Bitcoin mining is experiencing its deepest profitability crisis since the 2021 China ban. With the average cost to mine one bitcoin now estimated at $87,000 and spot prices hovering near $68,000, the entire industry is operating roughly 20% below breakeven. The result: the largest mining difficul...
"We are no longer a Bitcoin company. We are an infrastructure-first owner and developer for HPC/AI data centers across North America." — Ben Gagnon, CEO, Bitfarms (now Keel Infrastructure)
Bitcoin mining is experiencing its deepest profitability crisis since the 2021 China ban. With the average cost to mine one bitcoin now estimated at $87,000 and spot prices hovering near $68,000, the entire industry is operating roughly 20% below breakeven. The result: the largest mining difficulty drop in nearly five years, a 15–20% decline in network hashrate from October 2025 peaks, and the first full-company exit from Bitcoin mining in this cycle — Bitfarms, now rebranding as Keel Infrastructure, announced in February 2026 that it will wind down all mining operations and convert entirely to AI data centers.
This is not a temporary dip. The structural economics of post-halving Bitcoin mining have collided with a prolonged price correction — BTC has fallen more than 45% from its October 2025 all-time high above $126,000 — and an energy market reshaped by AI demand. The result is an accelerating bifurcation: miners are either converting their power infrastructure to serve the artificial intelligence industry or facing existential margin compression. Mining revenue per petahash has collapsed from $70 to roughly $35, hashprice has hit $0.034/TH/s (down 35% year-over-year), and new hardware ROI has stretched beyond 1,000 days — past the next halving horizon.
For the first time in Bitcoin's history, the more profitable use of mining infrastructure may not be mining Bitcoin.
The data paints an unambiguous picture of industry-wide distress:
| Metric | Current (Feb 2026) | Peak (Q3 2025) | Change | |--------|-------------------|-----------------|--------| | Bitcoin spot price | ~$68,000 | ~$126,000 | -46% | | Avg. production cost per BTC | ~$87,000 | ~$52,000 | +67% | | Hashprice ($/TH/s) | $0.034 | $0.055 | -38% | | Revenue per PH/s/day | ~$35 | ~$70 | -50% | | Network hashrate | ~950 EH/s | ~1.2 ZH/s | -20% | | New hardware ROI | 1,000+ days | ~400 days | +150% |
Bitcoin is trading approximately 20% below its average production cost, according to data from Checkonchain. Large public miners report all-in costs near $44/PH/s/day while generating only $35–38 in daily revenue. When miners spend more to produce bitcoin than bitcoin is worth, only three outcomes are possible: sell reserves, shut down machines, or find alternative revenue.
All three are happening simultaneously.
On February 9, 2026, Bitcoin's mining difficulty dropped 11.16% to 125.86 trillion at block height 935,424 — the largest single negative adjustment since China banned crypto mining in the summer of 2021, and reportedly the 10th largest in Bitcoin's entire history.
The drop was driven by two converging forces. First, the prolonged price decline forced marginal miners offline. Second, Winter Storm Fern in late January forced miners across U.S. power regions to curtail operations to support strained residential grids. Some companies reported daily bitcoin production falling by more than 60% during the harshest storm days.
The difficulty adjustment prior to this had already been negative, marking the seventh decline in the previous eight adjustment periods. The network hashrate fell to a seven-month low of 663 EH/s during the worst of the storm disruptions before partially recovering.
JPMorgan's mining research team noted in January that the network hashrate had fallen for two consecutive months in December and January — a pattern not seen since the 2022 bear market. Yet even as difficulty eased, the bank observed that miners "continue to face dwindling profits despite lower competition," because the price decline has outpaced the difficulty adjustment's ability to restore margins.
The most consequential development in the mining sector is not about mining at all. It is the wholesale conversion of Bitcoin mining infrastructure into artificial intelligence data centers.
Bitfarms — The First Full Exit. On February 6, 2026, Bitfarms declared it is "no longer a Bitcoin company." The firm announced plans to wind down all mining operations over 2026–2027, rebrand as Keel Infrastructure, relocate its legal base from Canada to the United States, and convert its 18 MW Washington State facility to host Nvidia GB300 GPUs with advanced liquid cooling. The pivot is backed by a fully funded $128 million deal with a major U.S.-based data center partner.
Core Scientific — The $6.7 Billion Template. Core Scientific, which emerged from bankruptcy in 2024, has built a 12-year, $6.7 billion hosting contract with CoreWeave, an Nvidia-backed GPU cloud provider. The company plans to deliver approximately 382 MW of infrastructure to CoreWeave by first-half 2026. Its AI hosting revenue is expected to dwarf its remaining Bitcoin mining income.
Hut 8 — The Google-Backed Transformation. Hut 8 signed a 15-year lease agreement with Fluidstack worth approximately $7 billion to provide data center space and power for AI computing. The deal represents one of the largest infrastructure commitments in the sector's history.
Cipher Mining — The AWS Pipeline. Cipher has secured a 15-year, 300 MW direct lease with AWS, expected to generate $5.5 billion in revenue. JPMorgan upgraded the stock to Overweight, projecting 480 MW of critical IT capacity by 2026.
