One in six uploads, one dollar in ninety: CGTrader's AI supply glut found no buyers

CGTrader, a 3D model marketplace with more than two million models listed for sale, opened its catalog to AI-generated assets about a year ago ^[1][2]. Its own year of sales data, covering June 2025 through May 2026, now shows one in six uploaded models is AI-generated while those assets produce one dollar in every ninety of revenue and 2.6% of sales ^[3][4][5][6].

Do the arithmetic the report leaves implicit. One dollar in ninety is about 1.1% of revenue ^[7]. One in six uploads is 16.7% of the catalog inflow, so AI supply is running roughly fifteen times ahead of AI revenue ^[8]. That is not a slow ramp. That is a catalog filling up with inventory that does not move.

The gap between the two published shares is the more useful number for anyone pricing a supply strategy. AI assets are 2.6% of sales but only about 1.1% of revenue, which means an AI sale brings in roughly 43 cents for every dollar the average sale on the platform brings in ^[9]. The AI listings that do clear are the cheap ones. Volume without price is not a business.

The company is unusually blunt about it. "AI is entering the catalog rapidly, but buyers aren't yet opening their wallets for it," the report said ^[10]. On experience, 5% of CGTrader's customers tried an AI model and found it worked well, against 20% who tried one and found it inadequate ^[11][12] - a four-to-one negative ratio among people who actually sampled the goods ^[13]. CEO Dalia Lasaite told Fortune that buyers are looking for high quality and as a result tend to prefer human-created 3D models, at least at this point ^[14].

Treat this as one marketplace's data, not a law. A 2025 Stanford study found that participants given a marketplace with both AI-generated and human-made art gravitated toward the AI pieces, with generative images rapidly increasing on the platform ^[15]. Stated preference runs the other way: a Pew poll last year found half of Americans liked a painting less after learning AI made it ^[16], and Pew reported on Tuesday that 52% of American adults are more concerned than excited about greater AI use in daily life, up from 38% in 2022 ^[17] - a 14 point move in four years ^[18].

There is a parallel in software. Dennis Zhang, a professor at Washington University in St. Louis's Olin Business School, measured app launches after the release of the coding agents Claude Code and Codex, and after controlling for other variables put the effect at about a 160% increase in apps by April 2026 versus two years earlier ^[19][20][25]. The number of apps with more than ten reviews dropped significantly over the same stretch, which he cautions is not a causal result ^[21][22]. His read: products helped by AI in production look less attractive than the human-crafted comparison set, though they still create utility for the market ^[23]. His broader claim is that consumers, not just workers, re-pivot toward the dimensions where humans matter more ^[24].

Watch three things. Whether CGTrader's AI upload share keeps climbing while the revenue share stays pinned near 1%, which would tell you sellers are responding to production cost rather than demand. Whether that 5%-versus-20% quality split narrows in the next annual cut, since model quality is the only variable that changes the conclusion. And whether marketplaces start labelling or demoting AI listings in ranking, which is what happens when unsold inventory begins to cost the platform search relevance.