Bitcoin's network hash rate passed 700 exahashes per second so long ago that the threshold no longer counts as news. As of mid-July 2026, the network is producing somewhere between 880 and 910 EH/s depending on whose estimator you trust, after spending much of the past 18 months at or above 1 zettahash per second — a thousand exahashes. The story worth telling is not a surge past an old milestone. It is that hash power is coming off the network, and that miners are increasingly choosing to let it go.
The July 11 difficulty cut
Bitcoin's difficulty adjustment is the network's automatic thermostat: every 2,016 blocks, the protocol retunes how hard it is to mine so blocks keep arriving roughly every ten minutes. When hash power leaves, blocks slow down, and the next adjustment makes mining easier.
That is what happened this month. According to Bitcoin.com News, difficulty fell 5% at block 957,600 on July 11, dropping from 133.87 trillion to 127.17 trillion — a decline of 6.70 trillion in a single reset. The epoch that triggered it ran 14 days, 18 hours and 9 minutes, with an average block time of 10 minutes and 32 seconds, about 5.1% slower than the protocol's target.
The same report put the seven-day average hash rate at roughly 908 EH/s on July 11, down from about 986 EH/s on July 1 — a 7.9% slide in ten days. It also noted that eight of the 14 difficulty adjustments so far in 2026 have been negative, with difficulty down roughly 14.22% on a compounded basis since January 8.
Why the hash rate numbers disagree
Readers who check two mining trackers and get two different numbers are not misreading anything. Hash rate is never measured directly — it is inferred from how quickly blocks are found, which is a statistically noisy process. Different smoothing windows produce different answers.
Hashrate Index's July 13 roundup pegged the seven-day simple moving average at 879 EH/s, down 6.4% from 939 EH/s the week before. Bitcoin.com News reported 908 EH/s for a similar period. Both describe the same trend — a mid-single-digit weekly decline — but land about 30 EH/s apart. Treat any single hash rate figure as an estimate with error bars, and watch direction rather than the decimal place.
Hashprice is the number that actually matters
For miners, raw hash rate is an input cost, not a scoreboard. The metric that determines whether a machine earns its keep is hashprice: the daily revenue a miner earns per petahash per second of capacity.
Hashrate Index put USD hashprice at $30.88 per PH/s per day on July 13. Bitcoin.com News reported $31.10 on July 11, against a one-year peak of $49.40 in October 2025 — meaning revenue per unit of computing power has fallen by roughly a third from its recent high, and is down 16.4% since the start of 2026. Hashprice touched about $27.20 in early June, its recent floor.
Transaction fees are not filling the gap. Miners collected an average of 0.0200 BTC per block per day in fees over the week to July 13, down 18% from 0.0244 BTC the prior week. With the block subsidy fixed until the next halving, thin fee revenue leaves miners almost entirely exposed to bitcoin's price and to their own power costs.
Efficiency now separates survivors from casualties
Hashrate Index's breakdown of operating margins by fleet efficiency shows how wide the gap has become:
- Under 14 J/TH: roughly $109 per megawatt-hour of operating margin
- 14–19 J/TH: about $79 per MWh
- 19–25 J/TH: about $59 per MWh
- 25–38 J/TH: about $41 per MWh
A modern fleet earns more than twice per megawatt-hour what an aging one does. For operators running older hardware on contracted power, that spread is the difference between a business and a slow liquidation — and it explains why curtailing machines during price weakness is now a rational choice rather than a distress signal.
The AI pivot behind the retreat
The July decline is not an isolated weather event. It fits a pattern that became visible earlier in the year, when CoinDesk reported Bitcoin's first first-quarter hash rate decline since 2020, ending five consecutive years of double-digit growth. The network was down about 4% year-to-date at the time, hovering around 1 ZH/s.
The arithmetic behind that reversal was brutal. CoinDesk cited production costs near $90,000 per bitcoin against a spot price closer to $67,000 — negative margins on newly mined coins. The same megawatt of interconnected, energized capacity can instead be sold to AI and high-performance computing tenants at rates mining cannot match, backed by multi-year contracts rather than probabilistic block rewards.
That matters more than usual because of who controls the machines. Publicly listed U.S. miners have accounted for over 40% of global hash rate, and those are precisely the companies with the balance sheets, the shareholders and the grid interconnects to redirect capacity toward AI workloads. When they reallocate, the network notices.
What to watch next
Hashrate Index's forward estimate projects a +2.74% difficulty adjustment on July 25, 2026, which would suggest hash rate stabilizing or modestly recovering after the July slide. In BTC terms, hashprice actually rose 5.0% over the week to July 13, reaching 0.00049723 BTC per PH/s per day — the difficulty cut improved miner economics in coin terms even as dollar revenue stayed soft.
Longer term, CoinShares has projected hash rate reaching around 1.8 ZH/s by the end of 2026 — but explicitly conditioned that forecast on bitcoin recovering toward $100,000. That conditionality is the whole point. Hash rate is not an independent measure of network health; it is a lagging function of miner profitability, and profitability is a function of price, power costs and what else that power could be doing.
Some perspective is still warranted. Even at 880 EH/s, the network commands roughly nine times the computing power it had five years ago, when it ran near 100 EH/s. A 7.9% pullback does not meaningfully change Bitcoin's security assumptions — the cost of mounting a majority attack remains far beyond the reach of any plausible adversary. What has changed is the direction of the trend, and the reason behind it. For the first time in Bitcoin's history, the marginal miner is not being pushed out by competition from other miners. They are being outbid for electricity by an entirely different industry.