The May 1 difficulty adjustment dropped to 132.47 trillion after sustained hashrate fell below the symbolic one-zettahash threshold for the first time since February. The dip says more about how public miners are reallocating capex than about network security.
Bitcoin's network difficulty adjusted downward by 2.3 per cent at block 947,520 on May 1, the sixth reduction of 2026 in nine epochs, after sustained hashrate fell below the symbolic one-zettahash threshold for the first time since February.
Difficulty now sits at 132.47 trillion. The hashrate at the time of the adjustment was around 899 EH/s, with seven-day averages hovering between 899 and 958. Block times had been running at roughly ten minutes and twenty-eight seconds, not catastrophic but enough above the ten-minute target to trigger a meaningful cut. Hashprice, the dollar revenue a miner earns per unit of computational work, climbed from $34.39 per PH/s to $37.52, a small but consequential bump for operators running at the margin of profitability.
The story underneath the numbers is the same one mining analysts have been telling since spring. Public miners sold more than 32,000 bitcoin in the first quarter of 2026, the heaviest miner distribution since the early aftermath of the 2024 halving. Bitfarms, Iris Energy, Hut 8 and Hive — none of them small operators — have all reduced fleet capacity or paused expansion plans this year. The ones still expanding are doing so on a different thesis. MARA's $1.5 billion deal for a 505-megawatt gas plant in Ohio, announced last week, is a power-acquisition story dressed up as a mining story; the underlying logic is that secured behind-the-meter electricity has more enduring value than the marginal hash it currently produces.
That pivot is now industry-wide. Core Scientific is selling AI compute capacity. Hut 8 has split its operations between bitcoin and HPC. CleanSpark continues to publicly resist the AI shift, but its peers are voting with capex. The hashrate dip below one ZH/s is partly a reflection of that reallocation. Machines coming offline are not always being replaced, because the spreadsheet for a new ASIC fleet does not pencil at current bitcoin prices and current power costs, while the spreadsheet for an HPC build does.
Bitcoin price has not been kind. It briefly cleared $80,000 on Tuesday after the CLARITY Act stablecoin compromise lifted equity-side sentiment, then was knocked back to $79,000 on a missile report near Jask in the Strait of Hormuz. The hashprice rebound this week reflects the price recovery as much as the difficulty cut; if BTC rolls over again, miner economics tighten immediately, regardless of what difficulty does next.
Network security remains comfortable in absolute terms. Even at 899 EH/s, the cost of a sustained majority attack runs into billions of dollars per day, and the largest concentrated pools — Foundry USA and AntPool — together still account for less than half the network. The dip is interesting more for what it says about the operator base than what it implies about resilience. The marginal miner has gone from running at break-even to shutting off when prices dip in roughly six months.
The next adjustment is estimated to land on May 16 or 17, at block 949,536, with difficulty rising slightly to about 133.14 trillion. That implies the network will tighten back up modestly. The machines still on are doing more work than the average since May 1, but the move does not signal an inflection. If hashrate stays in the 900–950 EH/s range and block times normalise around ten minutes and ten seconds, miners will be looking at small, regular oscillations through the summer rather than a clean break in either direction.
Glassnode flagged in its weekly note that bitcoin spot volume has dropped below $5 billion daily, the lowest level since October 2023. Combined with falling difficulty and rising AI revenue, the picture is of a network whose security budget is increasingly being subsidised by a different industry than the one it was built to be paid for. That is not the first time difficulty has been forced lower mid-cycle, but it is the first time the marginal seller has been a public miner with a clear non-bitcoin alternative on the same site.
The hashrate will move back above one zettahash. The question is what proportion of the machines doing the work in six months are owned by companies whose primary business is still mining — and what proportion belong to operators using bitcoin's security budget as a hedge against the AI capex they actually wanted to spend on.
