The Bitcoin mining industry is undergoing the most severe structural transformation in its history. With BTC prices down approximately 50% from October 2025 highs and average production costs per coin exceeding $87,000 — roughly 20% above current market prices — miners face a stark choice: adapt ...
"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)
The Bitcoin mining industry is undergoing the most severe structural transformation in its history. With BTC prices down approximately 50% from October 2025 highs and average production costs per coin exceeding $87,000 — roughly 20% above current market prices — miners face a stark choice: adapt or die.
The numbers tell the story. Revenue per petahash has halved from $70 to $35 in six months. Mining difficulty experienced its largest single drop since China's 2021 ban — 11% — as operators shut down en masse. Network hashrate has fallen 15% from October peaks, and daily industry-wide revenue has cratered to $28 million, its lowest since the months immediately following the April 2024 halving.
But from this wreckage, a new industry is being born. Former Bitcoin miners are pivoting to AI infrastructure at breathtaking speed, with companies like IREN securing $9.7 billion Microsoft contracts and MARA Holdings partnering with Starwood Capital to build 2.5 gigawatts of AI-ready data centers. The economic logic is irresistible: AI workloads generate up to 25 times more revenue per megawatt than Bitcoin mining. By end of 2026, mining revenue is projected to fall from 85% of total industry revenue to less than 20% for companies that have secured AI contracts. This is not a pivot — it is a mass industrial migration.
The current mining crisis is the product of three converging forces: the April 2024 halving, which cut block rewards from 6.25 to 3.125 BTC; a macroeconomic shock driven by President Trump's 15% global tariff announcement in February 2026 that cratered risk assets; and rising energy costs that have pushed average production costs above market price.
The numbers are severe. According to industry data, the average all-in production cost for one Bitcoin now sits near $87,000, while BTC trades around $65,000–$69,500. That represents a negative margin of approximately 20–25% for the average miner. Revenue per petahash per day — the industry's standard profitability metric, known as hashprice — has fallen from $70 at peak to approximately $35, a 50% decline in six months.
Daily industry-wide mining revenue dropped to $28 million in late January 2026, its lowest level since the months immediately following the halving. For context, this same metric sat above $60 million during the $126,000 BTC peak in October 2025.
The macro backdrop made everything worse. Bitcoin fell over 5% within hours of the February 23 tariff announcement, dropping below $65,000. The broader crypto market shed $800 billion in market capitalization during February alone. The Fear & Greed Index hit 11 — deep into "Extreme Fear" territory — and Polymarket prediction markets placed a 75% probability on Bitcoin falling below $55,000 at some point in 2026.
The hashrate data confirms widespread capitulation. Bitcoin's network hashrate has dropped 15% from its October 2025 all-time high, with mining difficulty posting an 11% decline in February 2026 — the largest single adjustment since China banned mining in mid-2021.
This is not merely a price-driven slowdown. Severe winter storms across the U.S. in January 2026, particularly impacting the Texas ERCOT grid where a significant concentration of mining operations are located, forced widespread curtailment. The combination of unprofitable economics and forced shutdowns created a cascading exit.
Multiple publicly traded miners have either closed entirely or liquidated their Bitcoin operations:
American Bitcoin Corp. reported a $59 million Q4 2025 loss with its stock falling 90% from September 2025 peaks. The company's BTC reserves suffered a $227 million unrealized loss as prices fell from $126,000 to ~$60,000.
Bitfarms announced a complete wind-down of Bitcoin mining operations in February 2026 after absorbing $46 million in losses in H2 2025. The company is rebranding as Keel Infrastructure and redomiciling from Canada to the United States, with completion expected by April 1, 2026.
Multiple smaller operations folded entirely, with some liquidating entire Bitcoin treasuries — selling all newly mined coins plus over 1,100 BTC in reserves — before shutting down.
The miner capitulation is visible on-chain as well. Miners are selling BTC at elevated rates to cover operational expenses, a classic indicator of industry stress that has historically preceded either market bottoms or prolonged bear markets.
The most consequential development is not the mining crisis itself — it is the industrial-scale migration from Bitcoin hashing to AI infrastructure. The economic rationale is overwhelming: AI workloads can generate up to 25 times more revenue per megawatt than Bitcoin mining, according to industry analysts.
This gap explains the staggering pace of capital reallocation. Sector-wide data center capital expenditure has increased 400% between March 2025 and February 2026. Mining revenue, which constituted roughly 85% of total revenue for diversified mining companies in early 2025, is projected to fall below 20% by end of 2026 for companies that have secured AI contracts.
The fundamental insight driving this migration is simple: Bitcoin miners already own what AI companies desperately need — megawatt-scale power capacity in permitted facilities with grid interconnections. These are assets that take years to develop from scratch. The power infrastructure that was built to run SHA-256 ASICs can be repurposed to host NVIDIA GPUs at dramatically higher margins.
