Bitcoin mining difficulty has fallen 19.9% from its November 2025 peak of approximately 156 trillion to 126.23 trillion as of the July 25, 2026 adjustment — the third-deepest drawdown since ASIC hardware became the industry standard. Network hashrate has declined roughly 12% from its late-2025 pe...
"Our thesis as a company is power fuels technology. When we first started chatting last year, we had zero contracted AI revenue. Now we have about $27 billion of AI contracted revenue." — Asher Genoot, CEO, Hut 8 (CNBC interview, June 2026)
Bitcoin mining difficulty has fallen 19.9% from its November 2025 peak of approximately 156 trillion to 126.23 trillion as of the July 25, 2026 adjustment — the third-deepest drawdown since ASIC hardware became the industry standard. Network hashrate has declined roughly 12% from its late-2025 peak above one zettahash per second to approximately 868 exahashes per second. Per-block dollar revenue has collapsed 74% year-over-year, from roughly $750,000 to $197,000, as the April 2024 halving (cutting rewards from 6.25 to 3.125 BTC) collides with a 47% decline in bitcoin's price from its October 2025 peak of $120,000 to approximately $63,100.
The industry response has been unambiguous: publicly traded miners sold 32,000 BTC in Q1 2026 alone — exceeding total 2025 sales and surpassing the 20,000 BTC liquidated during the 2022 Terra/Luna collapse. Simultaneously, mining companies have signed more than $70 billion in AI and high-performance computing (HPC) contracts, with some operators on track to derive up to 70% of revenue from AI by year-end 2026, according to CoinShares. Mining equities gained 56% in early 2026 while bitcoin fell 17% — a divergence that signals investors now value these firms as energy infrastructure plays, not bitcoin proxies.
Bitcoin mining difficulty has recorded nine downward adjustments in 2026. The June adjustment alone was -10.09%, bringing difficulty from 156.96 trillion to 124.93 trillion. The cumulative decline of 19.9% from the November 2025 all-time high of approximately 156 trillion ranks behind only two prior events: China's May 2021 mining ban (which triggered a roughly 28% drop) and the 2018 bear market.
According to Hashrate Index data tracked by Bitcoin Magazine Pro, the network has logged 287 consecutive days of downward trend — one of the longest sustained contractions in bitcoin's history. Difficulty has turned negative on a year-over-year basis for only the second time ever.
The next adjustment is estimated around August 9, 2026, with a projected further decline of approximately 2.58% based on recent average block times. This would mark the tenth negative adjustment in 2026.
Network hashrate has declined from its late-2025 peak above 1 ZH/s to approximately 868 EH/s by late July 2026. The network consumed an estimated 155–180 TWh annually as of early-to-mid 2026, with average fleet efficiency around 16 J/TH.
The current mining crisis results from two compounding forces. The April 2024 halving cut per-block rewards from 6.25 to 3.125 BTC. Transaction fees, which briefly spiked following the halving due to the Runes protocol launch, have since fallen back to low single-digit percentages of total block revenue — insufficient to offset the subsidy reduction.
Bitcoin's price compounded the damage. From an October 2025 peak near $120,000, the price declined 47% to approximately $63,100 by July 31, 2026. Combined, per-block dollar revenue fell 74% year-over-year, from roughly $750,000 to approximately $197,000.
Hashprice — the standard revenue metric measuring income per petahash per second per day — stood at approximately $32/PH/s/day in late July 2026. According to CoinShares' Q1 2026 mining report, any machine less efficient than an Antminer S19 XP running on electricity priced above $0.06/kWh is losing money at current hash price levels. The average production cost for publicly listed miners rose to approximately $79,995 per bitcoin in Q4 2025, according to CoinShares. With bitcoin trading in the $63,000–$70,000 range through mid-2026, many operators are estimated to be losing as much as $19,000 per BTC mined.
CoinShares estimates 15–20% of the global mining fleet is operating at a loss at current conditions. The Antminer S19 generation — once the industry workhorse — cannot generate positive cash flow under prevailing economics.
Publicly traded miners sold more than 32,000 BTC in Q1 2026 — a single-quarter record that exceeded their combined sales for all of 2025 and surpassed the approximately 20,000 BTC sold during the Q2 2022 Terra/Luna collapse.
Individual company data illustrates the scale:
| Company | BTC Sold (Q1 2026) | Value | Notes | |---------|-------------------|-------|-------| | Riot Platforms | 3,778 BTC | $289.5M | Produced only 1,473 BTC; sold from reserves | | Core Scientific | ~1,900 BTC | ~$175M | January sales alone | | Cango | 2,000 BTC | ~$143M | Proceeds used to retire bitcoin-backed loans | | MARA, Genius Group, Nakamoto Holdings | 15,000+ BTC combined | — | Single week in Q1 |
Riot Platforms' Q1 2026 results illustrate the pressure: bitcoin mining revenue fell 21.7% year-over-year to $111.9 million, while the all-in cost to mine a single bitcoin rose to $44,629. The company ended Q1 holding 15,679 BTC, with 5,802 coins pledged as collateral.
Aggregate miner reserves declined from 1.86 million BTC at end-2023 to approximately 1.8 million BTC by mid-2026. The drawdown is modest in absolute terms but represents a structural shift: miners are no longer accumulating bitcoin as a treasury strategy. They are liquidating to fund capital expenditures in AI infrastructure.
