Bitcoin mining is undergoing its most severe economic stress test since China's 2021 mining ban. With the average all-in production cost estimated at $87,000 per BTC and the spot price hovering near $70,000, miners are operating at a roughly 20% loss on every coin produced[^1]. The hashprice — th...
"Bitcoin mining is no longer a Bitcoin business. It is an energy arbitrage business that happens to produce Bitcoin when the math works — and hosts GPU clusters when it doesn't." — Sector analyst, February 2026
Bitcoin mining is undergoing its most severe economic stress test since China's 2021 mining ban. With the average all-in production cost estimated at $87,000 per BTC and the spot price hovering near $70,000, miners are operating at a roughly 20% loss on every coin produced[^1]. The hashprice — the daily revenue earned per petahash of deployed computing power — hit an all-time low of $33.31/PH/s in early February 2026, reducing even the most efficient public operators to break-even economics[^2]. On February 9, the network registered an 11% downward difficulty adjustment, the steepest single drop since China expelled its miners five years ago[^3].
But this is not merely a cyclical squeeze. It is a structural metamorphosis. Publicly traded mining companies have collectively announced over $65 billion in AI data center contracts, and industry projections suggest mining revenue will fall from 85% to under 20% of total sector revenue by late 2026[^4]. The companies that once secured the world's most valuable blockchain are quietly becoming landlords for NVIDIA's GPU clusters, raising a question the market has not yet fully priced: what happens to Bitcoin's security model when its miners no longer need Bitcoin?
This report examines the economics of the current mining crisis, the structural forces driving the AI pivot, and the long-term implications for Bitcoin's security budget and decentralization thesis.
The fundamental arithmetic of Bitcoin mining has inverted. According to data from Checkonchain and corroborated by JPMorgan research, the average all-in cost to produce one Bitcoin reached approximately $87,000 in late January 2026, while the spot market price had fallen to roughly $70,000 — a deficit of approximately $17,000 per coin, or a 20% loss on production[^1].
JPMorgan's subsequent estimate placed production costs at $77,000 following the difficulty adjustment and hashrate decline, suggesting some relief — but still above the spot price at the time of calculation[^5]. The divergence between these estimates ($77K vs. $87K) reflects the wide variance in mining economics: electricity costs range from $0.03/kWh for the most efficient operators in hydro-rich regions to $0.08/kWh or more for grid-dependent operations in the U.S. South.
The production cost breakdown (industry average):
| Component | Cost Per BTC | Share | |-----------|-------------|-------| | Electricity | $48,000–$58,000 | 55–67% | | Hardware depreciation | $14,000–$18,000 | 16–21% | | Facility & cooling | $6,000–$8,000 | 7–9% | | Personnel & G&A | $3,000–$5,000 | 3–6% | | Total all-in | $71,000–$89,000 | 100% |
The situation is particularly acute because the April 2024 halving cut block rewards from 6.25 BTC to 3.125 BTC per block, effectively doubling the per-coin infrastructure burden. Miners now earn approximately $28 million per day in total block rewards, down from a peak of over $75 million daily at Bitcoin's October 2025 all-time high of $126,000[^6].
Three converging forces created what analysts describe as a "perfect storm" for mining economics in February 2026:
Bitcoin's 44% drawdown from its October 2025 all-time high eliminated the post-halving profit cushion that miners had enjoyed during the Q3-Q4 2025 rally. The decline was driven by broader macro factors — rising U.S. recession expectations and a rotation out of risk assets — rather than mining-specific catalysts, but its impact on mining economics was devastating.
Severe winter storms swept across the U.S. South and Midwest in early February, triggering mandatory power curtailments from grid operators. Foundry USA Pool, the largest mining pool by hashrate, reported an approximately 60% drop in pool hashrate during peak disruption[^7]. Texas-based operations — which account for a disproportionate share of U.S. mining capacity — were forced offline as ERCOT prioritized residential heating loads over industrial consumers.
Some public mining firms reported daily Bitcoin output declining by more than 60% during the worst days of the storms.
Prior to the correction, Bitcoin's mining difficulty had climbed to 141.6 trillion — near all-time highs — as operators raced to deploy next-generation ASICs during the 2025 bull market. Network hashrate had surged to approximately 894.5 EH/s[^8]. This meant that even as revenue per coin collapsed, the computational cost to win each block remained elevated.
The result: on February 2, hashprice cratered to a record low of $33.31/PH/s per day[^2]. For context, even the most efficient public miners — CleanSpark and IREN — operate at total cash-based costs of approximately $30/PH/s and $26/PH/s respectively[^9]. At $33/PH/s hashprice, even tier-one operators are barely covering cash costs, with zero margin for depreciation, debt service, or capital expenditure.
The network responded on February 9 with an 11% downward difficulty adjustment — from 141.6 trillion to 125.86 trillion — the largest single drop since China's 2021 mining ban[^3]. This adjustment offered modest relief, but JPMorgan estimates that a sustained price above $77,000 is needed for the average miner to reach break-even[^5].
