Bitcoin's mining sector is operating at a loss. The weighted average cash cost to produce one BTC reached $79,995 in Q4 2025 according to CoinShares, while spot price has traded between $65,000 and $75,000 for most of Q1 2026. Hash price — the standard measure of miner revenue per unit of computa...
"Selling bitcoin could become a recurring element of our treasury strategy." — MARA Holdings, Q1 2026 Earnings Communication
Bitcoin's mining sector is operating at a loss. The weighted average cash cost to produce one BTC reached $79,995 in Q4 2025 according to CoinShares, while spot price has traded between $65,000 and $75,000 for most of Q1 2026. Hash price — the standard measure of miner revenue per unit of computational power — collapsed to $28–30/PH/s/day by March 2026, a post-halving all-time low and well below the $36–38/PH/s/day breakeven threshold for a significant share of the fleet.
The result: the first quarterly hashrate decline since 2020, three consecutive negative difficulty adjustments (the first such streak since 2022), and a coordinated liquidation of BTC treasuries by publicly listed miners totaling more than 15,000 BTC. The sector is not contracting randomly. It is restructuring around AI and high-performance computing (HPC) contracts now valued at over $70 billion cumulatively, with projections that up to 70% of public miner revenue could derive from non-mining compute by year-end 2026.
Bitcoin's network hashrate posted its first Q1 decline in six years. After peaking near 1,200 EH/s in late 2025, the hashrate fell to approximately 1,004 EH/s by early Q2 2026 — a 5.8% quarter-over-quarter contraction, according to data from CoinWarz and Hashrate Index. In March alone, the hash rate dropped 22% in two weeks, falling from 1,200 EH/s on March 5 to 813 EH/s on March 25, according to ABC Money, before partially recovering.
Mining difficulty followed. Three consecutive negative adjustments occurred in Q1 2026, the first such streak since the post-FTX crash in late 2022. Difficulty dropped 7.76% in March to 133.79 trillion. As of early April, difficulty stands at 138.97 trillion, with the next adjustment on April 18 projected to bring another reduction — estimated at 6.7% — to approximately 129.67 trillion.
Hash price — the dollar value of revenue per petahash per second per day — hit $27.89/PH/s/day at its nadir, per CoinShares. The weighted average for Q1 2026 was $29/PH/s/day. For context, the same metric was above $100/PH/s/day before the April 2024 halving. CoinShares estimates that at $30/PH/s/day, any miner operating hardware below an Antminer S19 XP with electricity costs at or above $0.06/kWh is losing money. This covers roughly 15–20% of the global mining fleet.
A contributing factor beyond price: the Iran-U.S. conflict lifted global energy prices in March 2026, according to CoinDesk, compressing margins for miners in regions with floating electricity rates.
According to the 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. All-in sustaining costs — including depreciation, interest, and SG&A — are materially higher.
Bitcoin's spot price as of April 8, 2026, was approximately $71,547, per CoinGabbar. This means the average publicly traded miner loses an estimated $8,000–$19,000 on every bitcoin produced at current economics. Token Metrics analysts estimate average losses of $19,000 per coin.
The breakeven calculation is sensitive to electricity rates:
Post-halving, one bitcoin now requires approximately 854,400 kilowatt-hours of electricity to produce, according to BestBrokers.com data compiled in late 2025. Only operators with sub-$0.06/kWh power contracts and latest-generation ASICs (such as the Bitmain Antminer S21 or MicroBT WhatsMiner M60) maintain positive margins at current prices.
The return on investment for new mining hardware now exceeds 1,000 days, according to CCN — more than 2.7 years to break even on a machine purchase at current economics, compared to under 12 months during peak profitability periods.
Public miners have collectively reduced their BTC treasuries by more than 15,000 BTC from peak levels, according to CoinShares. The liquidations are not speculative trades. They are funding debt repayment, operational costs, and the transition to AI infrastructure.
Core Scientific sold approximately 1,900 BTC (~$175 million) in January 2026 and announced plans to liquidate "substantially all" remaining holdings throughout Q1. The company held 2,537 BTC worth $222 million at year-end 2025. Proceeds funded its AI pivot and debt restructuring.
MARA Holdings sold 15,133 BTC between March 4 and March 25, 2026, generating approximately $1.1 billion at an average price of $72,689. Proceeds funded the repurchase of outstanding 0.00% convertible senior notes due 2030 and 2031. MARA simultaneously cut 15% of staff, per Unchained Crypto. The company stated that BTC sales could become a "recurring element" of its treasury strategy.
