For most of the generative-AI boom, 3D was the corner of the field that investors admired but rarely bankrolled at scale. That changed on July 21, when Meshy said it had closed a Series B of nearly $400 million at a $1.5 billion valuation — the largest disclosed financing to date for a company built specifically around AI-generated 3D, and Meshy's first publicly disclosed valuation.
The raise, announced from Silicon Valley, was backed by what the company called "a group of leading global investors," with every existing backer participating. Reporting around the deal identified IDG Capital, Matrix Partners China and Monolith among the leads, with prior investors including Granite Asia, Sequoia China, BAI Capital and Source Code Capital following on. Proceeds, Meshy said, will go primarily toward research and development and global market expansion.
What makes the number striking is the company it puts Meshy in. Text-to-3D has produced plenty of impressive demos and very little revenue; Meshy is pitching itself as the exception. As of July 2026, it says annual recurring revenue is growing roughly 12x year over year, with more than 12 million registered users and over 100 million models created. "In a field where most companies have yet to reach meaningful revenue," the company argued in its announcement, "Meshy has built its lead on real commercial performance."
From "looking right" to "ready for real production"
The core of Meshy's pitch is a claim about usefulness, not novelty. Founded by MIT PhD Ethan Hu, the company frames its progress as moving AI-generated 3D from "looking right on screen" to "ready for real production" — models that drop directly into game engines or come off a 3D printer as physical objects. From a single line of text or one image, Meshy says a usable model can be generated in about a minute. Work that "once took specialized skills, expensive software, and weeks," it said, "now takes about a minute and a dollar."
Alongside the funding, Meshy shipped a batch of product updates aimed at that production gap. The headline release is what it calls Meshy 3D Agent, billed as "the world's first AI agent for 3D creation," which turns a conversation, sketch or photo into a print-ready model in formats including FBX, OBJ, GLB and STL. The company claims a slicer success rate of up to 97% for 3D printing and exports compatible with Unity, Unreal and Blender. Two other features target the same seam between digital and physical: Auto Split, which breaks a model into watertight, reassemblable parts for printing, and Smart Topology, an in-house model that generates clean geometry in about 10 seconds with a controllable polygon count from 100 to 15,000. An 8K texture tool is rolling out soon.
The customer list is the other half of the argument. Meshy says teams inside five of the world's ten largest technology companies by market capitalization or valuation are building with its tools, without naming them. Its disclosed customers and partners span gaming — Nexon, NetEase Games and 37 Interactive Entertainment — the 3D-printing hardware world, including Bambu Lab, Creality, Elegoo, FlashForge and xTool, and consumer and cultural names such as Hugo Boss and Sweden's national museum of art and design.
Why it matters
Meshy's round lands in a category that has abruptly caught up with the rest of generative AI. For two years, text, image, audio and video models absorbed the capital while 3D lagged, held back by messy training data and the hard requirement that outputs be not just pretty but geometrically valid. That gap is now closing fast, and the money is following.
The competitive field is suddenly crowded and well-funded. Tripo AI raised roughly $200 million across Series A+ and A++ rounds in June 2026, on top of an earlier $50 million tranche backed by Alibaba and Baidu Ventures, and now claims more than 6.5 million creators. Deemos, maker of the Rodin and Hyper3D tools that rank highly for photorealism and production-ready output, and Kaedim occupy adjacent niches. Looming over all of them is Luma AI, whose roughly $900 million raise dwarfs the specialists and signals that the largest bets in spatial AI may come from companies treating 3D as a stepping stone to "world models."
That framing is the real stakes. Meshy itself gestures at it, describing 3D as "the starting point" toward experiences users can "be immersed in, genuinely feel, and remember afterward." As AR/VR hardware, robotics simulation and game pipelines all converge on the need for cheap, plentiful, physically accurate 3D content, whoever owns the generation layer sits upstream of a very large market. A $1.5 billion valuation for a still-young specialist is a bet that the layer will be worth owning — and that commercial traction, not just benchmark scores, will decide who owns it.
What to watch
The open questions are durability and disclosure. Meshy's 12x revenue growth is off an undisclosed base, and its "five of the ten largest tech companies" claim is unverifiable without names, so the next signal to watch is whether any of those enterprise relationships surface publicly or convert into deeper contracts. Equally important is how the specialists fare against Luma-scale players and the big model labs, any of which could fold 3D generation into a broader platform. If Meshy can keep translating a research lead into paying production use — in games, on print beds and inside AR pipelines — this round will look early. If 3D becomes a feature rather than a category, $1.5 billion will look like the top.
“Work that once took specialized skills, expensive software, and weeks now takes about a minute and a dollar.”— Meshy, company statement