AI Crypto Wheel

The AI crypto wheel picks one of twenty artificial intelligence tokens at random. The category covers compute marketplaces, model networks and data protocols, and it is the loosest label on this site — a token can qualify by doing AI work or merely by saying so.

The loosest label on this site

Every wheel here inherits its category from a market-data provider, and most of those categories describe something you could verify. A privacy coin either obscures transaction graphs or it does not. A base layer either settles its own transactions or it does not. Artificial intelligence is not like that. It is a claim about what a project intends, and intent is cheap.

Some of what lands on this wheel is doing recognisable work: renting out GPU capacity, routing inference requests, paying for labelled data, coordinating model training across machines that do not trust each other. Some of it adopted the label during a narrative cycle, when anything with those two letters in its description traded better, and simply never took it off. The tag cannot distinguish between them. Nothing automated can, which is exactly why the distinction is left to you rather than smuggled into a ranking.

Why the category is so crowded

Sector labels in crypto behave like tickers on a stock exchange during a mania: they attract flow, and flow attracts more claimants. A project that describes itself as infrastructure gets read as boring; the same project described as decentralised compute for AI workloads gets read as early. The incentive runs one way, so the population of the category grows faster than the population of things doing the work.

Classification also lags badly. A protocol pivots, the description changes, the tag persists for months. A project shuts down its model network and keeps the label because nobody removes tags. The list refreshes daily, which keeps the market-cap ordering current, but it cannot audit a project's actual output — no data feed does that, and any that claimed to would be selling an opinion.

What a random pick is good for here

More than usual, oddly. In a category where the labels are unreliable, the ordering you would naturally impose on it — by familiarity, by whichever name you saw most often this month — is not obviously better information than a coin flip. Attention in this sector is heavily path-dependent, and what you have heard of is largely a function of who bought advertising.

So if the intent is to read about one of these properly rather than to buy anything, a wheel is a defensible way to choose which. It has no marketing budget and no preference. It will hand you something obscure as readily as something you already recognise, which is the one advantage randomness genuinely has over intuition.

What the wheel has no idea about

Whether the compute network has customers. Whether the model weights are real or a wrapper around somebody else's API. Whether the token has any mechanical relationship to usage, or is simply the thing that was sold to fund the work. Whether the team shipped anything in the last year.

These are the questions that separate the projects on this wheel from one another, and none of them is answerable from a ticker and a rank. The wheel knows twenty names. Treating its output as a shortlist of one is fine. Treating it as a signal is a category error, and this page would rather say so plainly than dress the wheel up as research.

What is on this wheel

Market data provided by CoinMarketCap — CoinMarketCap. Rank, ticker and name only; this site stores no prices.

This list is a judgement about what belongs in the category, not a market-cap ranking, so it carries no date. See how the lists are chosen.

Questions

Which AI tokens are on this wheel?

All 20 of them are printed under the wheel: TAO, NEAR, ICP, RENDER, VVV, FIL, INJ, VIRTUAL, FET, VELVET, KITE, THETA, GRT, AKT, DEXE, TRAC, BAT, GLM, TAG and EGLD. Stablecoins and wrapped tokens are left off, and you can replace the whole list in Custom mode.

Do these tokens actually do anything with AI?

Some do: there are genuine compute marketplaces and inference networks in this category. Others attached the label during a narrative cycle and kept it. The tag cannot tell the two apart, and neither can a wheel.

Why is the AI category so crowded?

Because it pays to be in it. Sector labels in crypto attract attention and listings, so projects adopt whichever one is working, and the classification lags reality by a long way. Treat membership as a claim, not a credential.