Bitcoin mining difficulty fell 10.09% on June 13 at block 953,568 — the 11th-largest downward adjustment in network history and the second-largest drop of 2026. The new difficulty of 124.93 trillion is the lowest since July 2025. Network hashrate has declined from a peak of 1.151 ZH/s in October ...
"We want to be redirecting every dollar possible toward AI capex." — Gary Vecchiarelli, President & CFO, CleanSpark
Bitcoin mining difficulty fell 10.09% on June 13 at block 953,568 — the 11th-largest downward adjustment in network history and the second-largest drop of 2026. The new difficulty of 124.93 trillion is the lowest since July 2025. Network hashrate has declined from a peak of 1.151 ZH/s in October 2025 to approximately 886 EH/s by mid-June 2026, a 23% drawdown.
The proximate cause is a roughly 21% decline in BTC price since early May, from $81,000 to approximately $63,800, which pushed spot price roughly 25% below the estimated all-in production cost of $84,300. But the structural cause runs deeper: publicly listed mining companies are systematically reallocating power capacity and capital toward artificial intelligence and high-performance computing (HPC) infrastructure. According to CoinShares' Q1 2026 mining report, listed miners could derive up to 70% of revenue from AI by December 2026, up from approximately 30% at present.
The convergence of margin compression, capital reallocation, and AI infrastructure demand has produced the most significant structural shift in Bitcoin mining economics since the 2024 halving. VanEck's June 16 report identifies a $50 billion near-term funding gap facing miners attempting the transition, with as much as $221 billion in long-term capital requirements — raising questions about which operators survive the pivot and what it means for the network they are leaving behind.
Bitcoin's difficulty adjustment algorithm recalibrates every 2,016 blocks (roughly two weeks) to maintain a 10-minute average block time. When miners leave, blocks slow down, and difficulty falls to compensate.
The June 13 adjustment reduced difficulty from 138.96 trillion to 124.93 trillion — a 10.09% cut. For context, the three largest downward adjustments of 2026 have been:
| Date | Adjustment | New Difficulty | |------|-----------|----------------| | Feb. 7 | -11.16% | ~127T | | Mar. (mid) | -7.76% | ~117T | | Jun. 13 | -10.09% | 124.93T |
The largest downward adjustment in Bitcoin's history remains the 27.94% drop in July 2021 following China's mining ban. The current adjustment is the 11th-largest ever recorded.
The 7-day average hashrate declined from above 1,000 EH/s earlier in 2026 to 886 EH/s by mid-June — a loss of roughly 115 EH/s. Q1 2026 marked the first quarterly hashrate decline since Q1 2020, according to CoinDesk, breaking five consecutive years of double-digit growth.
The next adjustment, estimated for June 27, is projected to increase difficulty modestly to approximately 128.36 trillion, suggesting some hashrate has begun to return at lower difficulty levels.
The economics are straightforward. The April 2024 halving cut the block subsidy from 6.25 BTC to 3.125 BTC. At approximately $63,800, spot BTC sits roughly 25% below the estimated average all-in production cost of $84,300, according to data cited by multiple mining analysts.
For operators paying above $0.10/kWh — a common rate in much of the United States — most air-cooled SHA-256 hardware remains unprofitable even after the difficulty adjustment. The 10.09% cut raises bitcoin produced per unit of active hashrate by approximately 11%, but that improvement is insufficient to offset a 21% price decline.
Revenue per petahash per day (hashprice) hovers around $30/PH/s/day. CoinShares' Q1 2026 report notes production costs near $90,000 per bitcoin for many listed miners when accounting for depreciation, SG&A, and interest expenses — well above spot.
The result is a classic shakeout. Marginal miners with older hardware and higher power costs are shutting down. Efficient operators with sub-$0.06/kWh power and latest-generation ASICs (15-20 J/TH range) remain profitable, but the margin of safety is thin.
The miner-to-AI transition is no longer a strategy deck — it is a capital allocation event of significant scale. CoinShares reports more than $70 billion in cumulative AI and HPC contracts across the publicly listed mining sector. The logic is simple: mining facilities already have three things AI data centers need — power interconnections, cooling infrastructure, and physical space.
Key contract values across the sector:
| Company | AI/HPC Contract Value | Term | Partner/Tenant | |---------|----------------------|------|----------------| | Core Scientific | $10.2B | 12 years | CoreWeave | | TeraWulf | $12.8B | Various | Multiple | | Hut 8 | $7.0B | 15 years | Various (River Bend) |
The conversion is already showing in revenue mix. Core Scientific reported Q1 2026 AI colocation revenue of $77.5 million — now its largest business line — while crypto mining revenue fell to $30.1 million. AI colocation accounts for 39% of Core Scientific's total revenue. TeraWulf is at 27%.
However, execution remains the constraint. According to VanEck's June 16 report, the industry has delivered only approximately 25% of the AI and HPC capacity it has leased to customers. The gap between announced megawatts and energized megawatts is wide, and capital markets are beginning to differentiate.