More than $70 billion in AI and HPC infrastructure contracts have been signed by publicly traded bitcoin miners, according to CoinShares. The deals share a common structure: long-term leases (typically 10–15 years) with hyperscalers, chip companies, and AI compute customers, converting existing or planned power capacity from bitcoin mining to AI data centers.
Hut 8 has assembled the largest contracted portfolio. The company holds $26.6 billion in total contracted AI revenue, including a second 15-year Texas lease at its Beacon Point campus covering 352 MW with a $19.6 billion base term value. CEO Asher Genoot told CNBC the company now projects approximately $1.75 billion per year in EBITDA from AI contracts. Hut 8 held 597 MW of contracted AI data center capacity as of its latest disclosure. Its shares have quadrupled over the past 12 months.
Core Scientific announced a partnership with AMD on July 28, 2026, covering 529 MW under 15-year leases with projected base contracted revenue exceeding $14 billion. The company has approximately 1.1 GW of leased AI customer capacity and projects over $24 billion in potential contracted revenue. As part of the AMD deal, Core Scientific granted AMD a warrant for up to 30 million shares, with approximately 6.5 million shares vesting immediately. AMD secured rights to reserve up to an additional 1,925 MW of capacity through late 2028. CEO Adam Sullivan called it "a truly one-of-one type partnership in this market."
TeraWulf crossed an inflection point in Q1 2026 when AI/HPC hosting revenue of $21 million surpassed bitcoin mining revenue of less than $13 million for the first time. Total quarterly revenue was $34 million. CEO Paul Prager said the quarter was "about execution," converting the company's platform of sites, contracts, capital and strategy into operating performance and recurring revenue. The company had 60 MW of critical IT capacity energized at Lake Mariner.
HIVE Digital Technologies announced a CAD 3.5 billion ($2.55 billion) AI gigafactory project in the Greater Toronto Area on May 18, 2026. The facility will cover approximately 25 acres of land (acquired for $58 million), with a 320 MW power allocation designed to host more than 100,000 GPUs. Operations are targeted for the second half of 2027. Shares rose more than 35% on the announcement.
The most telling indicator of the industry's transformation is stock performance. A basket of mining equities gained 56% in early 2026 while bitcoin declined 17%. This marks an unprecedented divergence from the historical correlation between miner stocks and bitcoin's spot price.
Year-to-date performance through late July 2026:
| Company | YTD Return | Market Cap | |---------|-----------|------------| | Riot Platforms (RIOT) | +86.6% | $8.94B | | Core Scientific (CORZ) | +66.8% | $7.72B | | MARA Holdings (MARA) | +38.5% | — |
Wall Street analysts have responded accordingly. Bernstein raised its price target for Core Scientific to $32 from $24 (Outperform) and Riot Platforms to $30 from $25 (Outperform) in mid-2026, while cutting MARA Holdings to $17 from $23 (Market Perform). The divergence among the miners themselves reflects the AI pivot's uneven adoption: MARA, which has been slower to diversify, posted an 18% revenue decline in Q1 to $174.6 million and a $1.3 billion net loss on bitcoin writedowns.
The market is pricing these companies as power and infrastructure businesses. Hut 8's $26.6 billion AI contract portfolio dwarfs its bitcoin mining revenue. Core Scientific's $24 billion-plus pipeline exceeds its total bitcoin mining revenue across its entire operating history.
The underlying economic logic of the pivot is straightforward. Bitcoin mining monetizes low-cost, often stranded electricity at thin margins that fluctuate with bitcoin's price. AI compute monetizes the same electricity at higher, more predictable margins under long-term contracts.
The current mining fleet consumes an estimated 155–180 TWh annually. At an average production cost approaching $80,000 per bitcoin and a spot price around $63,100, mining operations are collectively destroying value on a per-unit basis. AI hosting contracts, by contrast, offer contracted revenue streams spanning 10–15 years with creditworthy counterparties (AMD, hyperscalers, enterprise customers).
The transition is not without friction. Bitcoin mining facilities require substantial retrofitting for AI workloads, which demand higher power density, liquid cooling, and different network infrastructure. Hut 8's CEO has noted the company is developing new purpose-built sites for AI rather than converting existing mining facilities. Capital expenditures for AI-grade data centers run significantly higher per megawatt than mining operations.
The energy consumption profile also differs. Bitcoin mining can tolerate intermittent power and geographic isolation. AI data centers require near-100% uptime and robust network connectivity. Miners with grid-connected sites and existing utility relationships have structural advantages in the pivot.
The bitcoin mining industry is undergoing a structural transformation driven by economic necessity. The combination of halving-induced revenue compression, a sustained price decline, and the availability of higher-margin AI workloads has accelerated a capital reallocation that began as an experiment in 2024 and has now become the dominant strategy for publicly traded operators.
The data describes an industry that discovered its core asset was never the bitcoin it mined but the power infrastructure it built. Hut 8's $26.6 billion AI contract portfolio, Core Scientific's AMD partnership, and TeraWulf's revenue crossover are not speculative pivots — they represent signed, long-term agreements with large counterparties. The market has responded: mining stocks have decoupled from bitcoin's price for the first time, trading on contracted AI revenue rather than hash rate or BTC holdings.
The implications for bitcoin's network are less clear. A 19.9% difficulty decline has reduced the cost of mining and, absent further price declines, should eventually re-establish equilibrium. But the capital that has exited — redirected into 10- and 15-year AI contracts — is unlikely to return to mining even if bitcoin's price recovers. The industry's center of gravity has shifted permanently. What remains will be a leaner, more efficient mining network supported by operators for whom bitcoin is one revenue line among several, not the sole business.