Bitcoin's mining industry is undergoing a structural transformation. Hashrate dropped approximately 145 EH/s between late May and early June 2026, triggering a 10.09% difficulty adjustment on June 14 — the second-largest downward retarget of the year. Hash price fell to approximately $28.26/PH/s/...
"We are no longer a Bitcoin company." — Ben Gagnon, CEO, Bitfarms
Bitcoin's mining industry is undergoing a structural transformation. Hashrate dropped approximately 145 EH/s between late May and early June 2026, triggering a 10.09% difficulty adjustment on June 14 — the second-largest downward retarget of the year. Hash price fell to approximately $28.26/PH/s/day, near a five-year low, before recovering to $32.31/PH/s after the retarget.
The proximate cause is economic: with average production costs near $80,000 per bitcoin (CoinShares Q1 2026 data) and spot prices around $61,800, an estimated 15–20% of global hashrate operates at a loss. But the structural cause runs deeper. Public mining companies have announced over $70 billion in cumulative AI and high-performance computing (HPC) contracts, redirecting power capacity, capital expenditure, and management attention away from bitcoin mining. Industry projections from CoinShares suggest listed miners could derive up to 70% of revenue from AI by end of 2026, up from approximately 30% at the start of the year.
The result is a two-track mining industry: one side mines bitcoin at diminishing margins, the other leases GPU infrastructure to hyperscalers. Both share physical facilities. The question is whether bitcoin's network security can sustain the capital reallocation.
Bitcoin's network hashrate peaked near 1,030 EH/s in late May 2026 before shedding approximately 145 EH/s over a two-week span — the steepest drawdown of the current cycle. By early June, hashrate had fallen to approximately 885 EH/s, according to CoinWarz data. This triggered a 10.09% downward difficulty adjustment at block 953,568 on June 14, reducing difficulty from 138.96 trillion to 124.93 trillion.
The decline followed a record difficulty peak of 144.4 trillion reached in February 2026, representing a 15% increase that was the largest percentage jump since 2021. From that February peak to the June trough, difficulty fell 13.5%.
Hash price — the dollar revenue per petahash per second per day — slid approximately 27% in 30 days to $28.26/PH/s/day before the retarget. CoinShares characterized this level as a five-year low, noting it had pushed 15–20% of legacy mining rigs into unprofitable territory. Post-retarget, hash price recovered to $32.31/PH/s as the lower difficulty made remaining machines slightly more productive.
The next difficulty adjustment, estimated for June 27, is projected to increase difficulty from 124.93T to approximately 133.38T, suggesting hashrate is already beginning to recover as marginal miners return.
Bitcoin's hashrate posted its first Q1 decline since 2020, breaking five consecutive years of double-digit growth. CoinShares projects hashrate could reach 1.8 ZH/s by end of 2026, but only if bitcoin returns above $100,000 to make mining economically viable across a broader operator base.
According to CoinShares' Q1 2026 Mining Report, the weighted average cash cost to produce one bitcoin among publicly listed miners rose to approximately $79,995 in Q4 2025, with costs continuing to climb through Q1 2026 as difficulty remained elevated. JPMorgan separately estimated that bitcoin had traded roughly 19% below production cost for five consecutive months as of their analysis period.
The cost-price squeeze originates from three converging pressures:
May 2026 miner revenue totaled approximately $1.12 billion, down 26% year-over-year. Bitcoin mining ROI timelines have stretched to approximately 1,000 days for new hardware deployments, according to CCN analysis, rendering fresh capital expenditure on mining equipment difficult to justify against AI infrastructure alternatives.
Public bitcoin miners have collectively announced over $70 billion in cumulative AI and HPC contracts, according to CoinShares' tracking. The scale of these commitments signals a permanent capital reallocation, not a temporary hedge.
The largest disclosed contracts:
| Company | Contract | Value | Duration | |---------|----------|-------|----------| | TeraWulf | HPC revenue pipeline | $12.8B | Multi-year | | Core Scientific | CoreWeave expansion | $10.2B | 12 years | | Hut 8 | River Bend AI lease | $7.0B | 15 years | | IREN | Microsoft partnership | $1.94B/yr annualized | 5 years |
These figures represent contracted revenue commitments, not speculative projections. CoreWeave's expanded deal with Core Scientific alone exceeds the entire public mining sector's annual bitcoin revenue by a factor of roughly nine.
The economic logic is straightforward. IREN's Microsoft contract carries an 85% project-level EBITDA margin, according to the company's disclosures. Bitcoin mining margins for publicly listed operators have been negative at spot prices for months. A megawatt of power capacity dedicated to AI GPU hosting generates materially more revenue per unit of energy than a megawatt of ASIC mining at current bitcoin prices.
Adam Sullivan, CEO of Core Scientific, framed the shift directly: "The ability for miners to convert to AI is one of the largest infrastructure shifts of this decade."
