Tripo AI Raises $50 Million to Advance Native 3D Spatial Generation Models

New updates have been reported about Tripo AI.

Tripo AI has secured $50 million in new funding to accelerate development of its 3D foundation models, as it shifts AI 3D generation from sequential token prediction to native spatial geometry modeling. The round, led by strategic investors including Alibaba and Baidu Ventures, is intended to scale research, expand the company’s global developer platform, and strengthen its role as core infrastructure for programmable 3D content.

The company now serves more than 6.5 million creators and 90,000 developers worldwide, who have collectively produced nearly 100 million 3D assets via subscriptions, creator tools, and APIs integrated into production workflows. This installed base positions Tripo AI to monetize demand from gaming, robotics, manufacturing, and immersive media customers seeking scalable, production-ready 3D asset generation.

At the technical core of this strategy are the new Tripo H3.1 and Tripo P1.0 model families, which natively represent vertices, edges, and faces in a unified three-dimensional probabilistic space instead of linear sequences. By reasoning about geometry and topology globally, the models aim to eliminate structural inconsistencies, reduce topology errors, and cut generation times to as little as two seconds for production-grade polygon meshes.

Tripo H3.1 targets high-fidelity use cases such as industrial design, cinematic assets, and high-resolution 3D printing, while Tripo P1.0 is optimized for real-time graphics and robotics, generating topology-aware meshes that are directly usable in game engines and XR environments. Both are trained on roughly 50 million high-quality 3D assets, giving the company one of the largest structured polygon mesh datasets in the market.

Founder and CEO Simon Song argues that native spatial modeling better reflects the holistic and symmetric nature of 3D space, addressing a fundamental representation gap in current generative AI systems. Looking ahead, Tripo AI is also developing Tripo W1.0, an early-stage world model designed for AI systems that can simulate and interact with dynamic environments, positioning the company as a potential infrastructure layer for future spatial computing and embodied AI applications.