AI’s surging demand for power-hungry information facilities is popping listed Bitcoin miners into strategic infrastructure performs, with Bernstein flagging practically $90 billion of introduced AI partnerships that might redraw the sector’s economics.
Bernstein’s newest analysis argues that the AI information middle construct‑out is colliding head‑on with a constrained U.S. energy grid, elevating Bitcoin (BTC) miners from “speculative hash‑factories” to important gatekeepers of huge‑scale compute. Within the report, seen by The Block, analysts say miners are “shock winners within the AI infrastructure growth” as a result of they already management large, energised websites in energy‑wealthy areas. Bernstein tallies over $90 billion in AI infrastructure collaborations introduced by hyperscale cloud suppliers, AI clouds, and chipmakers, involving roughly 3.7 GW of capability, and concludes that “Comply with the Gigawatts” is now the organizing precept of the AI construct‑out.
The report highlights IREN, Riot Platforms, CleanSpark, and Core Scientific as prime beneficiaries, assigning every an “outperform” score. For IREN, Bernstein units a $100 goal worth, implying roughly 98% upside from current ranges, whereas CleanSpark is given a $24 goal, about 78% above the place the inventory at present trades. The logic is blunt: “energy beats Bitcoin,” with Wall Road more and more valuing miners on their contracted megawatts and AI internet hosting offers quite than cash mined. In keeping with Bernstein, Bitcoin miners with energetic AI contracts commerce round $6 million per deliberate megawatt of capability, double the roughly $3 million per MW implied for pure‑play Bitcoin miners with out AI publicity.
Miners’ 27 GW edge in an influence‑starved AI race
Bernstein estimates that Bitcoin miners now management greater than 27 GW of deliberate energy capability globally, placing them in a structurally advantaged place versus new‑construct AI campuses that should navigate multi‑12 months interconnection queues. In elements of the U.S., the report notes, securing and energising a brand new 1 GW grid connection can take so long as 50 months, whereas many mining websites already sit on gigawatt‑scale substations and transmission strains. That actuality is why miners are more and more recast as suppliers of “heat powered shells” for AI and excessive‑efficiency computing: industrial‑scale land, energy, and information‑middle‑grade buildings, prepared for GPUs.
IREN is central to this narrative after pivoting from pure Bitcoin mining to AI compute and signing a sweeping partnership with NVIDIA to deploy as much as 5 GW of AI infrastructure constructed on the chipmaker’s DSX AI Manufacturing facility structure. Beneath that deal, IREN will roll out NVIDIA‑accelerated compute throughout a worldwide information middle portfolio, starting at a 2 GW campus in Sweetwater, Texas, whereas NVIDIA receives a 5‑12 months possibility to purchase as much as 30 million IREN shares at $70 every and commits to about $3.4 billion in GPU cloud spend over 5 years. On the identical time, Riot has agreed a ten‑12 months, $311 million lease with AMD that begins with 25 MW of information middle capability and may scale to 200 MW at its 700 MW‑linked Rockdale, Texas, web site, deepening the miner’s AI and excessive‑efficiency computing profile.
Upside, dangers, and the Bitcoin cycle commerce‑off
Bernstein’s math suggests the market continues to be discounting this energy optionality: Bitcoin miners total commerce at a roughly 90% valuation low cost to established AI information‑middle operators on some metrics, regardless of being “integral elements of the AI worth chain.” In a single instance, the brokerage attributes $3 billion of enterprise worth to Riot’s deliberate 1 GW Corsicana web site alone out of a $9 billion whole goal, though the power has but to generate significant income, underscoring how future AI internet hosting is driving the story. For Core Scientific, Bernstein estimates that 86% of goal enterprise worth stems from its AI enterprise, with solely 14% tied to legacy Bitcoin mining, reflecting how rapidly investor focus has shifted towards compute infrastructure.
The report is just not blind to dangers. Bernstein warns that new AI campuses stay hostage to environmental opinions, grid‑capability bottlenecks, and native allowing fights that may derail or delay megaprojects. It additionally cautions that if miners over‑tilt towards AI internet hosting and away from hash fee, they threat sacrificing upside in a future Bitcoin bull market simply as rewards from halving occasions and potential demand shocks might re‑fee pure mining economics. Nonetheless, with AI information‑middle demand rising quicker than utilities and regulators can add new gigawatts, Bernstein’s message is stark: miners sitting on low-cost, switchable energy are “energy gamers” within the AI arms race — and the market is just starting to cost it in.
