There is a number that has been haunting me since I first read it in a BeInCrypto report: the human brain runs on 20 watts. An AI supercomputer, by contrast, draws 42.2 megawatts. That is not a comparison; it is an indictment. We have built machines that think by burning enough electricity to power a small town, and we have built an entire financial narrative around the companies that own those machines. But the more interesting story is not the machines. It is the people who own the power they consume.
I spent six months in 2017 auditing whitepapers for governance tokens, chasing the illusion of permissionless consensus. I learned that the most valuable asset in crypto was never the code. It was the story. And the story now being told about Bitcoin miners is one of the most compelling narrative shifts I have witnessed in twenty-five years of watching this industry. It is a story about how the people who spent a decade being dismissed as energy parasites have become the most sought-after landlords in the artificial intelligence gold rush.
The Grid Is the New GPU
Let me be precise about what changed. For years, the bottleneck in AI was compute. NVIDIA shipped GPUs, and whoever had the most GPUs won. That era ended sometime in late 2025, when Microsoft CEO Satya Nadella admitted something extraordinary: his company owned processors it could not plug in. There was no electricity available to power them. The constraint had shifted from silicon to substations.
The data confirms this. The International Energy Agency estimates that data centers consumed 485 terawatt-hours in 2025, and AI-specific facilities grew by 50% in that single year. By 2030, that number is projected to triple. Meanwhile, the median time from interconnection request to commercial operation for a new grid connection now exceeds five years. Five years. In an industry where model generations turn over every six months, a five-year wait is an eternity.
This is where the narrative pivots. Bitcoin miners have spent the last decade doing something that now looks prescient: they secured grid interconnection rights. They built power purchase agreements. They erected physical sites with transformers, cooling systems, and security perimeters. They solved the problem that Microsoft, Google, and Amazon are now desperately trying to solve. And they did it while the world was mocking them for wasting electricity on a digital collectible.
Chaos is just data waiting for a story. The chaos of the 2022 crypto winter, when miners were selling rigs at fire-sale prices, was actually the moment when the most important infrastructure assets in the AI economy were being consolidated into the hands of a few publicly traded companies.
The Cost of a Megawatt
VanEck's analysis is the clearest articulation of this shift. Retrofitting an existing mining facility costs between $3 million and $4 million per megawatt. Building a greenfield data center costs between $10 million and $12 million per megawatt. That is a 60-70% cost advantage. In a market where capital efficiency determines survival, that advantage is not a moat; it is a fortress.
Publicly traded miners have already signed AI contracts worth more than $70 billion. Let me repeat that number because it deserves emphasis: $70 billion. Core Scientific, IREN, Marathon Digital, Riot Platforms — these companies, which were trading at distressed valuations two years ago, have become the upstream suppliers to the AI economy. They are not building models. They are not writing algorithms. They are selling something far more scarce: the right to consume electricity.
I have audited enough balance sheets to know that contract announcements are not the same as revenue. But the structural logic here is sound. The miners hold the physical assets that AI companies cannot build quickly. The interconnection rights, the substations, the power purchase agreements — these are not replicable in a quarter. They take years to acquire. The miners have them now.
The Neuromorphic Red Herring
A significant portion of the source analysis is devoted to neuromorphic computing — chips designed to mimic the human brain's efficiency. Intel's Hala Point, with its 11.5 billion artificial neurons across 1,152 chips, sits in a Sandia National Laboratory as a research prototype. The lab director admitted they do not know how to map a large language model onto it. BrainChip, a commercial neuromorphic company, reported $700,000 in customer receipts in Q1 2025 against $5.3 million in operating outflows. Rain AI failed to raise a $150 million round and explored a sale in 2025.
This is not a technology sector; it is a graveyard of good intentions. The reason matters for the miner thesis: if neuromorphic computing were viable, the energy bottleneck would eventually dissolve, and the miners' power assets would lose their scarcity premium. But the evidence says otherwise. No neuromorphic or analog chip system has ever trained or run a frontier model in production. The von Neumann architecture remains dominant, and it is voracious. The energy bottleneck is not a temporary condition; it is the defining constraint of the AI era.
We build bridges in the silence after the noise. The noise around neuromorphic computing has faded, and in the silence, the bridge between miners and AI has become visible.
The Hidden Tension in the Dual Income Model
The contrarian angle here is not whether miners will succeed in pivoting to AI. The evidence suggests they will, at least partially. The contrarian angle is what this pivot costs the Bitcoin network itself.
