The text-to-3D startup closed the largest funding round ever for an AI-3D company on July 21, 2026, reaching a $1.5 billion valuation on the back of 12x year-over-year revenue growth and 12 million registered users.
A minute and a few dollars. That is now roughly what it costs to generate a print-ready 3D model in FBX, OBJ, GLB, or STL format using Meshy, a startup whose technology accepts a text prompt, a photograph, or a hand-drawn sketch and returns something a game studio or product designer can actually put to work. On July 21, investors decided that capability is worth $1.5 billion, backing the company with nearly $400 million in a Series B round led by IDG Capital, Matrix Partners China, and Monolith Management, with existing shareholders including Sequoia China, Granite Asia, BAI Capital, and Source Code Capital also participating.
It's the largest round ever raised by a company built specifically for AI-generated 3D. That distinction matters because it signals something the venture community has been watching closely: text-to-3D is no longer a novelty demo. It is a production pipeline.
The speed of Meshy's ascent is worth sitting with. At the Game Developers Conference in March 2026, the company announced it had hit $30 million in annual recurring revenue, doubling in three months, and had crossed 10 million global users. By July, ARR is growing roughly 12x year over year, registered users have climbed past 12 million, and the platform has generated more than 100 million models in total. Those are not metrics a company manufactures to dress up a fundraise. They are the kind of numbers that close a $400 million round.
Alongside the funding announcement, Meshy launched what it calls the Meshy 3D Agent, billed as the first AI agent purpose-built for 3D creation. Where previous versions of the platform took a single input and returned a model, the 3D Agent is designed to behave more like a collaborator: you can have a back-and-forth conversation, refine the output iteratively, and arrive at something production-ready without any prior 3D experience. No Blender skills required. No freelance artist on retainer.
That last point is the one the gaming and product design industries are processing right now. A skilled 3D modeler working traditionally can spend days, sometimes weeks, on a single complex asset. Meshy's Pro plan runs $20 a month, and each generation costs 20 credits from a pool of 1,000 , meaning you can produce roughly 50 models per month on that plan alone. The economics are not a close call for studios churning out environment assets or manufacturers iterating on product prototypes.
Why this round landed when it did #
The timing reflects more than Meshy's own growth. The broader AI tooling market has been looking for proof that generative models can move up the stack from text and images into spatial computing. Text-to-image crossed that threshold years ago with Midjourney and Stable Diffusion. Video followed, slowly, with tools like Sora. 3D has been the harder problem: spatial coherence, watertight geometry, texture mapping that holds up at game-engine resolution. Competitors including Tripo AI and Rodin from Deemos have been chasing the same target, but Meshy's user numbers and revenue trajectory suggest it's moved furthest into real production use.
The investors backing this round are not making a bet on the technology in isolation. They're making a bet on distribution. Twelve million registered users is a meaningful moat in a market where switching costs are low and every model needs retraining on new data. The more models Meshy generates, the better its training data gets, which is a flywheel that pure compute spending cannot replicate quickly.
Gaming is the obvious first market, and it's clearly where Meshy has found early traction. Meshy Labs, the experimental incubator it announced at GDC, produced Black Box: Infinite Arsenal, the first title built entirely through the platform. But the use cases extend well beyond interactive entertainment. Architects generating massing models from sketches, e-commerce teams creating product visualizations without photo shoots, manufacturing engineers iterating on part geometry before committing to tooling , all of them share the same bottleneck Meshy is removing. Frankly, any workflow that currently ends with
Also read: The White House says Moonshot AI trained Kimi K3 on banned Nvidia chips routed through Thailand • The US Army burned through its AI token budget in months and every enterprise should take note • Glow raises $180 million at a $1.2 billion valuation to secure the AI agents that CrowdStrike wasn't built to see