Bitcoin’s network hashrate has now spent 316 consecutive days beneath its all-time high, the longest such stretch in roughly a decade, based on the rolling average published by Blockchain.com.
The smoothed metric peaked near 1,151 EH/s in late October 2025 and sat around 914 EH/s at the end of August 2026—a gap of roughly 20% that has persisted despite a rising Bitcoin price and back-to-back downward difficulty adjustments.
The drought stands out because hashrate typically follows price higher within weeks, as miners bring idle machines back online. This cycle, that reflex has weakened, and public-company disclosures now point to a structural cause rather than a temporary lull tied to weather or seasonality.
The Numbers Behind the Drought
Live estimates from CoinWarz and Blockchair place current hashrate in the 910–930 EH/s range, with daily readings swinging by tens of EH/s depending on the estimation window. Difficulty—the protocol-adjusted parameter that governs mining competition—sits at 125.81 trillion following a 1.31% downward adjustment at block 963,648 on August 23, according to on-chain data from Mempool.space. That is roughly 19% below the network’s all-time difficulty high of about 156 trillion set in November 2025.
Ten of the seventeen difficulty adjustments in 2026 have moved lower, a pattern consistent with capacity leaving the network. Winter Storm Fern in February drove one of the sharpest sequences on record: an 11.16% cut on February 7 followed by a 14.7% rebound twelve days later. A further ~10% cut on June 13 followed a wave of shutdowns tied to summer power curtailments and worsening unit economics after the 2024 halving.
The drought is also metric-specific. Instantaneous readings briefly touched 1,157–1,164 EH/s in mid-October 2025, and CoinWarz’s own methodology showed a September 2025 peak nearer 1,442 EH/s. On the smoothed seven-day series most commonly used to declare hashrate records, however, no new high has been printed in nearly eleven months.
Why the Rebound Is Stalling: The AI Pivot
The clearest evidence that this is not a normal cycle comes from listed miners’ own filings. In its FY26 annual report filed with the Securities and Exchange Commission (SEC), IREN Limited confirmed installed Bitcoin mining capacity of approximately 23.2 EH/s across roughly 380 megawatts as of June 30, 2026, down from around 50 EH/s a year earlier. The company plans to substantially complete its transition of that data-center capacity to AI Cloud Services by the end of 2026.
IREN’s separate FY26 earnings release reported $4 billion in contracted annualized run-rate revenue for 2026 AI capacity, alongside multi-year contracts with a frontier AI lab and customers including Cohere, Perplexity, and Figure AI. Non-cash impairments of $638.8 million for the fiscal year, largely tied to decommissioned mining hardware, quantify what the shift looks like on the balance sheet.
Similar reallocation has been disclosed across the sector, with mining equities significantly outperforming Bitcoin in 2026 as investors reward the pivot to high-performance computing.
The broader implication is that power and data-center infrastructure leaving Bitcoin now has a competing, contractually locked-in buyer. Long-duration AI compute revenue changes the calculus of waiting for hashprice to recover; a five-year hyperscaler agreement cannot simply be paused when Bitcoin rallies.
For the network, this does not threaten security in absolute terms—900 EH/s remains an extraordinary level of computational commitment, and difficulty adjusts downward to protect surviving miner margins—but it dampens the historic feedback loop between rising prices and returning hashrate.
Whether the late-2025 peak is retaken in 2026 will depend on whether improving mining margins can outbid hyperscaler contracts for the same megawatt.
Also read: Is Bitcoin’s 4-Year Cycle Ending? Willy Woo Points to a 6-8 Year Rhythm
