Bitcoin's network hashrate fell from 1,030 EH/s on May 28 to 861 EH/s by June 10, 2026 — a 16.4% decline in 13 days. On June 13, the protocol executed a -10.09% difficulty adjustment (138.96T → 124.93T), the 11th largest negative adjustment in Bitcoin's history and the second such drop exceeding ...
"The power landlords of AI — these miners have what hyperscalers want most: permitted power capacity at scale." — Gautam Chhugani, Managing Director, Bernstein
Bitcoin's network hashrate fell from 1,030 EH/s on May 28 to 861 EH/s by June 10, 2026 — a 16.4% decline in 13 days. On June 13, the protocol executed a -10.09% difficulty adjustment (138.96T → 124.93T), the 11th largest negative adjustment in Bitcoin's history and the second such drop exceeding 10% this year. The immediate trigger: BTC's price decline to $60,000 in early June compressed hashprice to approximately $23.9 per PH/s per day, the lowest reading since 2018.
The structural driver is more consequential. Listed mining companies — Core Scientific, TeraWulf, Riot Platforms, IREN, and MARA Holdings — are permanently reallocating power capacity from SHA-256 computation to AI/HPC data center hosting. Collectively, these five firms have secured over $35 billion in AI contracts and are building toward 10+ GW of data center capacity. Mining stocks have outperformed Bitcoin by 70% year-to-date, reflecting market repricing of these companies as infrastructure operators rather than commodity miners.
The economic logic is straightforward: AI workloads generate 3–25x more revenue per megawatt than Bitcoin mining, with 80–90% operating margins versus the 20–50% margins available to even efficient miners at current hashprice. The question is no longer whether miners will pivot, but what happens to Bitcoin's security budget when they do.
At block height 953,568 on June 13, 2026, Bitcoin's difficulty algorithm reduced the mining target by 10.09%. The adjustment followed a period in which average block times stretched past 11 minutes, indicating sustained hashrate loss.
Key metrics:
This was the second >10% negative adjustment in 2026. February saw an 11.16% drop. Before 2026, only three difficulty reductions exceeded 10% in Bitcoin's entire 17-year history: June 2021 (China ban), March 2020 (COVID crash), and November 2018 (hashwar).
The dual occurrence in a single year signals a structural shift beyond normal cyclical pressure. Miners are not merely pausing operations during a price dip — they are physically repurposing infrastructure.
Five listed mining companies account for the majority of the capacity reallocation:
| Company | AI Contract Value | Current AI Capacity | Target Capacity | AI Revenue Share | |---------|------------------|--------------------|-----------------|--------------------| | Core Scientific (CORZ) | $10.2B (CoreWeave) | 350 MW energized | 3.0 GW pipeline | 39% of Q4 revenue | | TeraWulf (WULF) | $12.8B total | 39 MW IT capacity | 2.9 GW (5 sites) | 27% of Q4 revenue | | Riot Platforms (RIOT) | ~$2B+ (AMD lease) | 600 MW evaluating | 1.0 GW total ERCOT | 20% of Q1 revenue | | IREN | $9.7B (Microsoft) | AI Cloud operational | Multiple GW-scale | 9% of Q4 revenue | | MARA Holdings | JV with Starwood | First 100 MW Q3 2026 | 2.5 GW expansion | Transitioning |
The aggregate contract pipeline exceeds $35 billion. Core Scientific's 12-year CoreWeave agreement alone commits 590 MW at full build-out by early 2027. TeraWulf's contracts span five physical locations. Riot's "Project Ditto" at Corsicana, Texas involves a $400 million, 335,430 sq-ft data center filing submitted in April 2026 with completion targeted for 2028.
To fund these transitions, miners have liquidated significant BTC holdings. According to Blockchain Magazine, MARA Holdings sold 13,210 BTC, Riot Platforms divested 4,026 BTC, and Core Scientific liquidated 1,992 BTC. MARA separately shed 15,000 BTC to clear $1 billion in debt in March 2026.
The revenue differential between Bitcoin mining and AI/HPC hosting has reached a point where continued mining-first strategies represent capital misallocation for companies with permitted power access.
