Publicly listed Bitcoin miners have collectively liquidated over 15,000 BTC from peak treasury levels and signed more than $70 billion in aggregate AI and high-performance computing (HPC) hosting contracts since late 2025. The shift marks the most significant structural reorientation the mining s...
"Bitcoin mining investment doesn't make a lot of sense at current hashprices compared to the returns available in AI infrastructure." — CleanSpark executive, Q1 2026 earnings call
Publicly listed Bitcoin miners have collectively liquidated over 15,000 BTC from peak treasury levels and signed more than $70 billion in aggregate AI and high-performance computing (HPC) hosting contracts since late 2025. The shift marks the most significant structural reorientation the mining sector has undergone since the April 2024 halving.
The economics are straightforward. With weighted average production costs near $80,000 per BTC and spot prices hovering around $72,000 as of April 9, 2026, miners face negative unit margins on every coin produced. Hashprice — the standard measure of daily miner revenue per petahash — collapsed to a five-year low of approximately $28/PH/s/day in late February 2026, according to CoinShares' Q1 2026 mining report. Meanwhile, AI workloads generate 3x to 25x more revenue per kilowatt-hour than Bitcoin mining, with margins between 80% and 90%. The capital is following the spread.
The result: Bitcoin's network hashrate posted its first Q1 decline in six years, falling roughly 4% year-to-date to approximately 1 zettahash per second (ZH/s). CoinShares estimates 15% to 20% of the global mining fleet is now operating below breakeven. The question is no longer whether miners will diversify — it is how much of the sector will remain committed to Bitcoin at all.
The post-halving economics that began compressing margins in April 2024 have reached a critical threshold. According to CoinShares' Q1 2026 Bitcoin Mining Report, the weighted average cash cost to produce one bitcoin among publicly listed miners rose to approximately $79,995 in Q4 2025. That figure excludes depreciation, stock-based compensation, and corporate overhead — the all-in cost is substantially higher.
Bitcoin's price trajectory has worked against miners. After peaking near $126,000 in October 2025, BTC slid to roughly $65,000 by February 2026 — a 48% drawdown. As of April 9, 2026, the price has partially recovered to approximately $72,000 following a U.S.-Iran ceasefire announcement on April 8 that lifted risk assets broadly.
Hashprice, which measures daily miner revenue per petahash of deployed hashrate, fell to a record low of approximately $27.89/PH/s/day in late February before recovering slightly to the $30-$35 range through March. For context, miners running mid-generation hardware (S19j Pro-class at ~29.5 J/TH) at average industrial electricity costs of $0.05/kWh were already operating well below breakeven in Q4 2025, according to CoinShares. Conditions have since worsened.
Mining difficulty itself reflected the exodus. On March 21, 2026, Bitcoin's difficulty fell 7.76% to 133.79 trillion at block height 941,472, marking the second-largest negative adjustment of 2026. The network's 30-day average hashrate fell to 1,004 EH/s in early Q2, down from 1,066 EH/s the prior quarter — a 5.8% decline, the sharpest quarterly contraction in years.
The ROI timeline for new mining equipment has stretched to approximately 1,000 days, according to CCN analysis — more than doubling the sub-400-day returns seen during the 2024 bull market.
The most visible symptom of the pivot is the mass liquidation of bitcoin treasuries by the same companies that once marketed themselves as leveraged BTC exposure vehicles.
Bitdeer (BTDR) reduced its bitcoin holdings to zero as of February 20, 2026, selling its entire remaining 1,132 BTC — including 943.1 BTC from company reserves and 189.8 newly mined coins. The company simultaneously priced a $325 million convertible notes offering and a $43.5 million equity raise to fund data center expansion. Despite holding no bitcoin, Bitdeer has maintained its position as the largest publicly traded miner by self-managed hashrate, having surpassed MARA Holdings.
Core Scientific (CORZ) disclosed in February 2026 that it expected to monetize "substantially all" of its bitcoin holdings during the year, with the majority of sales in Q1. The company sold approximately 1,900 BTC in January alone for $175 million, implying an average sale price of $92,100 per coin. Core Scientific's AI colocation revenue already accounts for 39% of total revenue.
In aggregate, publicly listed miners have reduced their BTC treasuries by over 15,000 BTC from peak levels, according to CoinShares. The sell-off appears to be nearing exhaustion for some operators, but the strategic direction is clear: bitcoin is being treated as working capital, not a reserve asset.
The scale of AI infrastructure commitments from former pure-play Bitcoin miners is substantial. According to CoinShares, miners have signed GPU co-location and cloud service deals with hyperscalers worth over $70 billion in aggregate over the course of 2025 and early 2026.
The revenue differential drives the pivot. AI workloads generate 3x to 25x more revenue per kilowatt-hour than Bitcoin mining, with margins between 80% and 90%, according to Bitdeer's disclosure. Long-term hosting contracts with investment-grade counterparties provide revenue visibility that block rewards and transaction fees cannot match.
