Tripo AI announced $50 million in new funding as it unveiled a new generation of production-ready 3D model architectures designed to generate assets directly in native three-dimensional space. The round was backed by Alibaba and Baidu Ventures, with proceeds set to support continued research into large-scale 3D foundation models and expansion of the company’s global developer platform.
The company, which now serves more than 6.5 million creators and 90,000 developers worldwide, said it has generated nearly 100 million 3D assets to date. Its platform includes subscription tools, creator software, and APIs that allow developers and studios to integrate AI-generated 3D content directly into production workflows across gaming, robotics, manufacturing, and immersive media.
Alongside the funding, Tripo AI introduced new model families, including Tripo H3.1 and Tripo P1.0, marking a shift in how AI systems generate 3D geometry. Traditional approaches often rely on sequential token prediction adapted from language or image models, which can introduce inefficiencies and structural inconsistencies when applied to inherently spatial data.
Tripo AI’s approach instead models geometry directly within a unified three-dimensional probabilistic space. By representing vertices, edges, and polygon faces within a shared spatial field, the system maintains global context and symmetry during generation. This enables more consistent topology and reduces issues such as broken meshes or missing surfaces that can arise from step-by-step generation.
The company said its architecture also improves computational efficiency by enabling parallel processing across the entire geometry field. As a result, production-ready polygon meshes can be generated in as little as two seconds, representing up to a 100x improvement over earlier workflows. These gains are supported by a training dataset of approximately 50 million structured 3D assets.
Tripo H3.1 is designed for high-fidelity outputs suited to industrial design, cinematic assets, and 3D printing, while Tripo P1.0 focuses on lightweight, topology-aware meshes optimized for real-time applications such as game engines, robotics simulation, and extended reality environments.
Looking ahead, the company is also developing Tripo W1.0, an early-stage world model initiative aimed at enabling AI systems to simulate and interact with dynamic spatial environments. Tripo AI believes native 3D representation will become a foundational layer for future AI systems capable of understanding and operating within physical space.
KEY QUOTE:
“Much of today’s generative AI is built around sequences. But three-dimensional space is inherently holistic and symmetric. When geometry is forced into a sequence, artificial structure is introduced. Our approach models shapes directly in native spatial space, allowing structure to emerge coherently.”
Simon Song, Founder And CEO Of Tripo AI
