Ethereum co-founder Vitalik Buterin said at ETH Taipei that blockchain should not be treated as the final objective, but as one tool for solving problems. In his remarks, he revisited the direction of the crypto industry over the past decade and argued that the values that matter most sit below the technology itself: censorship resistance, openness, privacy, and safety.
Buterin also pointed to a recurring wealth cycle in crypto and warned against getting carried along by each round of "who suddenly got rich this year." Whether the focus was NFT and DeFi in the last cycle or AI in the current tech market, he said the long-term goals worth pursuing have not changed.
Buterin frames the debate around "CROPS"
The center of the talk was a framework he called "CROPS": Censorship and Capture Resistance, Open Source, Privacy, and Safety.
Those, he said, are the actual ends. Blockchain is only one technology that can help achieve them. He added that other technologies can also reach the same goals, while many blockchain-based implementations do not make the world more resistant to censorship, more open, more private, or safer.
His conclusion was straightforward: define the goal first, decide what problem needs to be solved, and then choose the right technology, rather than taking blockchain as a hammer and treating every problem like a nail.
Looking back at the NFT and DeFi boom
Buterin then looked back at the crypto bull market from roughly five or six years ago. He described an industry filled with expensive NFTs, large and flashy crypto conferences, Lamborghinis, and a strong visible wealth effect. In that period, many products were evaluated through a very simple logic: if it was on-chain, it was seen as innovative.
Personal data was supposed to go on-chain. Identity was supposed to go on-chain. Social relationships were supposed to go on-chain. NFTs were often treated as good largely because they "existed on the blockchain." Buterin joked that if the exact same NFT product had been launched by Twitter or Google at the time, many people in the room might have disliked it.
He said DeFi had a similar issue. Even when an oracle was highly centralized, or a cross-chain bridge was controlled by a multisig that could contain a backdoor, the market could still assume the product had value simply because it carried the "DeFi" label.
In his view, that mindset took hold for two reasons. One was "number go up" as prices kept rising. The other was that the industry had few tools available then, and blockchain was the most visible hammer, so every problem started to look like a nail.
"Crypto first" thinking has largely run its course
Buterin said plainly that this year in particular, the old habit of centering everything on blockchain has "basically reached its end." One practical reason, he said, is that blockchain is no longer the most eye-catching new technology in the global tech sector.
If someone walks into a university, an Asian technology company, or a large enterprise today and asks what technology is attracting the most attention, he said, the answer would obviously be AI rather than blockchain.
That does not mean the problems crypto originally set out to solve have lost their relevance. In fact, Buterin argued the opposite: after AI’s rapid rise, censorship resistance, privacy, safety, and trustworthy information sources matter even more than they did 10 years ago.
AI makes centralization and de-anonymization risks more visible
To explain the point, Buterin said that if a workflow depends heavily on proprietary large language models from companies such as OpenAI and Anthropic, it contains a centralized choke point. A provider can change policy, limit access, or stop offering the service altogether.
He also said the use of AI itself can create substantial privacy risk. Model providers can see query content, usage records may be analyzed, and users have limited ability to verify whether the training of a next-generation model could accidentally retain passwords or other sensitive data.
A larger issue, he said, is de-anonymization. Buterin referred to the early internet line, "On the Internet, nobody knows you're a dog." Today, he said, AI can infer a person’s identity from very small amounts of information. Even if a user tries to remain anonymous or changes the language used in a post, AI may still identify that person again through writing patterns, metadata, and posting times.
He added that even encrypted messaging tools such as Signal protect message content, but do not necessarily fully protect metadata such as who contacted whom, how often they communicated, and when messages were sent. As AI analysis gets stronger, he said, metadata alone may be enough to reconstruct large parts of a person’s life.
Blockchain and cryptography together remain central to the long term
Buterin also said that in the mainstream technology world today, the most advanced form of "freedom tech" may not even be blockchain, but open-weight LLMs.
That, however, does not mean blockchain is disappearing. He said blockchain will remain for the long run and that many people are underestimating its importance. More precisely, what matters ahead is not only blockchain itself, but the combination of blockchain and cryptographic technology.
Five years ago, many people might have described themselves simply as "blockchain people," he said. Today, the Ethereum ecosystem is also helping push cryptographic technologies such as zero-knowledge proofs and fully homomorphic encryption, or FHE. He said future work could extend to more advanced cryptographic directions, including indistinguishability obfuscation.
In his view, blockchain and cryptography will strengthen each other, and tools such as zero-knowledge proofs are already expanding what blockchains can do.