CoinShares' 2026 mining outlook projects that for companies with AI contracts, mining revenue will plummet from approximately 85% of total revenue in early 2025 to less than 20% by end of 2026. The economics are straightforward: AI hosting offers contracted, dollar-denominated, multi-year revenue streams — a fundamentally different risk profile than the volatile, commodity-dependent economics of Bitcoin mining.
The current environment is creating a three-tier hierarchy among Bitcoin miners:
Tier 1: AI Converts — Companies like Core Scientific, Bitfarms/Keel, Hut 8, Cipher, and TeraWulf that have secured long-term AI/HPC contracts. These companies' stock prices have generally outperformed pure-play miners. They are effectively becoming power and infrastructure companies that happen to still mine some Bitcoin.
Tier 2: Efficient Pure-Play Survivors — Miners with electricity costs below $0.05/kWh, next-generation ASICs under 20 J/TH, and strong balance sheets. CleanSpark falls into this category, having been upgraded by JPMorgan based on 200 MW of HPC potential at its Texas site. Marathon (MARA) and Riot Platforms are straddling this tier and the AI convert tier, maintaining large mining fleets while exploring HPC opportunities.
Tier 3: The Walking Dead — Miners with older hardware, higher electricity costs, and limited capital. The ZeroLend shutdown on February 17, while a DeFi protocol rather than a miner, illustrates the broader principle: in a margin-compression environment, entities with thin economics and high operational overhead simply cannot survive. For mining, this means operators running S19-class or older ASICs at electricity costs above $0.07/kWh are mathematically unprofitable at current hashprices.
Goldman Sachs' 13F filing for Q4 2025 revealed a 39.4% reduction in Bitcoin ETF holdings — the bank trimmed its IBIT position from 33.9 million to 20.7 million shares. While Goldman simultaneously entered XRP and Solana ETF positions, the Bitcoin reduction signals institutional awareness that the mining-dependent security model is under pressure.
Amid the carnage, one technical indicator has attracted significant attention. The Hash Ribbon — which tracks miner capitulation by comparing the 30-day and 60-day moving averages of Bitcoin's hashrate — inverted on November 29, 2025, shortly after Bitcoin bottomed near $80,000. It subsequently flashed a buy signal in early January 2026 near $95,000.
Historically, Hash Ribbon buy signals have preceded significant price recoveries. The 2020 signal coincided with a 194–567% return within 12 months. Since 2013, 14 buy signals have been recorded with a 64% win rate. However, the current signal was followed by a further decline to $68,000, and BTC's price action has not yet confirmed the typical post-capitulation recovery pattern.
The signal's logic is sound in principle: when inefficient miners capitulate and difficulty drops, the remaining miners become more profitable, reducing sell pressure and creating conditions for price recovery. But this cycle introduces a variable that previous cycles did not: the AI pivot means that departing miners are not merely shutting down — they are permanently reallocating their power infrastructure away from Bitcoin. This capacity may never return to the network, structurally capping future hashrate growth and altering the long-term security model.
Viewed through an economic value lens, the mining crisis exposes the fragility of Bitcoin's security budget. The network currently generates approximately $3.1 billion in annualized transaction fees — a fraction of the roughly $18 billion in annual mining issuance that subsidizes network security. When miners operate below breakeven, the issuance subsidy that was designed to bootstrap the network is instead being spent below cost.
This creates a perverse dynamic: the more BTC's price falls below production cost, the more the issuance subsidy is effectively being wasted — transferred to miners who are destroying economic value with every block they produce at a loss. The rational economic response is exactly what we are observing: miners exit, difficulty drops, and the network finds a new equilibrium at lower security levels.
The AI pivot adds a structural dimension. When miners convert facilities from Bitcoin mining to AI hosting, they are making a revealed-preference statement about the relative economic value of securing the Bitcoin network versus serving AI compute demand. Core Scientific's $6.7 billion CoreWeave contract alone represents roughly 2x the Bitcoin network's entire annual transaction fee revenue. The market has priced AI infrastructure work as fundamentally more valuable than Bitcoin block production.
With the next halving expected in 2028, cutting the block subsidy to 1.5625 BTC, the mining industry's current trajectory raises legitimate questions about whether Bitcoin's security budget can sustain adequate hashrate protection without either a dramatic price increase or a fundamental shift in fee revenue.
The Bitcoin mining industry is not experiencing a cyclical downturn. It is undergoing a structural transformation. The combination of post-halving economics, a 46% price decline from all-time highs, and the emergence of AI as a higher-value use of power infrastructure has created conditions where the rational economic decision for mining companies is to stop mining Bitcoin.
This transformation carries profound implications for Bitcoin's long-term security model. If the most well-capitalized, lowest-cost mining operators are systematically redirecting their infrastructure toward AI, the assumption that hashrate will continue to grow in line with Bitcoin's market cap deserves serious scrutiny. The network's security budget — currently 85–90% subsidized by block rewards rather than transaction fees — faces its most significant stress test since inception.
For investors and institutions, the mining sector now requires evaluation not as a Bitcoin-derivative play but as an infrastructure allocation decision. The companies that will command premium valuations are those with the most MW under contract for AI/HPC, not those with the most exahashes pointed at SHA-256. In the economic value hierarchy, the market has spoken: artificial intelligence compute is worth more than Bitcoin block production.
The miners who survive this crisis will likely be unrecognizable as mining companies. And the Bitcoin network will need to answer a question it has never faced with this urgency: if the miners leave, who secures the chain?