Tripo AI Raises $150M Series A3 for 3D Foundation Models and World Model Research

Tripo AI, a Beijing-founded company building AI-native 3D generation technology, completed a $150 million Series A3 financing round on July 2, 2026, according to a GlobeNewswire press release. The raise came roughly one month after the company closed approximately $200 million across two earlier extensions — its Series A+ and A++ — in June 2026. In total, Tripo AI has raised approximately $400 million in 2026 alone, establishing it as one of the best-funded companies in generative 3D content.

The Company and Founders

Tripo AI was founded in 2023 by Simon Song (also written Song Yachen), who previously worked in the CEO’s office at SenseTime — the large Chinese computer vision company — and co-founded Minimax, one of China’s so-called “AI tiger” unicorns. The company is registered as “Vast” in the Cayman Islands and operates primarily from Beijing.

The founding thesis was that 3D content creation was a significant bottleneck in gaming, robotics, and virtual environments, and that large-scale generative models could reduce the cost and time required to produce production-quality 3D assets by orders of magnitude. Tripo positioned itself as a builder of foundation models for 3D — the 3D equivalent of what Stable Diffusion or Midjourney did for 2D imagery.

Products and Models

Tripo’s core product is its 3D generation platform, which generates textured 3D assets from image or text inputs. Its flagship open-source model, TripoSR, was developed in partnership with Stability AI and can generate a high-quality 3D model from a single image in 0.5 seconds, per SiliconANGLE.

In March 2026, Tripo launched Tripo H3.1 and Tripo P1.0, two next-generation models:

  • H3.1 is a high-fidelity model focused on structural precision — suitable for products, vehicles, architectural components, and objects where geometry accuracy matters in manufacturing and CAD workflows.
  • P1.0 targets physics-aware generation, producing assets that behave realistically in simulation environments — relevant for robotics training and interactive entertainment.

The platform also introduced 8K Texture Generation, which produces physically-based rendering (PBR) textures at production resolution, and Segmentation V2, which automatically decomposes a generated model into labeled, editable parts — a key feature for game pipelines that need to rig, animate, or swap components.

Exports support USD, FBX, OBJ, and STL — the standard formats for game engines, VFX pipelines, and manufacturing software.

Project Eden: World Model Research

In parallel with its 3D generation work, Tripo AI introduced Project Eden — a world model research initiative — alongside the June 2026 funding, per VoxelMatters. Project Eden is described as a persistent, multiplayer world model that breaks from existing paradigms by decoupling the underlying world state from visual rendering.

Where most “world models” in AI research either simulate state or render visuals as a combined process, Project Eden separates those two layers: the state of the world (physics, object relationships, entity positions) is maintained independently from how that state is visualized. The result is a system where the same simulated environment can be rendered in different visual styles, or rendered at different fidelity levels, without changing the underlying simulation.

The company positions Project Eden as serving three use cases:

  1. Interactive entertainment — persistent multiplayer environments for game developers who want to generate rather than hand-author worlds
  2. Embodied AI research — simulation environments where robotic agents can train in procedurally generated, physically realistic worlds
  3. Multi-agent evaluation — environments for testing AI agents at scale without relying on real-world deployment

Project Eden is currently a research preview, not a generally available product.

Series A3 Investors

The $150 million Series A3 round drew a diverse investor base across four industry verticals, per Unite.AI:

  • Automotive: Geely Capital (the investment arm of Geely, which owns Volvo and Polestar)
  • Gaming: 4399 Network, Tanwan, and Giant Network — all established Chinese gaming publishers with large content pipelines
  • Strategic: Fosun Capital, Orinno Capital
  • Financial: CoStone Capital, Addor Capital, T-Capital, Muhua Tech Ventures
  • Existing shareholders increasing: INCE Capital and Genesis Capital

The automotive participation is notable. Geely’s involvement suggests industrial manufacturers see value in AI 3D generation for vehicle design, digital twins, and potentially AR/VR visualization — workflows that currently require expensive human artists and CAD engineers.

The gaming publisher participation is more straightforward: companies like 4399 and Giant Network operate large game studios that consume massive volumes of 3D assets. If Tripo’s models can produce production-quality assets at scale, the cost and time reduction for their content pipelines is direct.

Context: A Rapid Fundraising Pace

Three rounds totaling roughly $400 million in four months is an unusually compressed fundraising timeline, even by 2026 AI standards. The pace suggests one of two things: either Tripo is finding traction quickly enough that investor demand is outpacing its ability to plan traditional round structures, or the company is strategically acquiring investor relationships across multiple sectors to build a broad strategic ecosystem before a larger raise or IPO.

3D Printing Industry notes that the Series A3 capital will be used for AI 3D and world model research teams, algorithm development, infrastructure, and global product expansion — the same stated use as the June rounds. The continuity of stated purpose across rounds suggests the company is simply scaling faster than any single round can accommodate.

The Market

Generative 3D is earlier-stage than generative 2D. Text-to-image models reached commercial adoption in 2022-2023; text-to-3D models are still working through the quality and reliability curve that image models completed. The gap matters: gaming, architecture, film VFX, and manufacturing all require 3D assets, and the manual production cost is significant. The total addressable market for 3D content creation software exceeds $10 billion annually by most estimates.

Tripo’s closest public competitors include Meshy, CSM AI, and Shape.e-style approaches from larger AI labs. The company’s open-source presence via TripoSR and its partnerships with Stability AI give it developer ecosystem credibility that purely closed competitors lack. Whether that translates into a durable platform position — the kind that a sustained world model research effort requires — depends on whether Project Eden becomes a genuine computing environment or remains a research demonstration.

ChatForest reviews AI companies and funding events from publicly available sources. We do not invest in or receive compensation from the companies we cover.