The mining industry's existential response to this structural squeeze is not to mine harder. It is to stop mining — at least partially — and repurpose infrastructure for artificial intelligence workloads.
By February 2026, publicly traded Bitcoin miners had announced over $65 billion in cumulative AI and high-performance computing (HPC) contracts with major technology companies[^4]. The transformation is visible across the sector's leading names:
| Company | Key AI Deal | Contract Value | AI Revenue Trajectory | |---------|------------|---------------|----------------------| | Core Scientific (CORZ) | CoreWeave colocation | $1.2B expansion | 26% gross margin on colocation vs. mining | | Hut 8 (HUT) | Fluidstack data center conversion | $7B / 15-year lease | Converting Louisiana campus entirely | | Applied Digital (APLD) | HPC cloud services | Multiple contracts | Exiting mining for HPC | | IREN (IREN) | GPU cluster hosting | Undisclosed | Converting power capacity to AI |
The financial logic is unambiguous. AI data center contracts generate approximately 3x the revenue per megawatt compared to Bitcoin mining, with operating margins of 80–90% on new HPC hosting contracts versus the current near-zero (or negative) margins on mining[^10]. Mining hardware payback periods now exceed 1,200 days — over three years — while AI hosting contracts offer predictable multi-year revenue streams with 99.999% uptime requirements and built-in escalation clauses[^4].
Industry projections suggest that for publicly traded miners as a group, Bitcoin mining revenue will decline from roughly 85% of total revenue to under 20% by late 2026[^4]. This represents the most dramatic business model transformation in the 15-year history of industrial Bitcoin mining.
Core Scientific exemplifies the trajectory: in Q3 2025, digital asset self-mining accounted for 70% of revenue, hosted mining 11%, and colocation 18%[^11]. By 2026, the company projects annualized colocation revenue of $360 million, with colocation margins (26% and rising toward 80% on new CoreWeave contracts) structurally superior to the mining segment.
The shift is backed by capital flows, not just press releases. Sector-wide data center capital expenditure increased by 400% between March 2025 and February 2026[^4]. Publicly traded miners no longer describe themselves as "Bitcoin companies" in investor presentations. The preferred nomenclature is now "digital infrastructure provider" — a term that captures power procurement, real estate development, and data center operations as primary competencies, with Bitcoin production relegated to a secondary revenue stream deployed when hashprice economics permit.
The AI pivot raises an uncomfortable question for Bitcoin maximalists: if miners' primary business is hosting NVIDIA H100 clusters, how much computing power will they dedicate to securing Bitcoin's $1.3 trillion network?
Bitcoin's security budget — the total revenue paid to miners for securing the network — currently runs at approximately $10.2 billion annualized (block rewards plus transaction fees at current prices)[^12]. This sounds substantial, but represents only 0.8% of Bitcoin's market capitalization, down from 1.6% before the halving and well below the 3–4% levels that characterized pre-2020 security spend ratios.
Transaction fees contribute a negligible share. Despite the network processing roughly $115 million in annualized fees, this represents less than 1.2% of total miner compensation[^13]. The remaining 98.8% comes from inflationary block rewards — newly minted Bitcoin diluting all existing holders.
Every megawatt allocated to AI hosting is a megawatt not mining Bitcoin. The economic incentive is stark: a megawatt of power allocated to AI HPC hosting can generate $300,000–$500,000 in annual revenue with 80%+ margins, while the same megawatt mining Bitcoin at current hashprice and difficulty produces approximately $100,000–$150,000 in revenue with near-zero margins[^10].
This creates a structural "competing bid" for the energy resources that have historically secured Bitcoin. As AI demand for power continues to scale — driven by hyperscaler capital expenditure that Alphabet, Microsoft, Amazon, and Meta have collectively budgeted at over $300 billion for 2026 — the opportunity cost of mining Bitcoin with that power will only increase.
The concentration of mining infrastructure among a small number of publicly traded companies — now primarily in the business of AI hosting — represents a potential centralization vector. If five to seven public companies control a majority of industrial-grade power capacity and choose to allocate it based on AI contract economics rather than Bitcoin hashrate, the network's security could become a byproduct of secondary economic calculations rather than a primary mission.
Bitcoin mining now consumes an estimated 170–180 TWh annually, comparable to the total electricity consumption of Poland or Thailand[^8]. This represents approximately 0.7–0.8% of global electricity production. The energy required to mine a single Bitcoin has climbed to approximately 854,400 kWh — enough to power 80 average U.S. households for a year[^14].
The geographic concentration of mining in the United States (37.8% of global hashrate) makes the sector particularly vulnerable to U.S. energy policy, grid reliability, and weather events — as the February 2026 winter storms demonstrated[^15].
Approximately 55–65% of Bitcoin mining reportedly utilizes renewable or sustainable energy sources, though industry self-reporting in this area remains difficult to independently verify.
