crypto

Vitalik Says Slow the Servers Down. He's Not Wrong.

Vitalik Buterin has never been someone who keeps his opinions to himself, and his post on 8 November 2026 defending opponents of large AI data centres is...

8 November 2026
Vitalik Says Slow the Servers Down. He's Not Wrong.

Vitalik Buterin has never been someone who keeps his opinions to himself, and his post on 8 November 2026 defending opponents of large AI data centres is exactly the kind of statement that makes the crypto world uncomfortable. Not because it is radical, but because it is reasonable, and reasonable does not sell well when there are billions of dollars of infrastructure deals being signed. The Ethereum co-founder looked at the wave of criticism aimed at anti-data-centre movements and decided, publicly, that the critics were wrong. What followed was a genuinely interesting argument about algorithmic efficiency, societal risk, and how fast we actually need to build the machines.

What Did Buterin Actually Argue?

The first thing to understand is what Buterin did not say. He did not call for existing facilities to be shut down. He did not propose a government ban. He did not predict doom. What he said, calmly and without apocalyptic framing, was that slowing down the construction of new data centres is "quite net good," and that the opposition to such movements has been emotionally overwrought.

His argument rests on algorithmic efficiency, which is the idea that as AI developers improve their software, the same tasks can be performed using fewer computing resources. He made a specific point: capabilities that might require much larger facilities in two years could become achievable with today's existing infrastructure within four years, possibly sooner. That is not a small claim. It reframes the entire urgency argument. If the technology will get there regardless, just slightly later, then the costs of slowing construction, in terms of genuine lost progress, start to look a lot smaller than the industry would have you believe.

He also made a point about scientific research that tends to get lost in these conversations. AI computing power is one input into scientific progress, not the only one. Slowing down compute infrastructure by a year does not automatically mean scientific advances are delayed by a year. That is a distinction the data centre lobby has little incentive to make, which is precisely why it matters that someone with Buterin's profile is making it.

The Security Argument That Does Not Get Enough Air

Beyond the efficiency logic, Buterin raised something more serious: the need for society to strengthen its defences against cybersecurity threats and potential biological risks associated with increasingly capable AI systems before those systems scale further. This is not scaremongering. It is project management. If you are building something that creates new attack surfaces faster than you can close them, slowing down the build is rational.

Buterin has form here. In October 2026, he had been testing privacy-focused AI tools combining Tor routing, anonymous payments and locally processed prompts, specifically to reduce the personal information being shared with external AI services. That experiment, which involved separate network connections, additional obfuscating traffic and local software rewriting prompts before transmission, showed both how seriously he takes these risks and how early-stage the defensive infrastructure really is. He acknowledged the setup introduced delays and did not eliminate every vulnerability.

The gap between offensive capability and defensive readiness is real, and it is widening. Buterin's position is that buying time to close that gap, even at the cost of slightly slower infrastructure expansion, is a trade worth making. That is not an anti-technology argument. It is an argument for building things in the right order.

Riot and Cipher: The Money Moving in the Opposite Direction

While Buterin was writing his post, the Bitcoin mining companies were signing contracts that pointed in entirely the opposite direction. Riot Platforms announced on 4 November that it had locked in a 15-year lease covering 910 megawatts of critical computing capacity at its Corsicana facility in Texas. The estimated contracted revenue from that deal alone is 28 billion dollars. Including other leases, Riot's total contracted data centre capacity reached 1,151 megawatts, representing roughly 38 billion dollars in long-term revenue. Those are not numbers that make a company receptive to arguments about slowing down.

Cipher Digital followed a similar path. Its earnings release on 2 November confirmed the company's first high-performance-computing lease revenue after rent payments began at its Black Pearl site. Cipher extended its Barber Lake lease to 20 years, bringing expected contracted revenue from that site to approximately 9.1 billion dollars, with a development pipeline totalling around 5.9 gigawatts. These companies have sold their Bitcoin holdings, committed to the infrastructure model and are not going back. Riot sold over 20,000 Bitcoin in the first nine months of 2026, raising approximately 1.5 billion dollars, almost entirely to fund data centre development. The pivot is complete.

Texas Is Where the Fight Is Happening Right Now

Texas has become the focal point of this argument in the physical world, and it is messier than most coverage suggests. Governor Greg Abbott issued a directive on 21 September ordering state environmental regulators to pause permits for new data centre projects while officials reviewed their potential effects on electricity reliability and water supplies. That was already a significant intervention. Then on 3 November, Gina Hinojosa won the gubernatorial election, and the political ground shifted further.

Hinojosa has proposed suspending new data centre approvals until legislators establish community protections covering energy costs, water supplies and public participation in planning decisions. The day after the election, Cipher shares fell nearly 19% as investors processed what that could mean for future development. For context, that is a dramatic single-day move driven entirely by the prospect of meaningful regulation, which tells you exactly how much of these companies' valuations is built on the assumption that the permitting environment stays permissive.

It is worth noting that Hinojosa's moratorium has not yet become law. Her administration would need to design and implement the restrictions after taking office, which means there is still considerable uncertainty about the form they will take. The fall in Cipher's share price was a reaction to possibility, not reality. But the direction of travel in Texas is clearly towards more scrutiny, not less, and Buterin's public statement arriving in this context is not coincidental.

Is Anyone Listening to the Communities Actually Affected?

The electricity and water concerns are not abstract. Texas communities near these facilities have been raising alarms about the pressure on local grids and water resources for some time, and those concerns are what drove Abbott's September directive in the first place. When a Republican governor in a state that broadly welcomed business without much interference starts pausing permits, something real is happening on the ground.

Buterin did not specifically reference these community concerns in his post, but his argument supports the space to take them seriously. If a one-year slowdown in compute scaling costs little in terms of genuine scientific progress, as he argues, then the question becomes: why are we treating infrastructure expansion as an emergency that overrides everything else? The communities living next to these facilities, paying the electricity bills and watching their water supplies being stressed, are asking exactly that question. They tend not to get quoted in the same publications as the chief executives who are announcing 38-billion-dollar revenue contracts.

The Verdict

Buterin is right, and the reaction against him illustrates the problem. Whenever someone argues for measured pace in AI infrastructure, they get framed as obstructing progress. That framing serves the people with the largest financial interest in unlimited expansion, and it has no particular obligation to accuracy. The algorithmic efficiency argument is legitimate, the security argument is legitimate, and the community impact argument is legitimate. None of them require you to believe AI will end the world. They just require you to accept that "as fast as physically possible" is not automatically the right speed for everything.

The companies pivoting from Bitcoin mining to AI hosting are doing what companies do: following the money, signing long-term leases and making the political environment that suits them. Buterin posting a calm, reasoned defence of the people pushing back is a small counterweight to that momentum. Whether Texas's new administration turns those concerns into durable policy is the real question now. The billions are already committed. The contracts are already signed. The only variable left is whether the regulation arrives in time to mean anything.

Share:XFacebookWhatsApp