Bitcoin Miners Lose Grid Priority as AI Data Centers Lock Up Power Contracts

Bitcoin mining operations are being systematically displaced from U.S. power grids as tech giants secure multi-gigawatt electricity contracts for AI infrastructure, forcing miners to either pivot to AI hosting or migrate to lower-cost regions. The competition has intensified to the point where publicly traded miners now expect up to 70% of their revenue to come from AI by year-end, according to CoinDesk, marking an irreversible transformation of the mining sector.

The weighted average cash cost to produce one Bitcoin among publicly listed miners rose to approximately $79,995 in the fourth quarter of 2025, while Bitcoin traded between $68,000 and $70,000—creating inverted economics that have made continued mining operations financially untenable for many operators. This price compression, combined with AI infrastructure offering up to 25 times more revenue per kilowatt-hour depending on application, has triggered a wholesale industry transformation.

The Infrastructure Pivot

More than $70 billion in AI and high-performance computing contracts have been signed by mining companies in the past year. Core Scientific, Marathon Digital, Riot Platforms, and IREN have announced infrastructure pivots that repurpose existing Data Centers—originally built to house ASIC mining rigs—for Nvidia GPU clusters serving AI workloads.

CleanSpark stated in the first quarter that Bitcoin mining investment “doesn’t make a lot of sense” at current hashprices compared to the returns available in AI infrastructure, per insights4vc. The company is among several miners liquidating Bitcoin treasuries to fund the transition—public miners collectively reduced holdings by over 15,000 BTC from peak levels, with Core Scientific planning to liquidate substantially all remaining coin in the first quarter.

The migration is visible in network metrics. Bitcoin mining difficulty dropped 7.76% on 21 March—the second-largest decline of the year—as major miners redirected capacity to AI infrastructure, according to Techi. Network hashrate fell 4% in the first quarter alone, the first quarterly drop since 2020, driven by the combined pressure of miner pivots and geopolitical factors including the Iran-U.S. conflict raising energy costs in certain regions.

Grid Allocation Hierarchy Emerges

The displacement is not merely economic—it reflects a regulatory and utility preference for AI workloads over crypto mining. ERCOT’s interconnection queue now shows more than 250 gigawatts of large-load requests, roughly three times the grid operator’s current peak demand, creating massive bottlenecks that have forced a “Batch Zero” reassessment process affecting around 8.2 gigawatts of load, reported TheMinerMag in February.

In markets where interconnection approval can take 36-48 months, AI companies with deeper capital reserves are outbidding miners for both existing power contracts and queue positions. Meta committed to underwriting electricity costs for its Louisiana Hyperion data center before it begins operating—paying Entergy Louisiana to accelerate grid upgrades for an estimated 5 gigawatts of compute power, per WebProNews. This grid-prepayment model, previously unheard of in data center economics, signals how aggressively AI infrastructure providers are willing to deploy capital to secure power access.

Cumulative planned AI data center capacity in the U.S. now totals roughly 296 gigawatts across more than 70 projects exceeding 1 gigawatt of peak demand each, with Amazon leading at 22 gigawatts planned, according to a January survey by Data Center Knowledge. U.S. electricity demand from data centers has risen from roughly 23 gigawatts of new load in 2023 to about 42 gigawatts today, with another 32 gigawatts under construction—on target to surpass 90 gigawatts by 2030.

Anthropic Deal Signals Acceleration

On 6 April, Anthropic announced it had signed agreements with Google and Broadcom for multiple gigawatts of next-generation TPU capacity coming online starting in 2027, as its annual revenue run rate jumped to $30 billion from $9 billion at the end of 2025. The deal, reported by CoinDesk, represents one of the largest single compute commitments disclosed to date and underscores the scale at which AI labs are now procuring power and hardware—orders of magnitude beyond what even the largest mining operations could command.

The shift extends beyond economics into regulatory preference. Analysis from Tech Insider notes that “the bottleneck is no longer capital or demand—it is physical infrastructure,” with projects that secured land and financing years ago now stalled waiting for grid connections, transformers, and generation capacity that simply does not yet exist. In this environment, utilities and grid operators are prioritising loads they perceive as economically productive—AI training and inference—over crypto mining, which many regulators still view as speculative.

Iceland serves as a canary. The country capped crypto mining allocation at 120 megawatts total, with mining now accounting for roughly 8% of Iceland’s electricity and no new power allocation available for miners, according to ftfa-sao.org. This has forced miners to migrate to Central Asia and other lower-cost regions, a pattern now replicating across OECD markets as AI infrastructure claims priority.

What to Watch

Monitor ERCOT’s Batch Zero reassessment outcomes in the second quarter—if the grid operator cancels or delays large mining interconnection requests in favour of AI projects, expect similar policies to spread to California ISO and PJM Interconnection. Track public miner earnings through mid-2026 to verify whether the 70% AI revenue target materialises or if the pivot proves operationally harder than anticipated. Watch for federal or state-level regulatory intervention that formalises grid priority hierarchies—language distinguishing “productive compute” from “speculative workloads” in interconnection rules would cement the displacement. Finally, observe Bitcoin network security metrics: if hashrate declines persist beyond the first quarter, the network faces its first sustained security reduction since inception, with implications for transaction settlement assurance that could accelerate institutional pullback from the asset class.

The transformation is structural, not cyclical. Miners that fail to pivot will face stranded assets as power contracts expire and renewable PPAs get redirected to higher-paying AI tenants. The question is no longer whether Bitcoin mining can compete for grid capacity—it is whether the network can maintain adequate security as its most capitalised operators abandon mining for AI infrastructure hosting.