Bitcoin miners are directing billions of dollars and scarce power capacity to artificial intelligence as BTC’s slump pushes mining profitability near historic lows.
data from crypto slate Bitcoin is trading at around $64,000, indicating it is almost 50% below its October peak, while increasing network competition and low transaction fees continue to weigh on miners’ profits.
These pressures are making AI infrastructure increasingly attractive as data center customers can pay significantly higher prices for reliable power and long-term capacity. This gives miners another way to monetize power assets that are becoming less profitable if they focus solely on Bitcoin.
Some of the industry’s biggest operators have already converted equipment, signed multi-year computing contracts, and in some cases backed away from investing in new Bitcoin mining equipment.
But Andre Dragos, head of research at Bitwise Europe, said: crypto slate Miners may be making shifts at the wrong point in the cycle.
He argues that expectations for AI computing demand, including demand generated by autonomous agents, may take longer to materialize than current investments suggest. At the same time, he believes Bitcoin is nearing the end of its decline.
This combination creates potential timing issues. Just as a recovery in Bitcoin prices restores the economics of mining, miners could spend the next year putting capital and power capacity into AI.
Dragosh therefore believes that some miners who pivot today may regret their decision within the next 12 months.
Mining economics make AI pivot difficult to resist
The economic conditions facing Bitcoin miners make it difficult to heed Dragosh’s contrarian calls today.
The April 2024 Bitcoin halving reduced the block subsidy from 6.25 BTC to 3.125 BTC, reducing the number of new coins miners receive to secure the network. The recent drop in Bitcoin prices has further exacerbated that pressure by reducing the dollar value of these rewards.
Along with the price of Bitcoin, miners’ profits also deteriorated. VanEck reports that daily miner revenue is down nearly 40% year-over-year, with the 30-day average currently at about $28.5 million.
At the same time, the increasing hashrate of the network has intensified competition for shrinking rewards.
The hash price, which measures a miner’s revenue per unit of computing power, has fallen to around $30 per petahash/second/day after reaching an all-time low earlier this year, leaving older and less efficient machines below break-even depending on power costs.
For some miners, the gap has become large enough to justify abandoning further expansion of the Bitcoin fleet.
Core Scientific, once one of the largest Bitcoin miners, said it no longer plans to spend on new equipment to maintain or expand its hash rate. Instead, we intend to direct more power to high-density computing while extracting cash from our existing mining fleet.
The second quarter results demonstrate why this deal became attractive.
Colocation revenue jumped to $136.7 million from $10.6 million in the year-ago period and accounted for approximately 83% of total revenue. Bitcoin self-mining revenue fell 66% to $21.5 million, down from about 80% in the same period last year to just 13%.
The gap in profitability widened further. Core Scientific reported gross margins from colocation of 59% in the quarter, while gross margins from its self-mining business were negative.
Meanwhile, Core Scientific is not alone in this pivot.
Earlier this year, CoinShares announced that public miners had announced over $70 billion in AI and high-performance computing contracts. The company estimated that publicly traded Bitcoin miners could derive around 30% to up to 70% of their revenue from AI by the end of 2026.
MARA Holdings CEO Fred Thiel summed up the economic situation in a recent interview:
“You get a much higher return per electron doing it for AI than you do for Bitcoin mining.”
This transition does not include converting Bitcoin mining machines to AI hardware. ASICs built for Bitcoin cannot run workloads processed by GPUs.
What miners can reclaim is access to power, grid connections, land and data center infrastructure – assets that are becoming increasingly valuable as AI developers compete for sites that can support large computing clusters.
Why AI bets are harder to solve
Still, the risks of Dragosh’s paper are magnified by how expensive the transition will be.
CoinShares estimates that Bitcoin mining infrastructure costs about $700,000 to $1 million per megawatt, while AI facilities cost about $8 million to $15 million per megawatt.
This difference means that miners moving toward high-performance computing are making significantly larger and longer-term capital commitments than those expanding their Bitcoin mining sites.
These investments also come during one of the largest technology infrastructure spending cycles on record.
The Bank for International Settlements estimates that the five largest hyperscalers could spend more than $1 trillion in AI-related capital spending between 2025 and 2026.
The BIS has warned that intense competition among technology companies could lead to excessive investment if commercial profits cannot keep up with spending.
Disappointing AI revenues could therefore delay new infrastructure development, dampen demand for additional capacity, and make it more difficult to finance projects built around continued growth in computing demand.
Dragosh is not arguing that the demand for AI will disappear. While he believes AI will be transformative, he believes a transition of this scale may take longer to mature than investment cycles indicate.
In that scenario, infrastructure supply could expand faster than commercial demand, compressing the unusually attractive economic conditions currently drawing Bitcoin miners to the sector.
If Bitcoin mining profitability also recovers at the same time, the impact will be even more significant.
Rising BTC prices and improving hash prices could narrow the revenue gap between mining and AI, but operators who have already committed billions of dollars, power capacity, and infrastructure to long-term computing projects will have less flexibility to respond.
The risk therefore lies in how many options miners are willing to give up while pursuing businesses that offer better economics than they do today.
(Tag translation) Bitcoin