Miners are not simply adding a second revenue stream. They are making a choice about how to allocate a finite resource: their electricity. Every megawatt dedicated to AI hosting is a megawatt not dedicated to Bitcoin mining. The $70 billion in AI contracts comes with service level agreements — minimum availability requirements that prevent miners from simply switching back to mining when Bitcoin prices spike. The flexibility that made mining attractive — the ability to power down and sell power back to the grid — is constrained by contractual obligations to AI customers.
This creates a structural tension that the market has not priced. Bitcoin's security model depends on hash rate. If the largest miners redirect their power to AI hosting, the network's hash rate growth slows, or even reverses. The narrative of Bitcoin as a secure, decentralized network is partially built on the assumption that miners have an economic incentive to keep mining. That assumption is now conditional.
There is a deeper issue. The $70 billion in contracts may include memoranda of understanding and non-binding letters of intent. The actual binding commitments may be a fraction of that headline number. And even where contracts are binding, the margins may be thinner than expected. Miners are not cloud service providers. They lack the operational expertise, the customer relationships, and the reliability track record of an Equinix or a Digital Realty. They are landlords, not operators. The contracts may pay them rent, but they will not pay them the premium that comes from being a full-stack AI infrastructure provider.
The Geopolitical Dimension
There is another layer to this story that the original analysis touches on but does not fully develop: the geopolitical bifurcation of AI compute. China's LineShine supercomputer, which topped the global rankings for the first time since 2017, runs entirely on domestic CPUs. It uses no Nvidia chips, no American technology. The US export controls have not stopped China from building world-class compute; they have accelerated China's path to self-sufficiency.
This matters for miners because it means there will be two separate AI compute ecosystems, each requiring electricity. The miners' assets are jurisdiction-bound. A miner in Texas cannot serve a Chinese AI company. A miner in Kazakhstan cannot easily serve a US hyperscaler. The grid interconnection rights that make miners valuable are also what make them geographically locked. In a world of bifurcated AI supply chains, the miners' power assets are valuable, but their value is contingent on which side of the geopolitical divide they sit on.
Narrative is not what we say, but what remains. What remains after the AI hype cycle cools is a set of physical assets — substations, transformers, power agreements — that will be valuable regardless of which model wins, which chip architecture dominates, or which country leads the AI race. The miners own the one thing that every AI company needs: the right to consume electricity.
The Risk of Narrative Fatigue
The market has already begun to price this story. Mining stocks have re-rated from crypto beta to infrastructure plays. The question is whether the narrative can sustain itself. I have seen this pattern before. In 2020, DeFi was going to replace traditional finance. In 2021, NFTs were going to replace art. In 2022, the metaverse was going to replace reality. Each narrative collapsed when the underlying metrics failed to match the story.
The AI-miner narrative is different in one crucial respect: it is backed by physical assets and signed contracts. But it is also vulnerable to the same forces that killed previous narratives. If AI capital expenditure slows, if GPU prices fall, if cloud service prices decline, the miners' contracts will be renegotiated or abandoned. The $70 billion headline number will shrink to something more realistic. The stocks will correct.
In the void, we find the architecture of trust. The void between the narrative and the reality is where the actual value will be determined. The miners who can execute on their AI transformation — who can retrofit their facilities on time and on budget, who can meet their SLA obligations, who can build the operational expertise to serve enterprise customers — will emerge as the infrastructure backbone of the AI economy. The miners who signed MOUs and hoped for the best will be exposed.
The Takeaway
The 20-watt brain is a beautiful metaphor, but it is also a trap. It suggests that efficiency is the answer, that we will eventually build machines that think like humans without consuming megawatts. That may be true in fifty years. It is not true in the next five. In the next five years, AI will consume an ever-growing share of the world's electricity, and the people who own the right to that electricity will be the kings of the new economy.
Bitcoin miners spent a decade being told they were wasting energy. They were actually building the most valuable infrastructure asset of the 21st century: grid access. The question is not whether they will profit from this pivot. The question is whether they can survive their own success — whether they can manage the transition from mining to hosting without destroying the very network that gave them their original purpose.
Liquidity flows where meaning is clear. The meaning of a Bitcoin miner has changed. It is no longer just a guardian of a decentralized ledger. It is a landlord to the most powerful technology companies on Earth. The story is still being written, and the next chapter will be determined not by the price of Bitcoin, but by the execution of a $70 billion promise.
