Bitcoin mining is experiencing its most severe profitability crisis since the 2021 China ban. With spot prices near $70,000 against an average production cost of $87,000, the industry is bleeding cash. Hashprice — the key metric measuring daily mining revenue per unit of computational power — has...
"Bitcoin trades 20% below its production cost... miners are continuing to sell bitcoin holdings to fund day-to-day operations, cover energy expenses, and service debt." — CoinDesk Markets Research, February 2026
Bitcoin mining is experiencing its most severe profitability crisis since the 2021 China ban. With spot prices near $70,000 against an average production cost of $87,000, the industry is bleeding cash. Hashprice — the key metric measuring daily mining revenue per unit of computational power — has collapsed to record lows below $32 per petahash per second, halving from $70/PH/s at Bitcoin's October 2024 highs. The network's mining difficulty just posted an 11% downward adjustment, the largest single drop since China expelled its miners five years ago.
This is not merely a cyclical correction. The April 2024 halving permanently cut block rewards from 6.25 to 3.125 BTC, compressing revenues at the precise moment that hashrate surged to all-time highs above 1.1 ZH/s. The result is a structural squeeze that is now forcing the most consequential restructuring the mining sector has ever faced: a mass migration from proof-of-work to artificial intelligence infrastructure. What began as opportunistic diversification has become an existential pivot, with companies like Bitfarms dismantling mining operations entirely and Core Scientific reinventing itself as an AI bare-metal provider.
For an industry that consumed approximately 173 TWh of electricity in 2025 — more than Argentina — and required an estimated $18.1 billion in annual Bitcoin issuance subsidies just to sustain itself, the question is no longer whether mining is profitable. It is whether the Bitcoin mining industry, as a standalone business model, can survive in its current form at all.
The numbers tell a stark story. Bitcoin's hashprice — the most widely tracked mining profitability metric — fell below $32/PH/s on February 5, 2026, setting a new all-time low. For context, hashprice peaked near $70/PH/s when Bitcoin traded at its all-time highs, meaning revenue per unit of mining power has more than halved in under 16 months.
At the current spot price near $70,000, Bitcoin trades approximately 20% below the estimated average global production cost of $87,000. In many U.S. regions, where industrial electricity rates have climbed alongside AI-driven data center demand, the raw energy cost to mine a single Bitcoin now exceeds $100,000 — before accounting for capital expenditure, staffing, cooling, or facility overhead.
The math at the machine level is equally punishing. A farm of 10 liquid-cooled Antminer S23Hyd units — currently the most efficient hardware available at 9.5 J/TH — consumes approximately 854,400 kWh of electricity to mine one Bitcoin over roughly 150 days. At $0.06/kWh, that translates to $51,264 in pure electricity costs. But the global weighted average efficiency is far worse: approximately 28 J/TH across the network's installed base of older-generation ASICs. For operators running anything below S21-tier hardware, every block mined is mined at a loss.
The result has been a massacre in public mining equities. Marathon Digital Holdings (MARA) trades near $8.09, down approximately 43% over the past 12 months. Riot Platforms hovers around $13.52. CleanSpark has fallen to $9.94. Across the sector, mining stocks posted double-digit losses in the first week of February as investors priced in the reality that these companies are burning cash.
On February 9, 2026, Bitcoin's mining difficulty dropped 11.16% — from 141.6 trillion to 125.86 trillion at block 935,424. This was the largest single downward adjustment since China's sweeping mining ban in mid-2021 forced roughly 50% of global hashrate offline overnight.
The February 2026 drop was triggered by a convergence of two forces:
Price collapse. Bitcoin fell from highs near $126,000 in late 2024 to approximately $69,500 by early February 2026 — a 45% decline that pushed the vast majority of miners below breakeven.
Winter Storm Fern. An arctic weather system blanketed much of the United States with extreme cold, widespread power outages, and record electricity demand for heating. Grid operators in Texas — home to the highest concentration of U.S. mining capacity — issued emergency curtailment requests, forcing miners to power down to preserve grid stability. At the storm's peak, as much as 40% of the global network's hashrate went offline. The total computational power plunged from around 1.1 ZH/s to as low as 663 EH/s — a 30-40% drop in a matter of days. Some public miners reported daily Bitcoin production falling more than 60% during the worst of the storm.
The 11% difficulty reduction provided temporary relief, lifting hashprice back to approximately $35/PH/s. But that relief is about to evaporate. The next difficulty adjustment, estimated for February 19-20, 2026, is projected to increase difficulty from 125.86T to between 139T and 144T — a 10%+ upward swing — as hashrate has already recovered more than 20% in recent weeks, climbing back above 1,054 EH/s.
This whipsaw dynamic — relief followed by immediate re-compression — is the hallmark of a shakeout cycle. Marginal miners who returned to the network during the brief difficulty respite will face a fresh squeeze within days.
The financial stress has translated into the most aggressive miner selling in five years. Bitcoin miner reserves have dropped to approximately 1.801 million BTC, approaching historical lows, with about 6,300 BTC flowing out over just the past 60 days.
The selling has not been gradual. On February 5 alone, miners transferred 28,605 BTC — approximately $1.8 billion — off their balance sheets, marking one of the largest single-day miner outflows since November 2024. The following day, another 20,169 BTC ($1.4 billion) departed miner-associated wallets.
Notable forced sales include:
On-chain analytics firm Checkonchain described the movement as a "rapid capitulation event with high volume and large realised losses" that fundamentally shifted market psychology. This is the classic miner death spiral scenario: falling prices force treasury liquidation, which adds sell pressure, which drives prices lower, which forces more liquidation.
