Instead of chasing fleeting cryptocurrency profits, Core Scientific made a strategic decision earlier this year, inking a $3.5 billion agreement to host artificial intelligence (AI) driven data centers. The company, previously recognized for its extensive Bitcoin mining operations, is now embracing a growing trend: repurposing energy-intensive mining facilities into advanced AI infrastructure hubs.
Companies involved in Bitcoin mining, including Core, Hut 8 (HUT), and TeraWulf (WULF), are increasingly replacing their Application-Specific Integrated Circuit (ASIC) miners—specialized hardware for Bitcoin computations—with Graphics Processing Unit (GPU) clusters. This shift is fueled by AI’s rapid expansion and the challenging financial landscape of cryptocurrency mining.
Power Play
The energy demands of Bitcoin mining are substantial and represent a significant portion of the cost associated with producing new digital assets.
During the Bitcoin bull market of 2021, miners enjoyed substantial profit margins, sometimes as high as 90%, due to a low network hashrate and low mining difficulty. However, the subsequent crypto downturn, along with the Bitcoin halving event that halved mining rewards, made survival difficult for miners. Looking ahead to 2025, increasing hashrate and rising energy expenses are creating exceedingly thin margins for miners.
Paradoxically, the high energy consumption—the main expense—has turned into an unexpected advantage for these mining companies, requiring a shift in strategy to diversify revenue.
Heightened competition in the mining sector led miners to acquire more machines to maintain their positions, requiring more electricity at lower prices. These companies invested heavily in securing affordable power sources, such as hydroelectric facilities and stranded natural gas sites. They also cultivated expertise in handling high-density cooling and advanced electrical systems, skills acquired during the cryptocurrency boom of the early 2020s.
These capabilities caught the attention of AI and cloud computing entities. While Bitcoin mining relies on specialized ASICs, AI applications thrive on adaptable GPUs, such as Nvidia’s H100 series. These require similar high-power infrastructures but are designed for parallel processing tasks involved in machine learning. Rather than constructing new data centers, leveraging existing mining infrastructure with available power proved a quicker path to accommodate the growing need for AI-related infrastructure.
In essence, these miners are not just changing direction—they are adapting existing systems.
The cooling technologies, budget-friendly energy contracts, and high-density power structures developed during the crypto boom now serve a new function: powering AI models for companies like OpenAI and Google.
Entities such as Crusoe Energy divested its mining operations to focus entirely on AI, implementing GPU clusters in remote, energy-rich locations mirroring crypto’s decentralized philosophy while serving centralized AI hyperscalers.
Terraforming AI
Bitcoin mining has effectively “terraformed” the landscape for AI computing, providing scalable and power-efficient infrastructure vital for AI development.
Nicholas Gregory, Board Director at Fragrant Prosperity, observed, “It could be said that Bitcoin laid the foundation for digital dollar payments, as seen with USDT/Tether. It also appears Bitcoin has terraformed data centers for AI/GPU computing.”
This “terraforming” enables mining companies to quickly adapt facilities, often in less than a year, compared to the multi-year projects needed for standard data center construction. Firms like Crusoe Energy sold off mining assets to concentrate on AI, deploying GPU clusters in isolated areas abundant with energy, resembling the decentralized principles of cryptocurrency but supporting centralized AI powerhouses.
Higher Returns
In practical terms, it means miners can convert a facility in under a year, significantly faster than building a new data center which takes several years.
However, upgrading to AI is a costly endeavor.
Bitcoin mining setups have relatively lower costs, ranging from $300,000 to $800,000 per megawatt (MW) excluding ASICs, allowing for quick scaling based on market conditions. Conversely, AI infrastructure demands much higher capital expenditures due to advanced liquid cooling, redundant power systems, and the GPUs themselves, which can cost tens of thousands per unit and are subject to global supply constraints. Despite these higher initial investments, AI offers miners up to 25 times more revenue per kilowatt-hour than Bitcoin mining, making the switch economically attractive given rising energy costs and shrinking cryptocurrency profitability.
A Niche Industry Worth Billions
As AI continues its rapid growth and cryptocurrency profits decline, Bitcoin mining might become a specialized sector—reserved for regions rich in energy or for highly efficient operators, especially given that the next halving in 2028 could render many operations unprofitable without gains in efficiency or lower energy costs.
While projections suggest the global cryptocurrency mining market will grow to $3.3 billion by 2030 at a 6.9% compound annual growth rate (CAGR), this growth would be dwarfed by AI’s immense expansion. According to KBV Research, the global AI in the mining market is expected to reach $435.94 billion by 2032, expanding at a CAGR of 40.6%.
With investors already recognizing the financial potential in this shift, the prevailing trend indicates either a hybrid model or a complete transformation to AI, where stable contracts with hyperscalers promise long-term stability, in contrast to the boom-and-bust cycles of cryptocurrency.
This evolution not only reuses underutilized resources but also highlights how the past frontiers of crypto are paving the way for the future AI landscape.