This transition mirrors the economic-value-distribution dynamics observed across blockchain infrastructure more broadly. Just as on-chain value flows fragment across validators, MEV searchers, and infrastructure providers, the physical infrastructure layer is now fragmenting its revenue across mining and AI hosting — with the latter capturing an ever-larger share.
The deal flow in Q1 2026 has been extraordinary:
IREN (formerly Iris Energy) — The poster child of the pivot. IREN secured a landmark $9.7 billion AI cloud contract with Microsoft, covering 200 MW of IT load at its Childress campus, generating an expected $1.94 billion in annual recurring revenue. Q1 FY26 revenue hit a record $240.3 million, up 355% year-on-year. The company plans to scale from 23,000 to 140,000 GPUs by end of 2026, targeting $3.4 billion in AI cloud ARR — all while using only 16% of its 3 GW of secured power capacity.
MARA Holdings — On February 26, 2026, MARA announced a strategic partnership with Starwood Capital Group to jointly develop, finance, and operate digital infrastructure across MARA's power-rich portfolio. The partnership targets 1 GW of immediate capacity with potential expansion to 2.5 GW. MARA shares jumped 17% on the announcement — despite the company simultaneously reporting a $1.7 billion quarterly loss on its mining operations.
Riot Platforms — Activist investor Starboard Value LP released a public letter on February 18, 2026, arguing that Riot's 1.7 GW of available power capacity could generate more than $1.6 billion in annual EBITDA if converted to AI hosting, with a potential enterprise value of $21 billion. Riot has already signed a lease with AMD at its Rockdale, Texas facility — initially 25 MW of critical IT load with options to expand to 200 MW — guaranteeing approximately $311 million in revenue over the contract term, with total potential value reaching $1 billion.
Bitfarms / Keel Infrastructure — The most dramatic rebrand: completely exiting Bitcoin mining, redomiciling to the United States, and repositioning as a pure-play HPC/AI infrastructure company. CEO Ben Gagnon explicitly stated the company is "no longer a Bitcoin company."
Not every miner is exiting. The difficulty adjustment mechanism, which dropped 11%, is working as designed — by reducing competition, it improves margins for those who remain. But survival requires extreme efficiency.
The current break-even threshold demands:
Mining ROI timelines have stretched to approximately 1,000 days for new hardware deployments, compared to roughly 300 days during the 2024 bull run. This makes new capacity investment nearly impossible to justify unless BTC price recovers substantially.
The hashrate decline raises legitimate questions about network security, though the situation is far from critical. Even with a 15% decline from peak, Bitcoin's network hashrate remains orders of magnitude above levels from just two years ago. The difficulty adjustment mechanism ensures block production stays close to the 10-minute target regardless of how many miners exit.
However, the concentration of surviving hashrate among fewer, larger operators does introduce centralization risk. If the AI pivot continues at its current pace, Bitcoin mining could increasingly become a secondary revenue stream — or an afterthought — for companies whose primary business is AI hosting. This changes the incentive structures that have historically underpinned network security.
The long-term question is whether Bitcoin's fee market can mature enough to sustain mining economics once block rewards diminish further with the next halving in 2028. At current fee levels, the answer is clearly no.
Bitcoin mining is underwater: Average production cost of ~$87K per BTC vs. market price of ~$65–69K represents a 20–25% negative margin for the average miner.
The capitulation is historic: An 11% difficulty drop — the largest since China's 2021 ban — and 15% hashrate decline from peak confirm widespread shutdowns.
AI economics are irresistible: At 25x revenue per megawatt versus mining, the pivot is economically rational. Companies are securing billion-dollar AI contracts (IREN's $9.7B Microsoft deal, MARA's Starwood partnership, Riot's AMD lease).
The industry is bifurcating: By end of 2026, miners with AI contracts will derive less than 20% of revenue from mining. Pure-play miners face extinction unless BTC recovers above $90K.
Network security implications are real but manageable: The difficulty adjustment mechanism works as designed, but increasing hashrate concentration among fewer operators introduces long-term centralization risk.
The Bitcoin mining industry of 2024 — a sector defined by maximizing hashrate per dollar — is being replaced by something fundamentally different: a power infrastructure industry that happens to also mine Bitcoin. The economic gravity of AI compute demand is too strong for miners to resist, particularly when their core asset — permitted, grid-connected megawatt capacity — is precisely what AI hyperscalers need most.
For investors, this means evaluating former mining companies on an entirely different set of metrics: power pipeline, AI contract backlog, GPU deployment timelines, and hyperscaler relationships. For the Bitcoin network, it means confronting the reality that mining is becoming a by-product rather than a primary business for the industry's largest participants.
The miners who survive this cycle will not be those who mine the most Bitcoin. They will be those who most efficiently convert their power assets into AI revenue while maintaining Bitcoin mining as a flexible, opportunistic complement. The industry's identity has been permanently rewritten.