Bitcoin mining economics (June 2026):
AI/HPC hosting economics:
The divergence explains why mining stocks have outperformed BTC by 70% in 2026. A basket of listed miners gained over 50% year-to-date while BTC declined approximately 17%. The market values these firms on their power portfolios and AI contract backlog, not their hashrate contribution.
According to CoinShares' Q1 2026 Bitcoin Mining Report, listed miners could derive as much as 70% of revenue from AI/HPC by year-end, up from roughly 30% at the start of 2026. The report characterized the period as potentially the "toughest moment" for pure-play Bitcoin miners.
The 16.4% hashrate decline from peak raises questions about Bitcoin's security model. The theoretical cost of a 51% attack falls proportionally with hashrate reduction — a 30–40% decrease in double-spend cost during hashrate troughs, according to KuCoin's analysis.
However, several factors mitigate immediate risk:
Absolute hashrate remains high. At 861 EH/s, the capital expenditure required for a 51% attack remains in the billions of dollars. No known actor possesses this capacity.
Geographic distribution. Non-U.S. miners (Kazakhstan, Russia, Middle East) absorbed some load as U.S. operators pivoted. No single geographic concentration dominates.
No observed disruptions. No orphaned block spikes, no chain reorganizations, no double-spend attempts detected during the hashrate decline.
Difficulty adjustment functions as designed. The 10.09% reduction restores profitability for remaining miners, creating equilibrium incentive to stay online.
The longer-term concern is structural rather than acute. If the AI pivot continues at current pace, the network may face a sustained hashrate ceiling rather than cyclical recovery — a fundamentally different security posture than Bitcoin has operated under historically.
Bitcoin's security budget problem predates the AI pivot but is now accelerated by it. Current data:
According to The Block, transaction fees hit a 12-month low in 2026, underscoring continued reliance on the diminishing block subsidy. The industry consensus threshold — fees consistently accounting for >20% of miner revenue to ensure long-term security — remains far from achieved.
The next halving (estimated April 2028) will reduce the subsidy to 1.5625 BTC. If hashprice remains at current levels and fee revenue does not materially increase, mining economics will further deteriorate, potentially accelerating the AI pivot among operators who have not yet committed capacity.
Block space utilization remains near capacity with fees at 1 sat/vB according to BTC.network's April 2026 report, suggesting demand exists but willingness to pay elevated fees does not — a structural problem that no difficulty adjustment can resolve.
The AI pivot carries implications for Bitcoin's energy narrative. According to Spark Research's 2026 energy mix analysis:
Canaan's North American self-mining fleet achieved 17.9 J/TH average efficiency in May 2026, an 11% year-over-year improvement.
As U.S. miners with grid-connected power pivot to AI, the remaining hashrate may concentrate among operators using stranded gas, curtailed renewables, and off-grid sources — potentially increasing the sustainable energy percentage further. Companies like Giga Energy continue capturing flare gas in Texas, reducing flaring from 240,000 standard cubic feet per day to near zero at individual well pads.
However, AI data centers require 99.99% uptime, which favors grid power over intermittent sources. The infrastructure being repurposed for AI will likely run on baseload grid power, while remaining mining operations may increasingly rely on interruptible, stranded, or curtailed energy — creating a bifurcation in the industry's energy profile.
The June 2026 difficulty adjustment marks a point where Bitcoin mining's identity crisis becomes measurable on-chain. The network's self-correcting difficulty mechanism functions as designed — but it is now correcting for a fundamentally different phenomenon than temporary price-driven shutdowns.
Listed miners are not turning machines off because BTC dropped. They are physically demolishing mining racks and pouring concrete for AI data centers. The $35 billion in committed contracts represents permanent capacity reallocation, not cyclical optionality.
For Bitcoin, the economic question is binary: either the fee market develops sufficient revenue to compete with AI hosting economics for power capacity, or the network's long-term hashrate ceiling will be determined by operators for whom Bitcoin mining remains the highest-value use of their specific power assets — predominantly stranded gas, curtailed renewables, and jurisdictions without AI demand.
The market has already priced this divergence. Mining company equities trade on power portfolios and AI contract backlog. Bitcoin trades on monetary premium and ETF flows. These two valuations have decoupled. Whether Bitcoin's security model can tolerate that decoupling through another halving cycle remains the open question.