Headline: Биткоин-майнеры переходят к ИИ: новые горизонты для доходов и технологий Translation: Bitcoin miners shift to AI: new horizons for income and technology

Seven out of the ten largest miners by hash rate have reported generating revenue through activities related to AI or high-performance computing (HPC). The remaining three plan to follow suit, according to CryptoSlate.

Cryptocurrency miners are creating a secondary business model that competes with traditional ASIC mining, allowing them to secure stable income from clients utilizing GPUs.

In August, TeraWulf signed a 10-year contract with the cloud AI platform Fluidstack. Supported by Google, the miner will supply approximately 250 MW, amounting to $3.7 billion.

The company will offer its infrastructure at the Lake Mariner campus in Western New York for high-performance computing capabilities. The facility is designed specifically for liquid cooling of AI workloads and is intended to scale massively.

The agreement is expected to yield around $3.7 billion in revenue over the first ten years, with options for two extensions of five years each. If the contract is renewed, total revenue could reach $8.7 billion.

Google will provide guarantees for lease obligations up to $1.8 billion. In exchange, the corporation will receive warrants for approximately 41 million shares of TeraWulf — about 8% of its equity.

In June 2024, Core Scientific entered into a 12-year agreement with AI firm CoreWeave to provide infrastructure with a capacity of 200 MW for hosting Nvidia GPUs.

Core expects annual revenue from hosting services for CoreWeave to be around $290 million, totaling $3.5 billion over the contract term.

In October 2025, CleanSpark acquired 110 acres of land and 285 MW of electricity in Texas for a «next-generation» AI and HPC campus.

In August, MARA purchased a 64% stake in Exaion — a subsidiary of EDF — to enhance its global capabilities in AI and high-performance computing.

Riot is evaluating the conversion of about 600 MW in Corsicana for AI and HPC and has suspended all mining expansion plans. Bitfarms has hired consultants to explore the issue and is promoting its sites to clients in the AI sector.

Cipher Mining leased a data center in Colorado City, Texas to Fluidstack, a UK-based platform. Google agreed to guarantee lease obligations totaling $1.4 billion in exchange for a 5.4% stake in the mining company.

Phoenix Group from Abu Dhabi plans to increase its data center capacity to over 1 GW, focusing on AI. It is exploring the option of listing its shares in the U.S. to finance this expansion.

The total computational power of the Bitcoin network is around 1.08–1.1 EH/s. Each day, 144 blocks are mined, with miners receiving a block reward of 3.125 BTC plus transaction fees.

1 MW of power supplied to a modern ASIC with an energy efficiency of approximately 17 J/TH delivers a computational power of about 0.059 EH/s. Such a share in the network generates approximately $1–1.6 million annually, before electricity and operating costs, based on a Bitcoin price around $104,000.

The AI contract with TeraWulf anticipates an annual revenue of $1.85 million per MW of energy. More stable cash flows and high profitability have become significant advantages for companies betting on neural networks.

Utilities are adapting to the situation: American Electric Power has increased its five-year investment plan to $72 billion.

Shares of Bitcoin miners that are also involved in AI outperform those focused solely on cryptocurrency mining.

AI-oriented companies like Core Scientific, IREN, and Terawulf have achieved greater success due to investor confidence in diversified business models, more predictable revenue sources, and enhanced flexibility.

Analysts have noted that the hash rate growth of such miners may be slower than those focused exclusively on Bitcoin. However, AI-focused firms are expanding their data center infrastructure, opening up additional opportunities in the rapidly growing AI sector.

However, limitations exist. The availability of electricity and transformer supply, as well as the accessibility of gas turbines for new peak loads, dictate the pace of project launches. GPU supplies are also not unlimited.

In cryptocurrency mining, any change in network activity can lead to increased fees, narrowing the gap between revenues from this activity and AI hosting. An increase in the average fee of 0.5 BTC per block raises revenue by approximately $0.2–0.3 million per MW annually.

Contracts for computing power for AI and dollar returns per MW are becoming new key metrics for investors to monitor. A range of $1.5 million to $2 million per MW annually is becoming a benchmark for hosting centers in the U.S.

Capital plans of energy companies and updates on connection queues now hold equal importance for miners as the supply schedules of ASIC equipment.

As the electricity market situation in the U.S. tightens, miners with existing connections will be able to monetize their capabilities more quickly than players building facilities from scratch.

The aforementioned situation could lead to a slowdown in the overall hash rate growth of the Bitcoin network, particularly if a significant portion of new energy capacities is directed towards GPUs instead of ASICs.

Nonetheless, the high price of Bitcoin and spikes in fees continue to make the crypto industry promising. However, the hash rate has become a less reliable indicator of companies’ market value compared to previous cycles, according to CryptoSlate.

It is worth noting that in August 2024, the energy consumption of AI exceeded that of Bitcoin mining.