3 hours ago
When Core Scientific signed a $3.5 cardinal woody to big artificial quality (AI) information centers earlier this year, it wasn’t chasing the adjacent crypto token — it was chasing a steadier paycheck. Once known for its immense fleets of bitcoin mining rigs, the institution is present portion of a increasing trend: converting energy-intensive mining operations into high-performance AI facilities.
Bitcoin miners similar Core, Hut 8 (HUT) and TeraWulf (WULF) are swapping ASIC machines — the dedicated bitcoin mining machine — for GPU clusters, driven by the lure of AI’s explosive maturation and the harsh economics of crypto mining.
Power play
It's nary concealed that bitcoin mining requires an extended magnitude of energy, which is the biggest outgo of minting a caller integer asset.
Back successful the 2021 bull run, erstwhile the Bitcoin network's hashrate and trouble were low, miners were making retired similar bandits with margins arsenic overmuch arsenic 90%. Then came the brutal crypto wintertime and the halving event, which slashed the mining reward successful half. In 2025, with surging hashrate and vigor prices, miners are present struggling to past with razor-thin margins.
However, the request for power—the biggest input cost—became a blessing successful disguise for these miners, who needed a antithetic strategy to diversify their gross sources.
Due to rising contention for mining, the miners continued to procure much machines to enactment afloat, and with it came the request for much megawatts of energy astatine a cheaper price. Miners invested heavy successful securing these low-cost vigor sources, specified arsenic hydroelectric oregon stranded earthy state sites, and developed expertise successful managing high-density cooling and electrical systems—skills honed during the crypto roar of the aboriginal 2020s.
This is what captured the attraction of AI and unreality computing firms. While bitcoin relies connected specialized ASICs, AI thrives connected versatile GPUs similar Nvidia's H100 series, which necessitate akin high-power environments but for parallel processing tasks successful instrumentality learning. Instead of gathering retired information centers from scratch, taking implicit mining infrastructure, which already has powerfulness ready, became a faster mode to turn an expanding appetite for AI-related infrastructure.
Essentially, these miners aren’t conscionable pivoting—they’re retrofitting.
The cooling systems, low-cost vigor contracts, and power-dense infrastructure they built during the crypto roar present service a caller purpose: feeding the AI models of companies similar OpenAI and Google.
Firms similar Crusoe Energy sold disconnected mining assets to absorption solely connected AI, deploying GPU clusters successful remote, energy-rich locations that reflector the decentralized ethos of crypto but present substance centralized AI hyperscalers.
Terraforming AI
Bitcoin mining has efficaciously "terraformed" the terrain for AI compute by gathering retired scalable, power-efficient infrastructure that AI desperately needs.
As Nicholas Gregory, Board Director astatine Fragrant Prosperity, noted, "It tin beryllium argued bitcoin paved the mode for integer dollar payments arsenic tin beryllium seen with USDT/Tether. It besides looks similar bitcoin terraformed information centres for AI/GPU compute."
This pre-existing "terraforming" allows miners to retrofit facilities quickly, often successful nether a year, compared to the multi-year timelines for accepted information halfway builds. Firms similar Crusoe Energy sold disconnected mining assets to absorption solely connected AI, deploying GPU clusters successful remote, energy-rich locations that reflector the decentralized ethos of crypto but present substance centralized AI hyperscalers.
Higher returns
In practice, it means miners tin flip a installation successful little than a year—far faster than the multi-year timeline of a caller information center.
But AI isn’t a inexpensive upgrade.
Bitcoin mining setups are comparatively modest, with costs ranging from $300,000 to $800,000 per megawatt (MW) excluding ASICs, allowing for speedy scalability successful effect to marketplace cycles. Meanwhile, AI infrastructure demands importantly higher capex owed to the request for precocious liquid cooling, redundant powerfulness systems, and the GPUs themselves, which tin outgo tens of thousands per portion and look planetary proviso shortages. Despite the steeper upfront costs, AI offers miners up to 25 times much revenue per kilowatt-hour than bitcoin mining, making the pivot economically compelling amid rising vigor prices and declining crypto profitability.
A niche manufacture worthy billions
As AI continues to surge and crypto profits tighten, bitcoin mining could go a niche game—one reserved for energy-rich regions oregon highly businesslike players, particularly arsenic the adjacent successful 2028 could render galore operations unprofitable without breakthroughs successful ratio oregon vigor costs.
While projections amusement the planetary crypto mining market increasing to $3.3 billion by 2030, astatine a humble 6.9% CAGR, the billions would beryllium overshadowed by AI's exponential expansion. According to KBV Research, the planetary AI successful mining marketplace is projected to scope $435.94 cardinal by 2032, expanding astatine a compound yearly maturation complaint (CAGR) of 40.6%.
With investors already seeing dollar signs successful this shift, the broader inclination suggests the aboriginal is either a hybrid oregon a afloat conversion to AI, wherever unchangeable contracts with hyperscalers committedness longevity implicit crypto's boom-bust cycles.
This improvement not lone repurposes idle assets but besides underscores however yesterday's crypto frontiers are forging tomorrow's AI empires.
English (US) 