Finst

Vitalik Buterin Says AI Could Push Bitcoin Miners Out of Expensive Locations

Buterin points to the cost gap between Bitcoin mining and AI inference: miners are more sensitive to electricity prices, while AI relies mostly on expensive chips.

Vitalik Buterin Says AI Could Push Bitcoin Miners Out of Expensive Locations

Key Takeaways

  • Vitalik Buterin says Bitcoin mining is about half hardware costs and half electricity costs.
  • According to Buterin, AI inference relies much more on expensive chips than on electricity, so higher power prices matter less.
  • The growth of AI data centers is increasing competition for power and grid capacity, pushing miners and AI companies toward the same locations.

Vitalik Buterin explained on X that Bitcoin mining and AI use energy in different ways. According to the Ethereum founder, the cost of Bitcoin mining is roughly half hardware and half electricity. AI inference, he said, is almost entirely a hardware bill. That difference matters more as both sectors increasingly look at the same power locations.

What Bitcoin Mining Costs

Buterin shared a cost breakdown showing that the total cost per Bitcoin comes out to about $93,700 (€83,300). The biggest expense is electricity: power that goes directly to the machines costs $40,600 (€36,100) per coin according to his table, or 43.3% of the total. Another $2,400 (€2,130) in other electricity costs is added on top of that.

The hardware side also adds up fast. Compute capex, including equipment, cooling, power supply, and financing, comes to about $37,600 (€33,400) per coin, or nearly 40%. Labor, site rent, maintenance, insurance, taxes, and bandwidth make up the rest. That leaves electricity and hardware as very close cost items.

Buterin also pointed out that this estimate is above a recent JPMorgan forecast, which put Bitcoin's production cost at $85,000 (€75,500). The table is also above the current Bitcoin price, which means an average miner does not recover its full costs.

Why AI Works Differently

For AI inference, Buterin says the focus is much more on expensive chips than on power. That means an AI company can probably handle higher electricity prices more easily than a miner can. For Bitcoin miners, cheap electricity remains a much bigger part of the edge.

That contrast sits against a broader backdrop. Bitcoin mining already uses a large amount of electricity worldwide, while the rapid growth of AI data centers is pushing power demand even higher. Gartner expects data center electricity use to rise 26% in 2026 to 565 TWh, mainly because of AI workloads. AI servers themselves are also taking up an ever larger share of that demand.

Effects on Energy Locations

For European crypto readers, the key point is that the fight for power and grid capacity could get tighter. In some regions, data center projects are already being delayed because of power shortages and equipment scarcity. That makes it clear why miners and AI companies are increasingly looking at the same places, even though they use capital in very different ways.

Buterin said some miners already control permitted grid capacity that AI companies cannot build fast enough. At the same time, other players are looking for ways to combine both models. BitMEX founder Arthur Hayes, for example, is working on a Bitcoin-like AI mining network that pays GPU owners. Which side ends up stronger, Buterin said, seems to depend less on Bitcoin itself and more on who can finance hardware the fastest.


Disclaimer: This content is for informational purposes only and does not constitute financial, investment, legal, or tax advice. The information provided may be incomplete, inaccurate, or outdated and should not be relied upon as such. Nothing on this website should be considered a recommendation to buy, sell, or hold any cryptocurrency. Investing in crypto-assets involves risk of loss.