BTC Current Affairs Security

Bitcoin Miners' Billion-Dollar AI Gamble: When the Timing Goes Wrong

2026-07-29

Bitcoin miners are hemorrhaging money. After the 2024 halving, they're losing roughly $19,000 per coin produced, and it's dire enough that they're all looking at their assets—power contracts, land, cooling systems, entire operating teams—with fresh eyes. None of that was designed for Bitcoin alone. It could do something else. Something more profitable. Something called artificial intelligence.

The numbers add up, at least on paper. The halving slashed block subsidy from 6.25 BTC to 3.125 BTC, forcing miners to reconsider whether Bitcoin was even the best use of their infrastructure. GPU-based AI and high-performance computing workloads earn several times more revenue per megawatt than Bitcoin hashing. Bitcoin mining infrastructure costs roughly $700,000 to $1 million per megawatt. AI infrastructure runs $8 million to $15 million per megawatt. But AI offers structurally higher and more stable returns.

The transformation is happening fast. Core Scientific shifted dramatically: colocation generated $136.7 million, or 83% of its $164.2 million in Q2 revenue, while self-mining revenue crashed 66% to $21.5 million. Then came the mega-deals. Core Scientific signed a 15-year, 529 MW deal with AMD for AI infrastructure, potentially generating $14 billion in base contracted revenue, plus AMD rights to reserve up to 1.9 gigawatts of additional capacity through 2028. Others followed: Hut 8 inked a 15-year, $9.8 billion lease for an AI data center campus; TeraWulf's 20-year lease with Anthropic could generate roughly $19 billion.

Wall Street is eating it up. Miners with secured HPC contracts trade at 12.3 times next-twelve-month revenues, while pure-play Bitcoin miners sit at just 5.9 times. The market is paying more than double for the AI angle. But here's where the story gets uncomfortable.

The capital required is staggering. VanEck estimates a near-term capital shortfall of $50 billion to fund promised AI infrastructure, with long-term sector needs reaching $221 billion. Core Scientific is raising $3.3 billion through a junk-bond offering specifically to accelerate its pivot from Bitcoin mining to AI data centers.

The real risk isn't execution on building the infrastructure. The real risk is timing.

Contracted revenue and realized revenue are not the same thing. The gap between them is where skepticism concentrates, as deals move from announcement to actual cash flow. If AI demand growth slows—if it materializes more gradually than current investment cycles assume—while Bitcoin's price simultaneously recovers, miners will be locked into multi-year, multi-billion-dollar contracts that suddenly look like terrible trades.

McKinsey estimates global power demand for AI data centers will increase 250% from 2025 through 2030. That's impressive. But it's also an assumption. Bitcoin miners are betting their infrastructure advantage translates into durable, predictable cash flows before the capital markets lose patience. If that confidence proves wrong—if infrastructure glut and financing challenges arrive faster than expected—these miners will face a recovery in Bitcoin profitability they're no longer positioned to capture.

They may have just sold their future for a present that never comes.

Source & further reading:

Sources