Difficulty just did something it has done once before
Bitcoin network difficulty is now lower than it was a year ago. That has happened exactly twice in the history of the protocol.
At the July 25 retarget, difficulty settled at 126.23 trillion, roughly 1.1 percent below the 127.62 trillion recorded twelve months earlier. Luxor Hashrate Index confirmed the reading on July 28. It is only the second negative year-over-year difficulty print the network has ever produced.
The first was 2021, when China banned mining and roughly half the network hash rate went dark within weeks. That was a policy shock with an obvious cause and an obvious end date. Machines were unplugged in one jurisdiction, shipped, and replugged in another.
This time nobody banned anything.
The numbers
Difficulty peaked near 156 trillion in November 2025. The current reading of 126.23 trillion is 19.9 percent below that high, the third deepest drawdown since ASIC hardware replaced GPUs on the network.
Two adjustments account for most of the decline. A 10.09 percent drop at block 953,568 in June was the second largest negative adjustment of 2026, triggered by a price slide that pushed marginal machines offline. Subsequent retargets have continued to grind lower. The next adjustment is estimated for August 8 and currently projects roughly flat.
Hash rate has followed the same path, falling from records set late last year into the 900 EH/s range.
Why this drawdown is different
Luxor attributes the contraction to three overlapping forces: falling Bitcoin prices, compressed mining revenue, and the diversion of capital, power and operators toward artificial intelligence and high performance computing.
The third force is the interesting one, and it is structurally new.
For most of the history of Bitcoin mining, the alternative use for a megawatt of cheap, interruptible power sitting behind a remote substation was close to nothing. Mining was often the only bidder. That is no longer true. AI data centre demand now competes directly for the same power contracts, the same interconnection queues, and the same shells that miners spent a decade building out.
When an operator converts a site to host GPUs, that hash rate does not go offline temporarily waiting for a better price. It leaves. The economics of the conversion do not reverse because Bitcoin rallies ten percent.
What hashprice says about the next six months
The forward market is not pricing a recovery.
As of July 28, Luxor hashrate derivatives were pricing average hashprice at $31.85 per petahash per day through December 2026. That is only modestly above spot, and it sits below breakeven for a good deal of older equipment still running on the network.
Translated: the market that actually takes the other side of miner revenue risk expects the squeeze to persist through year end. Operators running older machines on unsubsidised power are not being offered a way out.
What it means for network security
The instinctive worry is security. Less hash rate means a cheaper attack.
That framing overstates the near-term risk and understates a longer question. In the short term, a network in the 900 EH/s range remains absurdly expensive to attack, and the difficulty adjustment is the mechanism working exactly as designed. It is a thermostat, not a warning light. Blocks keep arriving roughly every ten minutes because difficulty falls to meet the hash rate that remains.
The longer question is about who is left mining. If the marginal operator increasingly treats Bitcoin as a flexible load to be shed whenever AI hosting pays better, hash rate becomes more elastic to power markets and less anchored to Bitcoin price alone. That is a different security model than the one most holders assume.
It also concentrates. Operators with owned generation, long dated power contracts, or access to the newest hardware survive drawdowns that eliminate everyone else.
What to watch
The August 8 retarget currently projects roughly flat, which would suggest hash rate has stabilised near current levels rather than continuing to bleed.
Beyond that, watch conversion announcements rather than difficulty charts. Every public miner that signs an AI hosting deal removes future hash rate from the network in a way that a price rally will not bring back.
Bitcoin Gate Take
Difficulty falling year over year without a government banning anything is a genuinely new data point, and it deserves more attention than it is getting. The network is fine. The difficulty adjustment is doing precisely what it was built to do. But the composition of who secures Bitcoin is shifting toward operators who treat hashing as one revenue option among several.
That is not bearish. It is a reminder that mining is a commodity business subject to the same capital rotation as everything else, and that the security budget has always been a function of price rather than the other way around. If you want to understand how difficulty, hash rate and the security budget actually fit together, the mining section of the Bitcoin Gate course covers it end to end.