Changpeng Zhao echoed a growing sentiment among analysts, urging AI projects to take payments in existing tokens and only launch their own if they achieve significant scale.

In his recent post on X, Changpeng Zhao expressed an “unpopular opinion,” stating that not every AI agent needs its own token as agents can collect payments in existing tokens for their services. “Launch a coin only if you have scale. Focus on utility, not tokens,” he said.

According to CoinMarketCap, the total market cap of AI & Big data tokens has declined by approximately 22% in the past 30 days, currently at $27.44 billion. Many AI agents are down by double digits in the last month, with Virtuals Protocol (VIRTUALS) declining by 42%, Render (RENDER) by 30%, and Near Protocol (NEAR) by 26%.

However, these declines are likely driven by the broader crypto market downturn rather than a lack of utility, with macroeconomic factors—most notably Trump’s tariffs—playing a key role. The uncertainty over potential U.S. expansions of existing restrictions recently sent stocks of AI chipmaker companies tumbling, with Nvidia’s stock taking a significant 6% decline, adding further pressure on AI-related crypto tokens.

Although the decline in AI tokens’ market cap is driven primarily by macro factors, other experts echoed Zhao’s sentiment about the lack of utility of AI tokens.

For example, Coinbase research analyst David Han noted that the recent price rallies in AI coins were driven mainly by hype rather than actual utility. He attributed the hype around these tokens to the broader focus on the AI industry.

On-chain sleuth ZachXBT also claimed that 99% of AI cryptos are scams, saying that while memecoins acknowledge their lack of inherent value, AI projects often market their tokens as if they had utility.