Binance co-founder Changpeng Zhao, commonly known as CZ, believes artificial intelligence (AI) projects within the crypto industry are too “profit-oriented.” Taking to social media platform X on Wednesday, CZ called out developers for focusing too much on their tokens while neglecting the actual usefulness of their AI agents.
“Too many AI agent developers focus too much on their token and not enough on the agent’s usefulness,” Zhao wrote. “I recommend making a really good agent first. Only launch a token when there is product-market fit.”
The crypto AI market capitalization increased by 322% in the fourth quarter of 2024, from $4 billion to $15.5 billion. According to Coingecko data, the market valuation rose even further in 2025, climbing to $21 billion as of April 2.
Through comments on CZ’s post, which the former Binance CEO had limited to those he follows, community members on X engaged in a debate over the necessity of AI project tokens.
One user questioned Zhao about the difficulty research and development teams face in securing funding without token sales. They argued that many investment firms prioritize quick returns, pushing projects to launch a token prematurely because they do not have support from venture capitalists.
CZ said token-based fundraising is still important to projects but insisted that developers must first be committed to building real products. “Raising money using tokens is a powerful use case for crypto,” he continued. “But they should still focus on building after they raise, not just selling the tokens.“
Raising money using tokens is a powerful use case for crypto, but they should still focus on building after they raise, not just sell the tokens.
— CZ 🔶 BNB (@cz_binance) April 2, 2025
Satoshi Club, a group of market analysts, asked Zhao what he thought about “95% of AI agents” not needing a token, to which he replied: “Maybe 99.95%.”
Influencer Crypto_BN compared AI project tokens to stock ownership, arguing that tokens can help builders raise money, but he reckoned that without a solid product to support it, a token is “just a memecoin.”
Many others were of the sentiment that most AI-related crypto projects capitalize on market hype to launch tokens for a quick profit.
“At first, we expected the AI agent space to be entirely focused on providing real value and utility to users. However, it quickly turned into just another way to promote their tokens,” said one X user who quoted CZ’s post.
However, one commenter suggested that even if the token initially focuses on raising capital, a high-utility AI agent with a well-structured token system could attract users organically by rewarding early adopters and contributors. “It’s all about balance,” they concluded.
This week, several discussions over AI token legitimacy have taken over social media, partly started by the news of a new blockchain project backed by venture capital firm Andreessen Horowitz (a16z), claiming it could create a project that is “better than Bitcoin.”
Ambient, an AI-focused blockchain co-founded by Travis Good, is a proof-of-work network with AI-driven features designed to appeal to Bitcoin miners. “It’s a useful proof-of-work network, which we don’t think anyone has ever done well in crypto,” Good said.
According to the co-founder, Bitcoin’s encryption mechanisms are becoming outdated and could be “completely obsolete within five years.” Speaking at the ethDenver conference, he propounded that this presents significant problems for Bitcoin miners, who have invested billions in specialized hardware.
“You’ve got people who’ve invested billions of dollars in hash power for securing a network in ASICs,” Good surmised. “And the question is like, where do they all go?”
Ambient recently secured $7.2 million in seed funding from a16z’s crypto accelerator program and Delphi Digital, a venture capital firm known for its investments in AI-crypto crossovers.
According to Delphi Digital venture associate Alex Golding, the current issue of AI in crypto is centralized control. “Everyone in crypto is currently using centralized AI to power their apps, to power their frontend,” he said, arguing that this deprives users of transparency regarding AI models and the data they rely on.
Ambient claims it will rid the market of these concerns through a mechanism known as “verified inference,” which relies on miners to fact-check AI-generated responses. Good cautioned investors that without such verification, AI models could be manipulated by hostile actors.
“If you don’t have verified inference, you’re guaranteed to get rugged,” he said. He went even further, claiming that “nation-state actors are going to poison your model and just do fun stuff, like we saw with Lazarus,” referencing the North Korean hacking group.
Cryptopolitan Academy: Tired of market swings? Learn how DeFi can help you build steady passive income. Register Now