Riot Platforms sold 3,778 BTC in Q1 2026, generating $289.5 million at an average price of $76,626. As of quarter-end, Riot held 15,680 BTC.
CleanSpark produced 568 BTC in February and sold 553 BTC — 97% of monthly output — indicating a shift from accumulation to immediate liquidation.
The pattern is consistent: miners that previously operated under a "HODL" thesis — accumulating BTC on balance sheets to benefit from future price appreciation — have reversed course. The operational environment no longer supports the cost of holding.
The mining sector's pivot to AI and HPC is no longer theoretical. Over $70 billion in cumulative AI/HPC contracts have been announced across publicly listed miners, according to industry tracking data compiled by insights4vc.
IREN signed a $9.7 billion, five-year AI cloud agreement with Microsoft, involving the deployment of 76,000 NVIDIA GB300 GPUs across 200MW of capacity at its Childress, Texas campus. IREN targets a fleet of 140,000 GPUs by end of 2026, with its Sweetwater 1 site expansion scheduled for energization in April 2026. The company has effectively transformed from a Bitcoin miner to an AI infrastructure provider.
Core Scientific signed a $10.2 billion, 12-year hosting deal with CoreWeave. The company rejected a $9 billion buyout offer to maintain independence as a colocation provider. Core Scientific's strategic direction is now defined by AI hosting, not bitcoin mining.
MARA has begun converting mining facilities to AI data centers, per its Q1 disclosures, though the company is earlier in the transition than peers.
The economics are straightforward: a megawatt of power capacity allocated to AI hosting can generate 5–10x the revenue of the same megawatt allocated to bitcoin mining, depending on contract terms and utilization rates. Miners hold the two assets most difficult to acquire in the current AI buildout — high-voltage electrical infrastructure and physical data center shells with cooling capacity.
Industry projections suggest up to 70% of public miner revenue could derive from AI hosting contracts by end of 2026, up from approximately 30% at present. This represents a fundamental sector reclassification: "Bitcoin miners" are becoming compute infrastructure companies that happen to also mine some bitcoin.
Bitcoin's security model is directly tied to hashrate. A lower hashrate reduces the cost of mounting a 51% attack, at least in theory. The 22% intra-month decline in March 2026 prompted discussion about the network's resilience.
However, two structural factors mitigate the risk. First, the difficulty adjustment mechanism — hardcoded into Bitcoin's protocol — automatically reduces mining difficulty when hashrate drops, maintaining approximately 10-minute block intervals and ensuring continued block production regardless of miner participation levels. The system functioned as designed throughout Q1.
Second, the geographic distribution of mining may actually improve as U.S.-listed miners — which account for over 40% of global hashrate — redirect capacity to AI. This could lead to a more decentralized network, with a larger share of hashrate operated by independent miners in lower-cost jurisdictions such as Paraguay, Ethiopia, and Central Asia.
The counterargument: a smaller, more fragmented hashrate is cheaper to attack by state-level adversaries. At current levels (~1,000 EH/s), mounting a sustained 51% attack would still require hardware and electricity expenditures in the tens of billions of dollars. The network remains practically secure, but the margin of safety has narrowed.
The deeper concern is structural. If bitcoin mining continues to be economically subordinate to AI hosting for the companies that control the largest power infrastructure, the long-term incentive to mine bitcoin depends entirely on price recovery. Without it, the network's security budget erodes further.
Bitcoin mining in Q1 2026 is an industry operating below cost. The April 2024 halving cut block rewards from 6.25 to 3.125 BTC. Bitcoin's price declined 45% from its October 2025 all-time high of $126,272. Global energy prices rose on geopolitical tensions. These three forces converged to produce the worst margin environment in the sector's history.
The industry's response has been rapid and rational: liquidate BTC holdings, cut staff, and redirect power infrastructure toward AI hosting, where unit economics are multiples of what bitcoin mining can offer. This is not a temporary adjustment. The $70 billion in AI contracts represent multi-year commitments that will permanently alter how these companies allocate their most scarce resource — power capacity.
For Bitcoin the network, the implications are material but manageable in the near term. The difficulty adjustment mechanism ensures block production continuity. However, the long-term security budget — the economic incentive for miners to secure the network — is directly linked to BTC price multiplied by block rewards. At current price levels, that budget is structurally insufficient to sustain historical hashrate levels, let alone grow them.
CoinShares and industry analysts project that hashrate recovery to 1.8 ZH/s by year-end requires a BTC price recovery toward $100,000. Absent that, the mining sector will continue its metamorphosis into something the Satoshi whitepaper never envisioned: an AI infrastructure industry with a bitcoin mining side business.