Every time a new tech trend takes off, crypto finds a way to slap that word on a token and sell it. Right now that word is AI. Hundreds of AI-themed crypto tokens launched during the last cycle, and my next guest has been digging through the data to see what is actually real and what is not. Her verdict? Most of them are headed to zero. Joining me again is Alice Liu, Head of Research at CoinMarketCap. Alice, welcome back to Wall Street to Mena.
Great to be back. Thanks for having me.
What actually makes an AI token a real AI token?
Anyone can put AI in the token name. But the difference between the real one and the fake one is whether AI plays a role in the system itself, or if it is just wearing a t-shirt. In crypto, we generally put AI tokens into three buckets. Number one is infrastructure — for example decentralised computing. Number two is applications — utilities built on top of that infrastructure. And the third one is the costume bucket. Anything with AI or GPT or agent merely in the name — that is the buzzword. They do not have AI actually in the product or the token usage itself.
Can you explain the difference between a real project and a meme coin using AI clothing?
Yes. If you can find the token in the product itself, it is AI. If you cannot find it, that is a meme coin. On CoinMarketCap, under the AI and big data category, we have 945 tokens. Within those, only 81 tokens have over $20 million market cap and over $1 million daily liquidity. Those are the ones generally being traded and accepted as part of the AI narrative.
Are good AI projects being punished alongside bad ones?
You are right about this. The market always prices hype before fundamentals. So when AI as a category is traded as a whole in the crypto market, the bad ones poison the good ones. Whenever there is bad news on any so-called AI token being a scam, it drags the whole industry down. Another reason is that traditional finance investors already have an AI trade — they buy Nvidia and the AI supply chain stocks. So good projects in crypto AI end up trading at a discount.
The whole category is worth around $19 billion on CoinMarketCap. How much of that value is real?
I would say one tenth of those tokens are meaningful. So basically this sector has a good few billion in real value within a category priced at 21 billion. There are good ones at the top, but the majority of it is just a very long tail of lottery tickets.
Is this history repeating itself from 2017 ICOs or 2021 metaverse tokens?
In 2017 we had ICOs. Two years later, the majority of those tokens disappeared. But Ethereum was part of the ICO era. Chainlink stayed as well. Using that example — AI as a sector is getting started and the market is pricing the hype. A few years later, the good ones will stay. AI crypto is a top-heavy sector with a long tail of lottery tickets. But right now it is very easy to distinguish which ones are the good ones and which ones might be falling away.
If someone is holding an AI coin right now, what should they check?
Five things. First: does the token have a job — if the token is removed does the product stop working? Second: float market cap versus fully diluted value — token unlocks will change what the market cap looks like in future. Third: real liquidity — does it actually get traded? Fourth: signs of life beyond social followers — check GitHub commits, active network participants, active nodes. Fifth: the correlation test — does the token pump because Nvidia is pumping, or because the network itself had genuine activity? That last one is the one I like the most.
Thank you so much Alice for being here with us today.
Thank you very much.
