Ethereum co-founder Vitalik Buterin defended the stance against building large AI data centers, stating that slowing down the construction pace would give society more time to address cybersecurity and other emerging risks without significantly hindering technological progress.

In a post on November 8th, Buterin opposed the strong criticism directed at the anti-data center movement over the past week. He believes that improvements to the AI algorithm may ultimately enable existing computing facilities to accomplish tasks that would otherwise require much larger data centers.

At the time of his remarks, Bitcoin mining companies, including Riot Platforms and Cipher Digital, were expanding into the field of artificial intelligence infrastructure, while the community and policymakers expressed concerns regarding electricity demand, water usage, and the speed of data center construction.

Vitalik Buterin : AI Progress does not require endless data centers

Buterin questions the claim that "restricting new data centers will inevitably hinder AI developers from achieving their technical goals."

His argument revolves around algorithm efficiency, that is, as developers improve their methods, software can complete certain tasks with fewer computing resources.

He stated that the capabilities that would require larger-scale facilities in two years may be achievable with today's computing infrastructure within four years, or even sooner.

"Even if anti-data center advocates achieve extreme success, nothing close to the end of the world will happen," said Buterin.

This co-founder of Ethereum did not call for the immediate closure of existing facilities, nor did he propose any specific government bans. His viewpoint supports the idea that it is acceptable to slow down construction during the period when AI technology continues to develop.

He believes that society needs more time to strengthen its defenses against cyber security threats and potential biological risks associated with increasingly powerful artificial intelligence systems.

Buterin says that scientific research depends on a variety of factors beyond the computing power of AI. Therefore, slowing down the development of AI by one year does not necessarily mean that scientific progress will also be delayed by one year.

"The slowdown in the ultra-large-scale expansion of computing power seems to be a good thing overall," he wrote, expressing this view without relying on predictions that "advanced AI will inevitably lead to catastrophic consequences."

Bitcoin mining companies are entering the AI custody business through transactions worth billions of dollars.

At the time of this discussion, listed Bitcoin mining companies were aggressively expanding their AI custody services, hoping to generate revenue by utilizing existing land, power access, and computing infrastructure.

On November 4th, Riot Platforms disclosed that it has signed a 15-year lease for a facility in Corsicana, Dezhou, covering a critical computing capacity of 910 megawatts.

The agreement is expected to generate $28 billion in contract revenue during the initial period. Riot also indicates that, including other leases, the contracted data center capacity has reached 1,151 megawatts, corresponding to approximately $38 billion in long-term revenue.

The company sold 20,264 bitcoins in the first nine months of 2026, raising approximately $1.5 billion, which was mainly used to fund the development of data centers.

As previously reported by crypto.news, Riot has been reducing its Bitcoin holdings while expanding its infrastructure business.

Cipher Digital has also followed a similar path. The company's financial report on November 2 confirmed that after starting to collect rent from Black Pearl, the company recorded high-performance computing rental revenue for the first time.

Cipher extended the contract with Barber Lake for another 20 years, bringing the project's anticipated contract revenue to approximately $9.1 billion. The company disclosed that its total development pipeline amounts to about 5.9 gigawatts.

Texas data center restrictions raise questions about the AI project

Texas has become a focal point of opposition to new data centers, especially in communities that are concerned about electricity prices and water supply.

In an order on September 21st, Governor Greg Abbott requested that the state environmental regulatory agency suspend the licensing approval of data center projects, pending officials' review of their potential impact on power reliability and water resources.

After Gina Hinojosa emerged victorious in the gubernatorial election on November 3rd, these restrictions have once again sparked new political uncertainties.

Hinojosa proposes to suspend the approval of new data centers until legislators establish community protection measures, which include rules regarding energy costs, water supply, and public participation.

On November 4th, after the election results were announced, the stock price of Cipher fell by nearly 19%, with investors assessing the possibility of further development restrictions.

However, the suspension of approval measures proposed by Hinojosa have not yet become new state-wide laws. After her government takes office, it will still need to decide how the promised restrictions will be implemented.

Buterin has previously followed AI on Security and Privacy.

This comment by Buterin continues his previous work regarding the security and privacy risks associated with the more powerful AI system.

In October, he tested privacy-focused AI tools, which combine Tor routing, anonymous payments, and local processing prompts to reduce the sharing of personal information with external AI services.

This experiment utilized separate network connections to handle conversations, additional traffic to obscure requests, and local software to rewrite prompts before transmitting them.

Buterin acknowledges that such settings can lead to processing delays and cannot eliminate all possible privacy vulnerabilities.

His statement on November 8th put forward an independent argument regarding the construction speed of the AI infrastructure. He stated that concerns over cybersecurity and future biological risks are sufficient to justify giving more time for preparatory measures, even if this means a slower expansion of large-scale computing facilities.