CoinShares said in a quarterly report that a megawatt allocated to artificial intelligence can generate about $1.5 million in annual revenue, compared with roughly $500,000 for bitcoin mining, underscoring a widening economic gap that is reshaping parts of the mining industry.

The report comes as mining margins remain under pressure. CoinShares said the average cost to produce one bitcoin has risen to $75,500, while bitcoin was trading at $58,400 in the cited comparison. That leaves some operators mining at a loss under current conditions.

The changing economics are already affecting company decisions. Core Scientific canceled an order for mining machines worth $41.9 million, according to the report, while Keel has stopped bitcoin mining. At IREN, AI-related cloud revenue reached $70.5 million and exceeded revenue from mining, pointing to a business mix that is shifting beyond hash-rate generation.

CoinShares also said the global bitcoin hash rate has fallen 27% since October 2025 to 850 EH/s. At the same time, electrical infrastructure is becoming harder to secure. The report cited 225 moratoriums on data centers in the United States, highlighting growing constraints around grid access and data-center development.

Even so, the move from mining to AI infrastructure remains incomplete. CoinShares said companies converting mining sites into AI facilities are commanding higher valuations, but only $1.1 billion in annualized revenue has been realized from $100 billion in announced contracts. Separate industry reporting has also described AI and high-performance computing contracts as offering stronger revenue per megawatt, while requiring far higher upfront capital than traditional mining sites.

Why It Matters

The report points to a structural shift in how crypto mining assets are being valued. If power-connected mining sites can be redeployed into AI or high-performance computing infrastructure, the sector starts to look less like a pure bitcoin production business and more like a digital infrastructure business built around scarce electricity, land and interconnection capacity.

That has broader implications for both crypto and AI. For miners, weaker economics after rising production costs increase the pressure to diversify revenue. For the AI buildout, mining operators may offer one of the faster paths to energized capacity in a market where new power and data-center approvals are increasingly constrained.