The mining sector's response to this existential pressure has been the most dramatic strategic realignment in its history: a wholesale pivot toward artificial intelligence and high-performance computing infrastructure.
Bitfarms has announced it will exit Bitcoin mining entirely. The company posted a $46 million third-quarter loss on $68 million in revenue and has detailed a phased plan to dismantle crypto mining operations through 2026-2027 while building a dedicated AI-HPC business. Bitfarms has secured a $128 million, fully funded deal with a U.S.-based data center partner to retrofit its 18 MW mining farm for Nvidia GB300 GPUs with advanced liquid cooling, targeting completion by December 2026.
Core Scientific, which filed for bankruptcy in 2022 as a crypto miner, has reinvented itself as a bare-metal AI infrastructure provider. The company signed a 12-year, 200 MW deal with CoreWeave and plans over 300 MW of AI capacity by 2026, offering H100 cluster rentals to AI startups and research labs. Core Scientific's post-bankruptcy valuation is now driven primarily by its AI contracts, not its mining operations.
More than a dozen public Bitcoin miners have now shifted large portions of their operations to AI infrastructure since 2024. The economic logic is compelling: AI compute contracts offer 3-5 year terms with predictable revenue, versus Bitcoin mining's exposure to volatile spot prices, quadrennial halvings, and difficulty adjustments that can swing profitability by double digits overnight.
The pivot raises a fundamental question for Bitcoin's security model. If the most efficient, well-capitalized mining companies are rationally exiting proof-of-work for AI, who will secure the network? The answer, increasingly, is smaller, less efficient operators in low-cost energy jurisdictions — exactly the fragmentation that could undermine network resilience.
Viewed through the lens of economic value distribution, Bitcoin mining's crisis is structural, not cyclical. The foundational data is damning:
Bitcoin's security model requires approximately $54-72 billion annually in total economic input — mining hardware, electricity, facilities, staffing — to generate just $115 million in organic fee revenue. The remaining $18+ billion comes from inflationary block rewards that dilute all BTC holders. This is the most extreme subsidy dependency in the blockchain ecosystem.
The April 2024 halving cut block rewards in half, but network energy consumption continued rising as hashrate surged to accommodate bullish sentiment. The network now consumes approximately 25 gigawatts of continuous power — daily consumption exceeding that of Belgium. When prices crashed, this bloated cost structure became a liability rather than an asset.
The 854,400 kWh required to mine a single Bitcoin in 2026 represents one of the most energy-intensive economic activities on Earth per unit of output. At a time when electricity competition from AI data centers is driving industrial power costs higher in key mining regions (Texas, Quebec, Scandinavia), the economic viability of proof-of-work mining is being structurally undermined by a more lucrative alternative use for the same physical infrastructure.
Despite the carnage, Bitcoin's mining difficulty is about to snap back upward. The estimated 10%+ increase on February 19-20 indicates that hashrate has substantially recovered from Winter Storm Fern's disruption. The network's total computational power has climbed back above 1,054 EH/s, a 5.1% increase in just the past week.
This rebound is a double-edged signal. On one hand, it demonstrates the network's resilience — hashrate returned faster than after the 2021 China ban. On the other hand, it means the brief window of improved economics for surviving miners is closing rapidly. The difficulty increase will push hashprice back down toward the $30-32/PH/s danger zone, potentially triggering another round of capitulation.
JPMorgan's mining analysts have identified $77,000 as a critical support level — the price at which the most efficient miners can sustain operations. Below that threshold, the sector faces cascading shutdowns. Bitcoin currently trades approximately 10% below this line.
Bitcoin mining is operating at a structural loss. With spot prices 20% below the $87,000 average production cost and hashprice at record lows below $32/PH/s, the majority of miners are burning cash.
The 11% difficulty drop was the largest since China's 2021 ban. It provided temporary relief, but the upcoming 10%+ difficulty increase on February 19-20 will re-compress margins immediately.
Miners are liquidating treasuries at the fastest pace in five years. Approximately $3.2 billion in BTC left miner wallets in a two-day window (February 5-6), with total reserves approaching historical lows at 1.801 million BTC.
The AI pivot is no longer diversification — it is existential. Bitfarms is exiting mining entirely; Core Scientific has reinvented itself as an AI provider. More than a dozen public miners have materially shifted to AI infrastructure.
Bitcoin's security model faces a structural test. With fee revenue covering less than 1% of security costs and the most efficient operators migrating to AI, the long-term sustainability of proof-of-work mining depends on either a substantial price recovery or a fundamental rethinking of the economic model.
The February 2026 mining crisis is not a black swan. It is the predictable outcome of a halving cycle colliding with peak hashrate, rising energy costs, and a superior economic alternative in AI compute. The industry that was built to secure Bitcoin is now rationally choosing to secure Nvidia GPUs instead.
For Bitcoin, the implications are profound. The network's security has always been purchased with inflation — newly minted coins subsidizing miners to the tune of $18 billion annually. As that subsidy halves every four years while transaction fees remain negligible, each halving becomes a harder test of whether the market will support the price appreciation necessary to maintain mining economics.
The current crisis suggests the answer is no — at least not at $70,000. The miners who survive will be those with sub-$0.04/kWh power costs, latest-generation hardware, and diversified revenue streams. Everyone else is either pivoting to AI or heading for shutdown.
The question that should concern every Bitcoin holder: if the smartest money in mining is leaving for AI, who is left to secure the chain?