fundraising to roughly $446 million within less than six months, according to industry reporting — a pace of capital deployment that places the company's ambitions closer to those of a foundation-model developer than a conventional creative-software start-up.
VAST operates Tripo, a generative AI platform capable of producing three-dimensional models directly from text prompts and images, a capability with applications spanning gaming, film and entertainment production, industrial and product design, additive manufacturing, and increasingly, the simulation environments used to train embodied and robotic AI systems.
The financing syndicate reads as a deliberate bet by China's gaming, media and entertainment industries on generative 3D as a strategic capability rather than a novelty. Industrial participants including Perfect World, BlueFocus, Thundersoft and 37 Interactive Entertainment joined the round alongside financial investors such as CICC Capital, CMC Capital Partners, CDH VGC and existing backer Primavera Capital.
That mix of strategic and financial capital matters. Generative 3D sits at an earlier stage of commercial maturity than text or image generation, and the participation of established gaming and media companies signals that at least some of China's largest content businesses see near-term integration paths for VAST's technology within their own production pipelines, rather than viewing the category purely as a speculative long-term bet.
The scale of the round also reflects a broader recalibration under way in AI venture investing: as text-based large language models mature into a more commoditised layer, investor attention — and capital — is increasingly rotating toward adjacent generative categories, including 3D, video and embodied AI, where technical differentiation remains sharper and defensible moats appear more achievable.
For VAST, the challenge ahead lies in converting an enormous capital advantage into durable product usage across gaming studios, designers and manufacturers who have historically relied on specialised, purpose-built 3D modelling tools rather than AI-generated assets. If generative 3D follows the adoption curve seen in text and image generation over the past several years, VAST's early capital lead could prove decisive — but the technology must first prove it can meet the production-quality bar that professional content pipelines demand, a threshold considerably higher than the novelty-driven demos that have characterised much of generative AI's public reception to date.
