Publicly listed Bitcoin miners sold a record 32,000 BTC in Q1 2026, more than the total net sales in all of 2025 and exceeding the roughly 20,000 BTC liquidated during the Terra-Luna collapse in Q2 2022. The sell-off funded a capital migration toward artificial intelligence and high-performance c...
"We are no longer a Bitcoin company." — Ben Gagnon, CEO, Keel Infrastructure (formerly Bitfarms)
Publicly listed Bitcoin miners sold a record 32,000 BTC in Q1 2026, more than the total net sales in all of 2025 and exceeding the roughly 20,000 BTC liquidated during the Terra-Luna collapse in Q2 2022. The sell-off funded a capital migration toward artificial intelligence and high-performance computing infrastructure, where contracted revenue now exceeds $70 billion across the sector.
The economic logic is stark. Hashprice — the revenue a miner earns per petahash per second per day — fell to approximately $28/PH/s/day in early March 2026, a five-year low, according to CoinShares. The weighted average cash cost to produce one bitcoin among listed miners reached $79,995 in Q4 2025. With BTC trading near $78,000 in April, 15–20% of the global mining fleet is now cash-flow negative. Meanwhile, AI colocation contracts generate 3–25x revenue per megawatt with 80–90% EBITDA margins, compared to the sub-20% margins now typical in bitcoin mining.
The result is an industry bifurcation: a handful of low-cost, high-efficiency operators continue to mine bitcoin, while the majority of publicly listed miners are rebranding, selling bitcoin treasuries, and redirecting power capacity toward GPU clusters and data center leases. The implications for Bitcoin's network security model are material and largely unpriced.
The April 2024 halving cut block rewards from 6.25 BTC to 3.125 BTC. Eighteen months later, the downstream effects are compounding.
Hash price trajectory:
According to CoinShares' Q1 2026 Bitcoin Mining Report, the weighted average cash cost to produce one bitcoin among publicly listed miners reached approximately $79,995 in Q4 2025. At BTC prices near $78,000, mining operations are running at or below breakeven on a cash basis for the industry average operator.
Transaction fee revenue has offered no relief. Fees consistently accounted for less than 1% of total block rewards in Q1 2026, averaging approximately 0.018 BTC per block, according to CoinShares. This contrasts sharply with the brief fee spikes during Ordinals activity in 2023–2024 and removes what had been a potential profitability buffer.
Mining difficulty hit an all-time high of 155.97 trillion in late 2025. It has since adjusted downward — falling 7.76% on March 21, 2026, one of the largest single adjustments in Bitcoin's history — but remains elevated. The most recent adjustment on April 17 brought difficulty to 135.59 trillion, with the next adjustment on May 2 estimated to reduce it further to approximately 129.55 trillion.
Efficient miners running Antminer S21 XP hardware (13.5 J/TH) can remain operational at BTC prices as low as $55,000. Older fleets at 30+ J/TH begin shutting down below $75,000. CoinShares estimates 15–20% of the global mining fleet is currently unprofitable.
Public miners collectively sold over 32,000 BTC in Q1 2026. The scale is unprecedented: it exceeds total net sales for the full year of 2025.
Notable Q1 2026 sales:
| Company | BTC Sold | Approximate Proceeds | |---------|----------|---------------------| | MARA Holdings | 15,133 | ~$1.1B | | Riot Platforms | 3,778 | ~$289.5M | | Core Scientific | ~1,900 | ~$175M | | Other listed miners | ~11,189 | ~$800M+ |
The selling pattern marks a reversal of the "HODL" treasury strategy that dominated 2024, when miners accumulated bitcoin on their balance sheets as a strategic reserve. The shift is driven by two forces: compressed mining margins requiring BTC sales to cover operating costs, and the need for capital to fund AI infrastructure buildouts that require significant upfront GPU and facility investment.
MARA's $1.1 billion in March sales alone represents a strategic reorientation. CEO Fred Thiel has characterized the company's evolution as a shift toward "key infrastructure" for both bitcoin and sovereign computing, but the treasury data tells a more direct story: bitcoin is being converted to cash to fund diversification.
The aggregate value of AI and high-performance computing contracts announced by publicly listed bitcoin miners now exceeds $70 billion. The economic differential driving this migration is quantifiable.
Revenue per megawatt comparison:
Margin comparison:
According to S&P Global, the pivot is driven by three structural advantages that bitcoin miners possess: existing power purchase agreements (often at $0.03–0.05/kWh), large-scale facility permits, and grid interconnection capacity — all of which take 2–4 years to secure from scratch and are in acute short supply given AI infrastructure demand.
