Bitcoin's network hashrate has declined approximately 21% from its October 2025 peak of 1,160 EH/s to roughly 920 EH/s, triggering three consecutive negative difficulty adjustments — the first such streak since July 2022. The protocol has automatically reduced difficulty six times in 2026, cuttin...
"The conversion of just our Washington site to GPU-as-a-Service could potentially produce more net operating income than we have ever generated with Bitcoin mining." — Ben Gagnon, CEO, Bitfarms
Bitcoin's network hashrate has declined approximately 21% from its October 2025 peak of 1,160 EH/s to roughly 920 EH/s, triggering three consecutive negative difficulty adjustments — the first such streak since July 2022. The protocol has automatically reduced difficulty six times in 2026, cutting it 10.7% from the start of the year, with a further 9% reduction estimated around June 13. Q1 2026 marked the first quarter-over-quarter hashrate decline in six years, according to CoinShares.
The cause is structural, not cyclical. Public mining companies have collectively signed over $70 billion in AI and high-performance computing contracts, redirecting megawatts from block production to GPU co-location. CoinShares projects that listed miners will derive up to 70% of revenue from AI infrastructure by year-end 2026, up from approximately 30% at the start of the year. Mining stocks have responded: a basket of public miners rose more than 50% year-to-date while Bitcoin itself fell roughly 17%.
The economics are straightforward. The weighted average cash cost to produce one bitcoin among public miners reached $79,995 in Q4 2025, according to CoinShares. Bitcoin trades near $67,000. AI cloud infrastructure generates project-level EBITDA margins of approximately 85% with multi-year revenue visibility. Bitcoin mining margins have compressed from roughly 90% during the 2021 cycle peak to around 60% today.
Bitcoin's network computing power peaked at approximately 1,160 EH/s in early October 2025. By late December 2025, it had dropped roughly 10%. As of June 2026, hashrate sits near 895 EH/s, according to CoinWarz data at block height 952,625, with difficulty at 138.96 trillion.
The next difficulty adjustment, estimated for June 13, 2026, projects a decline from 138.96T to approximately 125.94T — a reduction of close to 9%. This would represent one of the larger downward moves logged in 2026. The protocol has already executed six negative adjustments this year.
Q1 2026 marked the first quarter-over-quarter decline in hashrate since Q1 2020, according to CoinDesk, citing data from multiple on-chain analytics providers. The drop of roughly 4% quarter-over-quarter broke a six-year streak of consecutive quarterly gains.
CoinShares projects that hashrate could recover to 1.8 ZH/s by year-end 2026 if Bitcoin rebounds toward $100,000. At current prices, that projection appears optimistic.
Hash price — the daily revenue per petahash of mining capacity — has collapsed. After peaking at approximately $63/PH/s/day in July 2025, it fell steadily through Q4 2025 to $35–37/PH/s/day by November. By early March 2026, it briefly touched $28/PH/s/day, a new post-halving all-time low.
At these levels, mid-generation hardware requires access to electricity priced below $0.05/kWh to remain cash-flow positive. CoinShares estimates that 15–20% of global mining capacity — primarily older-generation rigs — currently operates at a loss. Only latest-generation fleets with efficiency below 15 J/TH retain meaningful margin at typical industrial electricity rates.
The current hash price environment represents what CoinShares describes as "the tightest margin environment since the April 2024 halving." Further capitulation among higher-cost operators is expected in H1 2026 unless BTC price recovers materially.
The scale of capital reallocation is substantial. Public Bitcoin miners have announced over $70 billion in cumulative AI and HPC contracts since 2025. These are not letters of intent; they are signed multi-year leases and co-location agreements with hyperscale cloud providers.
The economic logic is clear. Bitcoin mining infrastructure costs $700,000 to $1 million per megawatt. AI infrastructure costs $8–15 million per megawatt. AI contracts generate approximately three times the revenue on a per-megawatt basis compared to mining operations, with EBITDA margins near 85% versus mining's compressed 60%.
Miners already hold two assets that AI infrastructure demands: large blocks of energized capacity and physical facilities designed for heavy compute and thermal management. Retrofitting these sites for GPU workloads is faster and cheaper than building greenfield AI data centers.
As of May 2026, nine publicly listed mining companies have formally announced AI pivots in various forms: Bitfarms (rebranded as Keel Infrastructure), IREN, Core Scientific, CleanSpark, HIVE Digital, TeraWulf, Cango, Cipher Mining, and Riot Platforms.
Core Scientific (CORZ): Expanded its CoreWeave contract to $10.2 billion over 12 years. AI colocation revenue already accounts for 39% of total revenue, the highest ratio among public miners. The company expects AI/HPC to reach 70% of revenue by year-end 2026.
TeraWulf (WULF): Has locked in $12.8 billion in total contracted HPC revenue. In May 2026, acquired the Muskie Data Campus in Eastern Kentucky, expected to support more than 1 GW of data center capacity. Carries $5.7 billion in total debt. AI revenue currently at 27% of total.
