When Core Scientific signed a $3.5 billion deal to host synthetic intelligence (AI) knowledge facilities earlier this yr, it wasn’t chasing the subsequent crypto token — it was chasing a steadier paycheck. As soon as identified for its huge fleets of bitcoin mining rigs, the corporate is now a part of a rising development: changing energy-intensive mining operations into high-performance AI services.
Bitcoin miners like Core, Hut 8 (HUT) and TeraWulf (WULF) are swapping ASIC machines — the devoted bitcoin mining pc — for GPU clusters, pushed by the lure of AI’s explosive progress and the tough economics of crypto mining.
Energy play
It is no secret that bitcoin mining requires an in depth quantity of power, which is the largest price of minting a brand new digital asset.
Again within the 2021 bull run, when the Bitcoin community’s hashrate and issue had been low, miners had been making out like bandits with margins as a lot as 90%. Then got here the brutal crypto winter and the halving occasion, which slashed the mining reward in half. In 2025, with surging hashrate and power costs, miners at the moment are struggling to outlive with razor-thin margins.
Nonetheless, the necessity for energy—the largest enter price—grew to become a blessing in disguise for these miners, who wanted a distinct technique to diversify their income sources.
Attributable to rising competitors for mining, the miners continued to acquire extra machines to remain afloat, and with it got here the necessity for extra megawatts of electrical energy at a less expensive value. Miners invested closely in securing these low-cost power sources, comparable to hydroelectric or stranded pure gasoline websites, and developed experience in managing high-density cooling and electrical techniques—expertise honed throughout the crypto increase of the early 2020s.
That is what captured the eye of AI and cloud computing companies. Whereas bitcoin depends on specialised ASICs, AI thrives on versatile GPUs like Nvidia’s H100 sequence, which require comparable high-power environments however for parallel processing duties in machine studying. As an alternative of constructing out knowledge facilities from scratch, taking up mining infrastructure, which already has energy prepared, grew to become a quicker option to develop an rising urge for food for AI-related infrastructure.
Basically, these miners aren’t simply pivoting—they’re retrofitting.
The cooling techniques, low-cost power contracts, and power-dense infrastructure they constructed throughout the crypto increase now serve a brand new objective: feeding the AI fashions of firms like OpenAI and Google.
Corporations like Crusoe Vitality bought off mining belongings to focus solely on AI, deploying GPU clusters in distant, energy-rich places that mirror the decentralized ethos of crypto however now gasoline centralized AI hyperscalers.
Terraforming AI
Bitcoin mining has successfully “terraformed” the terrain for AI compute by constructing out scalable, power-efficient infrastructure that AI desperately wants.
As Nicholas Gregory, Board Director at Aromatic Prosperity, famous, “It may be argued bitcoin paved the way in which for digital greenback funds as may be seen with USDT/Tether. It additionally appears like bitcoin terraformed knowledge centres for AI/GPU compute.”
This pre-existing “terraforming” permits miners to retrofit services shortly, usually in beneath a yr, in comparison with the multi-year timelines for conventional knowledge middle builds. Corporations like Crusoe Vitality bought off mining belongings to focus solely on AI, deploying GPU clusters in distant, energy-rich places that mirror the decentralized ethos of crypto however now gasoline centralized AI hyperscalers.
Larger returns
In follow, it means miners can flip a facility in lower than a yr—far quicker than the multi-year timeline of a brand new knowledge middle.
However AI isn’t an inexpensive improve.
Bitcoin mining setups are comparatively modest, with prices starting from $300,000 to $800,000 per megawatt (MW) excluding ASICs, permitting for fast scalability in response to market cycles. In the meantime, AI infrastructure calls for considerably greater capex as a result of want for superior liquid cooling, redundant energy techniques, and the GPUs themselves, which may price tens of 1000’s per unit and face world provide shortages. Regardless of the steeper upfront prices, AI presents miners as much as 25 occasions extra income per kilowatt-hour than bitcoin mining, making the pivot economically compelling amid rising power costs and declining crypto profitability.
A distinct segment trade price billions
As AI continues to surge and crypto income tighten, bitcoin mining may develop into a distinct segment recreation—one reserved for energy-rich areas or extremely environment friendly gamers, particularly as the subsequent in 2028 may render many operations unprofitable with out breakthroughs in effectivity or power prices.
Whereas projections present the worldwide crypto mining market rising to $3.3 billion by 2030, at a modest 6.9% CAGR, the billions can be overshadowed by AI’s exponential enlargement. In response to KBV Analysis, the worldwide AI in mining market is projected to achieve $435.94 billion by 2032, increasing at a compound annual progress fee (CAGR) of 40.6%.
With traders already seeing greenback indicators on this shift, the broader development suggests the long run is both a hybrid or a full conversion to AI, the place steady contracts with hyperscalers promise longevity over crypto’s boom-bust cycles.
This evolution not solely repurposes idle belongings but in addition underscores how yesterday’s crypto frontiers are forging tomorrow’s AI empires.
