I spent four months auditing the smart contracts of EtherTrust back in 2017, staring at a reentrancy vulnerability that could have drained $4.2 million. I chose to publish the exposé rather than take a private bug bounty. That decision taught me something that still holds: the most dangerous vulnerabilities are not always in the code. They are in the incentives. Today, as I read about Nvidia connecting GPU companies with data center operators in the Nordics, I see a similar pattern. The physical layer of the AI economy is being built, and the question is not whether it is efficient, but who controls it and for what purpose.
Context: A New Infrastructure Playground
Nvidia’s move is deceptively simple. The company is brokering relationships between GPU cloud providers (like CoreWeave, Lambda Labs) and Nordic data center operators who offer cheap renewable energy and natural cooling. The stated goal: “sustainable, cost-effective AI infrastructure.” This is a classic ecosystem play. By reducing the total cost of ownership for AI compute, Nvidia locks in demand for its GPUs. But the subtext is more profound. The Nordics, with their abundant hydro and wind power, have long been a haven for Bitcoin miners seeking energy arbitrage. Now, the same logic is being formalized for AI. The difference is that Nvidia is acting as the architect, not just a vendor.
This is not just about chips. It is about the physical layer of the blockchain—or rather, the physical layer of the compute economy that underpins all decentralized systems. Every AI model, every smart contract, every DeFi protocol relies on a data center somewhere. For years, crypto evangelists like me have preached decentralization. But the hardware remains centralized in the hands of a few hyperscalers. Nvidia’s Nordic initiative is a bid to reshape that reality, but in a way that consolidates power around its own ecosystem. “Trust is earned, not mined,” I often say. Here, the trust is being earned by Nvidia’s ability to deliver cheaper compute, but the mining is still happening on someone else’s land.
Core: The Technical Heartbeat of the Physical Layer
The technical details matter. Efficient cooling, especially liquid cooling, is a prerequisite for the next generation of GPUs (the B200, the GB200). These chips generate so much heat that traditional air cooling is inadequate. The Nordics offer natural cooling for much of the year, but liquid cooling is still required for peak loads. The combination of low ambient temperature and efficient liquid cooling can reduce energy consumption for cooling by up to 40%. That is not trivial. For a 100 MW data center, that could save millions annually. This is the same type of infrastructure that Render Network, Akash, and other decentralized GPU marketplaces rely on. But here is the twist: those networks are permissionless. Anyone can contribute compute. Nvidia’s model is permissioned. You need to be an approved partner. That is a fundamental difference.
I have seen this pattern before. In 2020, during DeFi Summer, I analyzed the Compound governance working group. The code was open, but the governance was captured by large token holders. The result was a system that claimed to be decentralized but was effectively controlled by a few. Now, Nvidia is building a similar structure at the physical layer. The infrastructure is more efficient, but the control is centralized. The “Soul in the machine” is missing. The soul is the ethos of permissionless participation. Without it, the infrastructure becomes just another tool for the powerful to extract value.
From my experience auditing protocols, I know that the most resilient systems are those that distribute power. The Nordics data center play is a brilliant technical move, but it is a step backward for the decentralization of compute. It creates a new single point of failure: Nvidia’s ecosystem. If the company decides to change its terms or prioritize certain partners, the entire network of GPU cloud providers becomes vulnerable. We saw this with AWS during the crypto winter: they delisted certain services, and projects died. Nvidia is building a similar dependency.
Contrarian: The Pragmatism Test
But let me play the contrarian. Perhaps I am too idealistic. The market is in a bull phase, and everyone is FOMOing into AI. The need for cheap, sustainable compute is urgent. The Nordics offer a tangible solution. My own community, the one I built during the NFT crash, has members who are GPU farmers. They are desperate for lower electricity costs. If Nvidia can deliver that, they will flock to it. The immediate benefit is real: lower carbon footprint, lower costs, more compute for researchers and developers. The contrarian question is: does this help the greater good, even if it centralizes power?
I think of the “Proof of Humanity” project I helped build in 2021. We used non-transferable tokens to verify identity. It was small, only 500 members, but it was ours. We controlled the keys. We trusted each other. That is the essence of decentralization: not just the code, but the human relationships. The Nordics data center is the opposite. It is a giant, efficient machine that will be managed by a few executives at Nvidia and their partners. The users—the AI developers, the crypto miners, the DeFi protocols—will have no say. They will be tenants, not owners.
I recall the “Long Winter” manifesto I wrote in 2022, analyzing why 80% of top projects failed. The common thread was not bad technology, but misaligned incentives. The founders built beautiful protocols but centralized the governance. The same is happening here. Nvidia is building a beautiful infrastructure, but the governance is centralized. The risk is that when the next bear market comes, or when Nvidia decides to pivot its strategy, the entire ecosystem built on top of this infrastructure will collapse.
Takeaway: Vision Forward
So what is the alternative? I am not saying we should reject Nvidia’s initiative. Efficiency is good. But we must demand that the physical layer of the AI economy be built on principles of openness and community ownership. Imagine a cooperative of GPU miners and data center operators, governed by a DAO, that owns the infrastructure in the Nordics. The energy is cheap, the cooling is efficient, but the control is distributed. That is the vision I hold. “Conscience over consensus.” The consensus of the market is that Nvidia’s move is a win. But my conscience tells me that we must build infrastructure that cannot be switched off by a single company. The code is not enough. We must also build the infrastructure with heart. The next step is not to admire Nvidia’s efficiency, but to replicate it in a decentralized way. That is the soul we need in the machine.
