When Core Scientific signed a $3.5 billion deal earlier this year to host AI data centers, the company wasn’t chasing the next crypto token, it was chasing a more stable salary. The company was once known for its huge fleet of Bitcoin mining rigs, but is now part of a growing trend of converting energy-intensive mining operations into high-performance AI facilities.
Bitcoin miners such as Core, Hut 8 (HUT) and TeraWulf (WULF) are swapping ASIC machines, a dedicated computer, for GPU clusters, with the explosive growth of AI and the temptation of cryptocurrency mining.
Power play
It is well known that bitcoin mining requires a large amount of energy, which is the biggest cost of mining new digital assets.
Back to the bull market in 2021, when the Bitcoin Network’s hashrate and difficulty was low, miners were making up like bandits with as many as 90% margins. Then came the harsh cryptocurrency winter, and a half-life event occurred, where mining rewards were halved. In 2025, miners are currently struggling to survive at a very thin margin due to the rising hash rates and energy prices.
However, the biggest input cost, the need for electricity, has been a hidden fortune for miners who needed a different strategy to diversify their revenue streams.
With the growing competition in mining, miners continued to source more machines to continue their business, but with this, they needed more megawatts of electricity at a cheaper price. Miners have invested heavily to ensure low-cost energy sources such as hydroelectric power plants and stranded natural gas power plants, and have developed expertise in managing high-density cooling and electrical systems. This skill was honeed during the cryptocurrency boom of the early 2020s.
This is what caught the attention of AI and cloud computing companies. Bitcoin relies on special ASICs, but AI succeeds with multi-purpose GPUs like Nvidia’s H100 series. These GPUs require similar high power environments, but require parallel processing tasks in machine learning. Rather than building a data center from scratch, taking over a mining infrastructure that is already ready for electricity, has become a way to increase the growing demand for AI-related infrastructure more quickly.
Essentially, these miners are not just pivoting, they’re doing some modifications.
The cooling systems they built during the cryptocurrency boom, low-cost energy contracts and high-power density infrastructure now serve a new purpose: supplying AI models for companies like OpenAI and Google.
Companies like Crusoe Energy have sold mining assets to focus solely on AI and deployed GPU clusters in remote areas with energy richness. This reflects the decentralized spirit of cryptocurrency, but is now promoting centralized AI hyperscalers.
Terraforming AI
Bitcoin Mining effectively “terraforms” the realm of AI computing by building the scalable, power-efficient infrastructure that AI desperately needs.
“You could also argue that Bitcoin paved the way for digital dollar payments, as you can see in USDT/Tether. It also looks like a Bitcoin terraforming data center for AI/GPU computing,” said Nicholas Gregory, director of Fragrant Prosperity.
This existing “terraforming” allows miners to rapidly renovate facilities, often within a year, compared to schedules that require a traditional data center construction. Companies like Crusoe Energy have sold mining assets to focus solely on AI and deployed GPU clusters in remote areas with energy richness. This reflects the decentralized spirit of cryptocurrency, but is now promoting centralized AI hyperscalers.
Higher profit
In reality, this means that miners can switch facilities in under a year. This is much faster than the yearly schedule for a new data center.
But AI is not a cheap upgrade.
Bitcoin mining setups are relatively modest, with costs ranging from $300,000 to $800,000 per megawatt (MW) that can be scaled quickly according to the market cycle. Meanwhile, AI infrastructure requires a large amount of capital investment, as it requires sophisticated liquid cooling, redundant power systems, and the GPU itself. These costs can cost tens of thousands of dollars per unit, facing a global supply shortage. Despite rising upfront costs, AI has brought miners up to 25 times the revenue per kilowatt-hour of Bitcoin mining, making the pivot economically attractive amid rising energy prices and declining profitability of cryptocurrency.
Niche industry worth billions of dollars
As AI continues to surge and cryptocurrency profits shrink, Bitcoin mining could become a niche game. In 2028, in particular, without groundbreaking advances in efficiency and energy costs, many businesses could become unprofitable, making it a game exclusively for energy-rich regions and efficient players.
The global cryptocurrency mining market will grow to $3.3 billion by 2030 at a gradual CAGR of 6.9%, but that billions will be overshadowed by the rapid expansion of AI. According to KBV Research, the global AI mining market is projected to reach $435.94 billion by 2032, with an average annual growth rate (CAGR) of 40.6%.
Investors are already aware of the dollar signs in this change, and the broader trend suggests that the future is either a hybrid or a complete shift to AI, with stable contracts with hyperscalers promising longevity during the cryptocurrency recession cycle.
This evolution not only reuses idle assets, but also highlights how yesterday’s cryptocurrency frontier is building up tomorrow’s AI empire.