The AI revenue transition is at different stages across major public miners:
| Company | AI/HPC Revenue Share | BTC Mining Share | Trend | |---------|---------------------|------------------|-------| | Core Scientific | 39% | 61% | AI growing | | TeraWulf | 27% | 73% | AI growing | | IREN | 9% | 91% | Scaling rapidly (200MW under construction) | | Marathon (MARA) | <5% (est.) | >95% | Transitioning |
Core Scientific's digital asset self-mining revenue fell to $30.1 million in Q1 2026, down from $67.2 million in Q1 2025 — a 55% decline driven by a 45% decrease in bitcoin mined as the company strategically shifted capacity to colocation services.
The revenue split matters because it reveals a sector-wide trend: the companies with the most infrastructure are systematically redirecting it away from bitcoin mining. They are not exiting the industry entirely — most maintain some hashrate — but the marginal dollar of capital expenditure is going to GPU racks, not ASIC deployments.
Marathon Digital Holdings (MARA) illustrates the volatility of miner treasury strategy in a bear market. In Q1 2026, MARA reported:
Then, on June 16, 2026, Marathon bought 1,000 BTC — weeks after liquidating over 20,000 BTC in Q1.
MARA is simultaneously planning a $1.5 billion acquisition of the Long Ridge Energy & Power campus in Ohio for AI infrastructure, signaling its own pivot toward HPC. The company's trajectory — selling bitcoin to service debt, buying bitcoin when prices dip, while simultaneously investing in AI data centers — reflects the contradictions embedded in the transition.
The miners' AI pivot raises questions about Bitcoin's long-term security model.
Hashrate concentration. Publicly listed U.S. miners have historically accounted for over 40% of global hashrate. As these operators redirect capacity, the remaining hash is increasingly concentrated among a smaller number of participants. Pool-level data shows the top two mining pools controlled 55% of global hashrate by December 2023, with the top six at 90%. Conditions have not materially improved.
ASIC manufacturing concentration. Three firms — Bitmain, MicroBT, and Canaan — dominate ASIC production, creating a supply-side bottleneck that limits new entrants' ability to replace departing hashrate.
Attack cost versus hashrate level. The theoretical cost to execute a 51% attack remains astronomically high at current hashrate levels. However, the rate at which hashrate is declining — not its absolute level — is the relevant security signal. A sustained 15–20% drawdown from peak, if not reversed, would represent a meaningful reduction in the economic barrier to attack.
Mitigating factor: difficulty adjustment. Bitcoin's built-in difficulty retarget mechanism serves as an automatic stabilizer. When hashrate drops, difficulty falls, making mining more profitable for remaining operators and incentivizing hashrate to return. The June 14 adjustment demonstrated this: hashrate began recovering within days of the retarget.
The network is not at imminent risk. But the dependency on a difficulty mechanism designed for organic hashrate fluctuations — not for a coordinated industrial exit from mining — warrants monitoring.
One unintended consequence of the U.S. public miner pivot may be geographic decentralization of mining. As American operators redirect power capacity to AI, hashrate is migrating to jurisdictions with cheaper energy and fewer AI infrastructure opportunities.
According to multiple industry analyses, Paraguay, Ethiopia, and Oman are each entering the top-10 hashrate rankings. These countries offer sub-$0.03/kWh electricity — often from stranded hydroelectric or natural gas resources — where bitcoin mining remains profitable at current prices.
This redistribution partially offsets centralization concerns. If U.S. public miners reduce their share of global hashrate from 40%+ to 25–30%, while new capacity comes online across multiple developing economies, the net effect on geographic distribution may be neutral or positive. However, regulatory and operational risks in emerging mining jurisdictions differ substantially from those in established markets.
The bitcoin mining industry is bifurcating. On one side, public companies with access to capital markets, power infrastructure, and hyperscaler relationships are converting their facilities into AI data centers. On the other, private operators in low-cost-energy jurisdictions continue to mine bitcoin at margins that public companies cannot match.
The economic logic of the AI pivot is difficult to dispute. An 85% EBITDA margin on a five-year Microsoft contract (IREN) versus negative margins on bitcoin mining at $61,800 is not a close call. The $70 billion in announced AI contracts dwarfs the entire mining sector's annual bitcoin revenue.
What remains uncertain is the equilibrium. Bitcoin's difficulty adjustment mechanism will rebalance the network toward profitability for remaining miners, but it cannot replace the absolute security provided by higher aggregate hashrate. If bitcoin's price recovers above production cost — roughly $80,000 or higher — some capacity will return. If it does not, the mining industry's structural transformation into an AI infrastructure sector will accelerate.
The market is pricing this transformation in real time. The question is not whether bitcoin miners will diversify into AI. They already have. The question is how much hashrate remains when the reallocation is complete.