Core Scientific (CORZ): The most advanced in the transition. Sold $208 million of bitcoin in Q1 2026 to fund AI infrastructure buildout. Acquired Polaris DS LLC for $421 million, adding 440 MW of contracted power in Oklahoma for AI expansion. Issued $3.3 billion in bonds in April 2026 to finance further data center construction. The CoreWeave relationship — a 590 MW contract expansion projected at $10.2 billion over 12 years — is the anchor.
MARA Holdings (MARA): Reported Q1 2026 revenue of $174.6 million, down 18.3% year-over-year, with a net loss of $1.3 billion. Approximately $1 billion came from unrealized mark-to-market adjustments on digital assets under FASB fair value rules. The company is building AI capacity but has not yet secured contracted tenants at the scale of Core Scientific or TeraWulf. The AI premium in the stock is priced on potential, not contracted revenue.
CleanSpark (CLSK): Posted a $378.3 million net loss in fiscal Q2 (ended March 31, 2026), more than doubling the $138.8 million loss in the prior year period. A $224.1 million fair-value loss on bitcoin holdings accounted for nearly 60% of the quarterly loss. Revenue declined to $136.4 million from $181.7 million year-over-year. Management has signaled intent to redirect capital toward AI but is earlier in the transition than peers.
Riot Platforms (RIOT): Scrapped its 600 MW Phase II Bitcoin mining expansion at the Corsicana facility, trimming its 2025 hashrate growth target from 46.7 EH/s to 38.4 EH/s. The company has engaged Altman Solon to assess feasibility of dedicating the 600 MW toward AI-focused cloud joint ventures. Riot has up to 1 GW of total approved capacity at Corsicana from the Electric Reliability Council of Texas, with 400 MW currently used for mining.
The departure of large-scale miners from Bitcoin hashing raises a recurring question: does the AI pivot compromise network security?
The hashrate has declined 23% from its October 2025 peak. In absolute terms, this represents the loss of roughly 265 EH/s — computational power equivalent to the entire network in early 2024. The 7-day moving average stands at approximately 886 EH/s, according to CoinWarz.
Several considerations are relevant:
Difficulty adjustment provides resilience. The protocol's built-in mechanism ensures that even with reduced hashrate, the cost of a 51% attack scales with the remaining network power. At 886 EH/s, Bitcoin remains by far the most computationally secured proof-of-work network.
Centralization risk may increase. Each miner that exits Bitcoin for AI reduces the pool of independent hashrate contributors, increasing the relative weight of remaining large pools. If the exit is concentrated among U.S.-listed miners, geographic distribution could shift, potentially increasing concentration in other jurisdictions.
The exiting miners are often the least efficient. Older-generation hardware operating at higher power costs tends to go offline first, leaving the network secured by more efficient, purpose-built mining operations. This is the system working as designed.
The scenario is not unprecedented. China's 2021 mining ban removed approximately 50% of global hashrate in a matter of weeks. The network adjusted within two months and hashrate recovered within six months. The current drawdown, at 23%, is less severe by comparison.
CoinShares still forecasts hashrate growth to approximately 1.8 ZH/s by year-end 2026, contingent on bitcoin recovering toward $100,000.
VanEck's June 16 report quantifies the challenge facing the mining-to-AI transition. The firm estimates a $50 billion near-term funding gap and as much as $221 billion in long-term capital requirements for miners attempting to build out AI infrastructure at the scale of their announced contracts.
The arithmetic is unforgiving. Converting mining facilities to Tier III or Tier IV data center standards requires significant capital expenditure per megawatt — networking equipment, redundant power systems, liquid cooling, and physical security upgrades. Miners' existing facilities were built for single-purpose SHA-256 computation, not the reliability and latency requirements of AI inference and training workloads.
VanEck notes that investor focus is shifting from contract announcements to three delivery metrics:
Companies missing construction milestones could face lasting valuation hits, according to VanEck. The market is bifurcating between miners with secured, funded, operational AI capacity and those still in the planning or early-construction phase.
Core Scientific's $3.3 billion bond offering and $421 million Polaris acquisition demonstrate the scale of capital deployment required. Not every miner has the balance sheet or capital market access to execute at this level.
The Bitcoin mining industry is undergoing a structural repricing. The combination of post-halving economics, a sustained BTC price decline, and the gravitational pull of AI infrastructure demand is redrawing the sector's capital allocation map. The June 13 difficulty adjustment — the second time in 2026 that difficulty has dropped by more than 10% — is a symptom, not a cause.
The underlying shift is economic. AI data center contracts offer longer duration, higher visibility revenue relative to the volatile, subsidy-dependent economics of proof-of-work mining. Core Scientific's transformation — from bankruptcy in 2023 to a company generating more revenue from AI colocation than from mining bitcoin — illustrates the speed and scale of the transition.
The open questions are execution and selection. VanEck's $50 billion funding gap estimate suggests that many miners lack the capital to complete the pivot. The market will increasingly sort miners into two categories: those with energized AI capacity and investment-grade tenants, and those still mining bitcoin with diminishing margins while hoping to finance a transition they may not be able to afford.
For the Bitcoin network itself, the difficulty adjustment mechanism continues to function as designed. Hashrate loss is real but manageable. The greater risk is not a security failure but a gradual shift in the identity of the mining industry — from Bitcoin-native operators securing the network to infrastructure companies that view BTC mining as a residual activity while their primary business moves to AI.