CoinShares projects that some publicly listed miners could derive up to 70% of their total revenue from AI hosting by end-2026, up from roughly 30% at the start of the year. The sector is undergoing a valuation reframe: mining stocks are increasingly priced on AI infrastructure potential rather than BTC production capacity.
| Company | AI/HPC Capacity | Key Contract | Status | |---------|----------------|--------------|--------| | Core Scientific | 590 MW (CoreWeave) | $10B+ / 12 years | ~350 MW energized, ~200 MW billing | | TeraWulf | 886 MW pipeline | Core42/Fluidstack, Lake Mariner, Cayuga Lake | $13.85B valuation at 5.5% cap rate | | Riot Platforms | 200 MW (expandable) | AMD 10-year lease, $311M revenue potential | 25 MW initial at Rockdale | | IREN (Iris Energy) | 23,000 GPUs operational | Canadian AI cloud build-out | Expanding to 60,000 GPUs | | CleanSpark | 300 MW planned | Brazoria County AI/HPC facility | Development stage | | Bitdeer | Data center expansion | $325M convertible + $43.5M equity | Zero BTC treasury |
Core Scientific's CoreWeave contract alone is worth north of $10 billion over 12 years covering 590 megawatts of hosting capacity. As of early 2026, approximately 350 MW has been energized — more than half the contract fulfilled — with close to 200 MW already in the billing cycle.
Riot Platforms signed a 10-year lease with AMD for 25 MW at its Rockdale, Texas site in January 2026, expandable to 200 MW. The AMD lease is expected to generate up to $311 million in revenue, with contract extensions potentially reaching $1 billion. Riot also completed a $96 million land acquisition at the same site to support expansion.
TeraWulf's 886 MW AI/HPC pipeline — spanning contracts with Core42, Fluidstack, and facilities at Lake Mariner and Cayuga Lake — has been valued at $13.85 billion at a 5.5% capitalization rate, according to JonesResearch.
The hashrate migration raises network security questions, though the immediate risk profile remains contained.
Bitcoin's hashrate peaked at approximately 1,160 EH/s in October 2025. By February 2026, it had fallen to roughly 850 EH/s — a 27% decline from peak. It has since recovered to approximately 1,000 EH/s. The difficulty adjustment mechanism is functioning as designed: lower hashrate triggers lower difficulty, which reduces the cost for remaining miners and attracts marginal hashrate back.
A downward adjustment in difficulty reflects fewer machines mining, which theoretically decreases the economic cost of a 51% attack. However, no attacks have been observed, and global distribution — while concentrated — remains dispersed enough to prevent coordinated manipulation.
The more nuanced risk is centralization. According to Hashrate Index data, the top six mining pools consistently account for 80% to 90% of all blocks. Three countries — the United States (37.4%), Russia (16.9%), and China (12.0%) — now control approximately 65% of global hashrate.
CoinShares maintains its hashrate growth forecast: 1.8 ZH/s by end-2026 and 2 ZH/s by end-March 2027. The firm pushed the 2 ZH/s milestone back by one month from its prior estimate, reflecting the current softness.
The geographic distribution of mining power is shifting in tandem with the AI pivot. The United States gained approximately 2 percentage points of hashrate market share quarter-over-quarter in Q1 2026, largely because U.S.-based operators have the infrastructure, grid access, and capital markets proximity to execute AI pivots while maintaining some mining operations.
Emerging markets have entered the global top 10 in hashrate share. Paraguay attracted miners like HIVE Digital Technologies, which deployed 300 MW of capacity. Ethiopia and Oman also entered the rankings, with Bitdeer operating a 40 MW facility in Ethiopia. These jurisdictions offer low electricity costs — often below $0.03/kWh — that make mining viable even at current hashprices.
The January 2026 hashrate drop was partially driven by Winter Storm Fern, which forced Texas-based miners to power down rigs to relieve grid pressure. The event exposed the sector's geographic concentration risk: a single weather event in one U.S. state caused a measurable dip in global network security.
The Bitcoin mining industry is repricing itself as digital infrastructure. The shift from hoarding BTC as a reserve asset to liquidating it as working capital — while signing multi-billion-dollar AI hosting contracts — represents a fundamental change in how these companies generate and allocate economic value.
For Bitcoin's network, the immediate security implications are manageable. The difficulty adjustment mechanism continues to function, and hashrate, while down from peak, remains at historically high absolute levels. The longer-term question is whether the network can sustain security if the economic incentives for dedicated mining continue to deteriorate relative to alternative uses of the same infrastructure.
CoinShares' forecast of 1.8 ZH/s by year-end suggests the industry believes bitcoin mining is not dying — it is consolidating around the most efficient operators while the marginal capacity finds higher returns elsewhere. The miners that survive will be those that can straddle both worlds: maintaining hash commitments while monetizing excess capacity through AI. The ones that cannot will become data center companies that happen to have once mined bitcoin.