The current shakeout is creating a clear two-tier structure in the mining industry:
Tier 1: Infrastructure Diversifiers — Companies with secured AI/HPC contracts, low electricity costs (<$0.04/kWh), and modern ASIC fleets. These firms treat Bitcoin mining as one of several revenue streams, deploying hashrate opportunistically when profitable. Examples: Core Scientific, IREN, Hut 8.
Tier 2: Pure-Play Miners — Companies dependent primarily on Bitcoin block rewards, with higher electricity costs and older hardware. These firms face existential pressure and are candidates for acquisition, shutdown, or forced pivoting. The February difficulty adjustment offers temporary relief, but does not solve the structural margin problem.
Key survival metrics:
| Metric | Tier 1 (Diversified) | Tier 2 (Pure-Play) | |--------|---------------------|-------------------| | Electricity cost | $0.03–0.04/kWh | $0.06–0.08/kWh | | Revenue mix (mining) | 20–50% | 80–100% | | AI contract pipeline | $1B+ | None | | Break-even BTC price | $55,000–$65,000 | $80,000–$95,000 | | Cash runway | 12–24 months | 3–6 months |
Bitdeer's Q4 2025 results illustrate the tension: revenue surged 225.8% year-over-year to $224.8 million, but gross profit collapsed to $10.6 million (4.7% margin) as lower Bitcoin prices and higher electricity costs consumed nearly all gains from expanded hashrate[^16].
Bitcoin mining is operating below production cost. The average all-in cost of $87,000/BTC against a ~$70,000 spot price represents a structural, not cyclical, deficit post-halving. JPMorgan estimates break-even at $77,000 after the difficulty adjustment.
The February 2026 difficulty drop was the largest since China's 2021 ban. An 11% decline from 141.6T to 125.86T reflects genuine miner capitulation — not just temporary curtailment.
The AI pivot is irreversible. With $65 billion in announced HPC contracts and 3x revenue-per-megawatt economics favoring AI hosting, mining companies are structurally reorienting away from Bitcoin as a primary business line.
Bitcoin's security budget faces a competing bid. Every megawatt diverted to AI hosting is a megawatt not securing the Bitcoin network. As AI hyperscaler demand scales, the opportunity cost of mining Bitcoin will continue to rise.
A two-tier mining industry is emerging. Diversified infrastructure operators with AI contracts will survive and thrive. Pure-play miners without diversification face existential margin pressure and potential consolidation.
Transaction fees remain inadequate. At roughly 1.2% of total miner compensation, user-generated fees are nowhere close to sustaining the security budget as block rewards continue their halving schedule toward zero.
The Bitcoin mining industry of February 2026 bears little resemblance to the sector that existed even 18 months ago. What was once a singular pursuit — deploying maximum hashrate to earn Bitcoin block rewards — has fractured into a multi-product infrastructure business where Bitcoin production is increasingly the least profitable use of scarce energy and capital resources.
This transformation is economically rational for individual mining companies. The pivot to AI hosting offers higher margins, more predictable revenue, and reduced exposure to Bitcoin's price volatility. But what is rational for each miner individually may be corrosive to the network collectively. Bitcoin's security model assumes that miners are economically motivated to dedicate resources to securing the chain. When those same resources command a 3x premium doing something else entirely, the assumption weakens.
The next Bitcoin halving is scheduled for 2028, which will cut block rewards to 1.5625 BTC. At current fee levels, this will further compress the security budget at precisely the moment when AI infrastructure demand — and the competing bid for miner power capacity — will likely be even more intense.
The market has not yet fully reckoned with this structural shift. Bitcoin's $1.3 trillion valuation implicitly assumes that sufficient mining infrastructure will always be available to secure the network. February 2026's capitulation event suggests that assumption deserves significantly more scrutiny.
[^2]: TheMinerMag — "Bitcoin Hashprice Hits Record Low as BTC Falls Below $70,000" (Feb 5, 2026)
[^3]: CoinDesk — "Bitcoin mining difficulty drops by most since 2021 as miners capitulate" (Feb 9, 2026)
[^7]: IndexBox — "Bitcoin Mining Difficulty Drops 11% in 2026: Causes & Market Impact" (Feb 2026)
[^8]: Compare Forex Brokers — "The Real Cost of Bitcoin Mining in 2026" (Feb 2026)
[^9]: Archyde — "Bitcoin Hashprice Plummets to Record Low — Miner Pressure Mounts" (Feb 2026)
[^10]: Insights4VC — "Bitcoin Mining's AI Pivot: 2026 Thesis Update" (Feb 2026)
[^11]: Seeking Alpha — "Core Scientific: Undervalued Powerhouse In The AI Data Center Era" (2026)
[^12]: Cointelegraph — "Bitcoin Mining 2026: AI Pivot, Profitability Pressure & Consolidation" (2026)
[^13]: Based on annualized network fee data from blockchain explorers and the foundational economic value analysis (Maze2 SA, October 2025).
[^14]: Buy Bitcoin Worldwide — "61 Bitcoin Energy Consumption Statistics (2026)"
[^16]: Yahoo Finance — "Bitdeer Technologies Group Q4 Earnings Call Highlights" (Feb 12, 2026)