Hut 8 (HUT): Signed a 15-year lease agreement in May 2026 for the first phase of its Beacon Point AI data center campus in Texas, covering 352 MW of IT capacity with a base contract value of $9.8 billion. Also holds a $7 billion, 15-year lease at its River Bend campus.
IREN (IREN): Secured a $3.4 billion, five-year AI Cloud contract with NVIDIA for air-cooled Blackwell GPUs. Also holds a Microsoft partnership projected to generate $1.94 billion in annualized revenue at an 85% project-level EBITDA margin. Carries $3.7 billion in convertible notes. On track for 480 MW expansion with $3.1 billion annual recurring revenue under contract, targeting $3.7 billion ARR by end of CY2026. AI currently 9% of revenue.
Cipher Mining (CIFR): Holds a 15-year, 300 MW direct lease with AWS, expected to generate $5.5 billion in revenue. Announced $11.4 billion in total contracted revenue and approximately $787 million in average annualized contracted NOI. Issued $1.7 billion in senior secured notes; quarterly interest expense jumped from $3.2 million to $33.4 million in Q4.
Riot Platforms (RIOT): Operates at 42.5 EH/s hashrate. Signed a 10-year AMD lease for AI infrastructure. Stock up 65–80% year-to-date.
Bitfarms (now Keel Infrastructure): Fully exited Bitcoin mining. CEO Ben Gagnon stated the Washington site conversion alone could produce more net operating income than the company has ever generated from mining.
The transition is not free. Miners are funding AI infrastructure build-outs through a combination of BTC treasury sales, debt issuance, and equity dilution.
Public miners collectively reduced BTC treasuries by more than 15,000 coins from peak holdings. Core Scientific sold approximately 1,900 BTC ($175 million) in January 2026. Marathon Digital expanded authorization for balance sheet BTC sales, with its loan-to-value ratio reaching 87%.
Leverage is rising across the sector. IREN carries $3.7 billion in convertible notes. TeraWulf holds $5.7 billion in total debt. Cipher issued $1.7 billion in senior secured notes, increasing quarterly interest expense more than tenfold.
The risk is execution. If AI contract ramp-ups lag power delivery timelines or if hyperscaler demand moderates, these balance sheets face strain. The current model depends on contracted revenue materializing on schedule.
A 21% decline in hashrate from peak raises questions about network security, though the risk level requires context. Bitcoin's hashrate at 895 EH/s remains orders of magnitude above historical norms. Even a further 20% decline would leave the network at approximately 716 EH/s — still higher than any point prior to mid-2024.
However, the concentration risk is shifting. As public miners redirect capacity toward AI, a larger share of remaining hashrate comes from private operators, many in jurisdictions with less transparency. The geographic distribution of mining power is evolving in ways that are not fully visible to on-chain analytics.
The three consecutive negative difficulty adjustments in early 2026 were the first such streak since July 2022. In isolation, difficulty adjustments are the protocol functioning as designed — self-correcting to maintain 10-minute block intervals. But the sustained direction signals a structural supply reduction, not a temporary fluctuation.
The market's verdict on the pivot is unambiguous. While Bitcoin declined approximately 17% through the first months of 2026, mining stocks moved in the opposite direction.
A basket of public miners rose more than 50% year-to-date, with top performers gaining over 70%. IREN climbed from an annual low of $31.62 to approximately $52, nearing all-time highs. Riot Platforms gained 65–80% year-to-date. Cipher Mining surged after its AI data center shift, backed by hyperscale leases.
The market is pricing these companies as AI infrastructure plays, not Bitcoin proxies. The correlation between BTC price and mining stock performance, historically tight, has broken down. Miners with larger AI contract backlogs trade at higher revenue multiples than those still primarily dependent on block rewards.
This repricing implies that the market assigns more value to $1 of contracted, high-margin AI revenue than to $1 of volatile, margin-compressed mining revenue.
The Bitcoin mining sector is undergoing a capital reallocation of historic scale. The economic arbitrage is straightforward: AI infrastructure generates 85% EBITDA margins with contracted multi-year visibility, while mining produces compressed 60% margins with full exposure to price and difficulty volatility.
The transition creates a temporary paradox. As hashrate declines, remaining miners benefit from lower difficulty and improved per-unit economics — until the price at which those economics work falls below market. The protocol's difficulty adjustment mechanism continues to function, but the assumption that hashrate growth is monotonic has been invalidated for the first time in six years.
The central risk is not to the Bitcoin network, which remains functional at far lower hashrate levels. The risk is to the miners themselves. The $70 billion in contracts depends on power delivery, GPU supply chains, and sustained hyperscaler demand. Any disruption to these variables creates balance sheet stress for companies that have already depleted BTC reserves and expanded debt to fund the transition.
What the market is pricing is a sector that found a more productive use for its primary asset — cheap, large-scale power infrastructure — than the one it was built for.