CoinMarketCap says AI-token flows are concentrating into a handful of liquid names
The AI & Big Data label spans about 945 tokens, but fewer than 10% clear basic size and volume screens, per CoinMarketCap research.
CoinMarketCap research is framing the AI-token trade as a rotation away from broad “narrative beta” and into a small set of liquid, catalyst-driven names. The data point doing the work is a liquidity screen that leaves most of the 945-token category effectively untradeable at scale.
CoinMarketCap: AI Label Stops Lifting the Whole Category
The AI label is no longer acting like a basket trade. CoinMarketCap Head of Research Alice Liu described a market that is now separating “AI” branding from actual market participation, with flows concentrating into a small subset of names that can carry size.
“More capital, fewer names. The AI tag is free. The market is finally charging for it,” Liu said in an email interview. Mechanically, that means the marginal dollar is showing up where liquidity is already there, not spreading evenly across the long tail of AI-tagged tokens.
CoinMarketCap’s framing is that this is a regime shift in how the theme trades. AI can still be a top-of-mind narrative, but the trade is moving from buying exposure to the label to underwriting project-specific catalysts and usage, with liquidity acting as the gate.
Liquidity Screen Explains the Dispersion: 945 Tokens, <10% Tradable at Scale
CoinMarketCap’s AI & Big Data category spans about 945 tokens with a combined market cap of $18.5 billion, described as roughly 0.7% of the $2.64 trillion crypto market. The category is huge by token count and small by capital share, which is exactly the setup where narrative rotations can look broad while being narrow in practice.
Liu’s liquidity filter makes that narrowness explicit. Fewer than one in 10 of the ~945 AI & Big Data tokens clears both a $20 million market-cap threshold and $1 million in daily trading volume, she said. Market cap is price times circulating supply, and daily trading volume is the last 24 hours of dollar turnover. Put together, those thresholds are a blunt proxy for whether a trader can put on meaningful size without paying for it in slippage.
That screen also explains why performance is dispersing instead of lifting the whole category. The examples Liu cited are all names that already sit on the “tradable” side of the line: Venice AI’s token was described as nearly doubling over the past month (as of Sept. 10) to about a $1.1 billion market cap, Bittensor’s TAO was described as up 24% over 30 days to a $2.8 billion market cap, and NEAR was described as gaining 27% over the week covered by the interview. TAO was described as the largest “pure-AI token” in CoinMarketCap’s data.
The catch is that the same interview also claims AI ranked among the top three crypto narratives over both seven-day and 30-day windows, without publishing the underlying narrative-tracking methodology or the exact measurement windows. That missing primitive matters because “AI is top-three” can mean very different things depending on whether it is based on social mentions, volume share, price momentum, or some blended score.
How I’d Translate This Into a Tradeable Watchlist
What AI crypto tokens: capital concentrates in Tells Me
The threshold that matters here is not whether “AI” stays a hot narrative, it is whether breadth improves above Liu’s “fewer than one in 10” liquidity screen. If the share of AI & Big Data tokens clearing both $20 million market cap and $1 million daily volume stays pinned under 10%, the category will keep trading like a handful of liquid leaders plus a long tail that cannot absorb size.
The real test is whether the cited leaders keep outperforming the $18.5 billion category aggregate, because sustained divergence is what “more capital, fewer names” looks like on a chart. If CoinMarketCap publishes a clearer breakdown of how it ranks narratives across the seven-day and 30-day windows, that would tighten the signal. If not, the practical takeaway remains simple: the AI label may still pull attention, but only liquidity and catalysts decide which tokens can actually take the next dollar.
