Core Scientific’s Shift to AI Data Centers
When Core Scientific signed a $3.5 billion agreement to host artificial intelligence (AI) data centers earlier this year, it wasn’t pursuing the next crypto token; it was in search of a steadier income. Once recognized for its extensive fleets of bitcoin mining rigs, the company has now become part of a burgeoning trend: transforming energy-intensive mining operations into high-performance AI facilities.
Transition from Mining to AI
Bitcoin miners such as Core, Hut 8 (HUT), and TeraWulf (WULF) are trading their ASIC machines—the dedicated bitcoin mining computers—for GPU clusters. This shift is spurred by the appeal of AI’s rapid growth and the challenging economics of crypto mining.
It is well-known that bitcoin mining demands a significant amount of energy, which constitutes the largest expense in creating a new digital asset.
The Changing Landscape of Bitcoin Mining
During the 2021 bull run, when the Bitcoin network’s hashrate and difficulty were low, miners enjoyed margins soaring as high as 90%. However, the harsh crypto winter and halving event, which halved the mining reward, changed the landscape. By 2025, miners grappled with surging hashrate and energy prices, with many now struggling to survive on razor-thin margins.
The need for power—the major input cost—ultimately became a hidden blessing for these miners, prompting them to adopt alternative strategies to diversify their revenue streams.
Investing in Energy Sources
With rising competition for mining, these miners sought to acquire more machines to remain viable, which led to an increased demand for more megawatts of electricity at lower costs. They heavily invested in securing low-cost energy sources, such as hydroelectric or stranded natural gas sites, and developed expertise in managing high-density cooling and electrical systems—skills sharpened during the crypto boom of the early 2020s.
This expertise caught the interest of AI and cloud computing firms. While bitcoin relies on specialized ASICs, AI benefits from versatile GPUs like Nvidia’s H100 series. These GPUs require similar high-power conditions tailored for parallel processing tasks in machine learning. Instead of constructing data centers from the ground up, leveraging existing mining infrastructure, which already has power readily available, proved to be a quicker route to meet the surging demand for AI-related infrastructure.
Repurposing Mining Infrastructure
The cooling systems, low-cost energy contracts, and power-dense infrastructure developed during the crypto boom now serve a different purpose: powering the AI models used by companies like OpenAI and Google.
More Read
Companies such as Crusoe Energy divested their mining assets to concentrate solely on AI, deploying GPU clusters in remote, energy-abundant locations that reflect the decentralized spirit of crypto yet now sustain centralized AI hyperscalers.
Bitcoin mining has effectively “terraformed” the landscape for AI compute by creating scalable, power-efficient infrastructures that AI urgently requires.
As Nicholas Gregory, Board Director at Fragrant Prosperity, stated, “It can be argued bitcoin paved the way for digital dollar payments as can be seen with USDT/Tether. It also looks like bitcoin terraformed data centres for AI/GPU compute.”
Adaptive Strategies in AI
This existing “terraforming” enables miners to retrofit facilities swiftly, often completing the process in under a year, in contrast to the multi-year timelines associated with traditional data center constructions. Firms like Crusoe Energy sold their mining assets to shift focus entirely to AI, reconfiguring GPU clusters in remote, energy-rich locales, adhering to the decentralized ethos of crypto while now servicing centralized AI hyperscalers.
In practice, this means that miners can transform a facility in less than a year—significantly faster than the multi-year timeframe for establishing a new data center.
Investment Costs and Revenue Potential
Bitcoin mining setups are relatively affordable, with costs ranging from $300,000 to $800,000 per megawatt (MW), excluding ASICs, allowing for quick scalability in response to market conditions. In contrast, AI infrastructure demands much higher capital expenditure due to the necessary advanced liquid cooling systems, redundant power setups, and the GPUs themselves, which can be priced at tens of thousands of dollars per unit and are currently facing global supply shortages. Despite the higher upfront costs, AI presents the potential for miners to generate up to 25 times more revenue per kilowatt-hour than bitcoin mining, making the transition economically attractive in light of increasing energy costs and dwindling crypto profitability.
The Future of Bitcoin Mining and AI
As AI continues to surge and crypto profitability declines, bitcoin mining may evolve into a niche industry—one that’s increasingly limited to energy-rich areas or highly efficient operators, especially as the next halving in 2028 could render many operations unprofitable without significant advances in efficiency or energy costs.
While forecasts suggest the global crypto mining market could grow to $3.3 billion by 2030, reflecting a modest 6.9% CAGR, this figure would be overshadowed by AI’s rapid escalation. According to KBV Research, the global AI in mining market is anticipated to reach $435.94 billion by 2032, progressing at a compound annual growth rate (CAGR) of 40.6%.
As investors begin to see potential in this shift, the overarching trend indicates that the future is either a hybrid model or a complete transition to AI, with stable contracts with hyperscalers offering longevity beyond the unpredictable cycles of crypto.
This evolution not only repurposes dormant assets but also highlights how the frontier of yesterday’s crypto industry is shaping the foundations of tomorrow’s AI enterprises.