CoinShares projects that listed miners could derive up to 70% of revenues from AI by December 2026, up from approximately 30% at the start of the year.
IREN (formerly Iris Energy): Secured a $9.7 billion deal with Microsoft for 76,000 NVIDIA GB300 GPUs across 200 MW at its Childress, Texas campus. The company targets $3.4 billion in annualized AI cloud revenue by year-end 2026, supported by an expansion to 140,000 GPUs across a 4.5 GW development pipeline. Current annualized revenue run rate under contract stands at approximately $2.3 billion.
Core Scientific: Operates under a 12-year, $10.2 billion agreement with CoreWeave for approximately 590 MW of HPC capacity across multiple sites. AI colocation already accounts for 39% of revenue. The company announced plans to sell the bulk of its remaining BTC holdings in 2026 to fund AI expansion.
TeraWulf: Has signed HPC contracts totaling $12.8 billion, with 27% of revenues already from AI. The company's power infrastructure at Lake Mariner gives it sub-$0.04/kWh costs that are competitive for both mining and AI workloads.
Keel Infrastructure (formerly Bitfarms): Completed its rebrand on April 1, 2026, redomiciled from Canada to the United States, and began trading on the TSX under ticker KEEL on April 6. The company sold its 70 MW Paraguay mining site for approximately $13 million in April. CEO Ben Gagnon stated the company has "100% of our focus on North American HPC infrastructure development" and expects to hold "no Bitcoin" in time.
HIVE Digital: Raised $115 million via a zero-interest convertible note on April 22, 2026, earmarked for GPU purchases and Tier III data center development. S&P estimates HIVE's HPC revenue could reach 15% of total revenue in 2026, up from 7% in 2024.
Hut 8: Signed a $7 billion, 15-year lease for AI infrastructure at its River Bend campus, backed by Google.
Marathon (MARA): Reported $238 million in Q1 revenue, up 64% year-over-year, with energized hash rate at 57.4 EH/s. Adjusted EPS of $1.84 represented a 1,068% year-over-year increase. However, MARA remains among the most aggressive BTC sellers, liquidating over 15,000 BTC in March alone.
The network-level consequences of the mining exodus are measurable. The 30-day average hashrate fell to 1,004 EH/s in Q1 2026, down from 1,066 EH/s the prior quarter — a 5.8% quarter-over-quarter decline and the first quarterly drop in six years, according to CoinDesk. As of mid-April, hashrate stood at approximately 940 EH/s, down from peaks above 1.15 ZH/s in early 2026.
Approximately 252 EH/s of computing power is now offline, with much of that capacity believed to be permanently retired.
The security implications are twofold:
1. Reduced attack cost. Difficulty adjustments follow hashrate downward, reducing the theoretical cost of a 51% attack. While Bitcoin's network remains by far the most expensive proof-of-work chain to attack, the margin of safety is narrowing as hashrate declines. The March 21 difficulty drop of 7.76% was one of the largest single adjustments on record.
2. Geographic concentration risk. The United States, China, and Russia together control approximately 68% of global hashrate. As US-listed miners redirect power capacity from mining to AI, the relative share of hashrate in less transparent jurisdictions increases. This is a byproduct of the pivot, not an intentional outcome, but the effect is the same: network security becomes more dependent on operators in countries with less regulatory oversight.
The counterargument — that difficulty adjustments make mining more profitable for remaining operators, attracting new hashrate — holds in theory. In practice, the current hashprice environment requires either sub-$0.04/kWh power costs or next-generation hardware (sub-15 J/TH) to generate positive cash flow. The pool of operators meeting both criteria is narrowing.
The bitcoin mining industry is undergoing an economic restructuring driven not by regulation or technological failure, but by comparative advantage. AI infrastructure generates higher returns per megawatt, per dollar of capital expenditure, and per unit of management attention than bitcoin mining at current hashprice levels.
The companies best positioned are those with the cheapest power, the newest hardware, and — increasingly — the largest AI contract backlogs. The companies at greatest risk are pure-play miners with older fleets and no credible path to AI revenue.
For the Bitcoin network, the question is whether the difficulty adjustment mechanism and potential future BTC price appreciation can attract sufficient hashrate to maintain security margins as the industry's most capitalized operators redirect their infrastructure elsewhere. The data from Q1 2026 suggests this is not a hypothetical risk but an ongoing transition.