January's zettahash records feel like another era. Half a year later the network is roughly flat — squeezed by thin margins, Texas summer curtailment, and miners rewiring their sites for AI. The plateau says more about economics than security.
The machines are still humming — they're just not multiplying the way they used to.
For five straight years, the only question about Bitcoin's
hashrate was how fast it would grow. In 2026 the answer, for the first time since 2020, is: it isn't. The network that crossed the zettahash line in January — briefly touching 1.1 ZH/s, a one with twenty-one zeros — has spent the summer oscillating around 900–930 EH/s, a few percent below where it started the year. A six-year growth streak is quietly ending, and the reasons are written in miners' income statements, not in the protocol.
The proximate cause is money. Hashprice — the industry's shorthand for daily revenue per petahash — has spent most of the year pinned near $30–35 per PH/s per day, close to the levels that triggered miner capitulation in February. At those rates, mid-generation hardware needs sub-5-cent power to stay cash-positive, and much of the air-cooled fleet doesn't clear the bar with bitcoin in the low $60,000s. The market's response showed up in
this month's difficulty adjustment, which fell 5% after June's 10% drop — the second-largest of the year and the eleventh-biggest downward move in the network's history.
Two structural forces are amplifying the squeeze. The first is seasonal and familiar: ERCOT's summer 4CP season began in June, giving large Texas miners a powerful incentive to shut down during projected peak-demand windows — curtailment that temporarily pulls tens of exahashes off the network from the largest mining market in North America. The second is new and probably permanent: public miners are re-fitting sites for AI and high-performance computing, a pivot that removes bitcoin hashrate even when the underlying power capacity stays fully used. Q1 marked the first first-quarter hashrate decline since 2020, and analysts attribute much of it to that capital reallocation rather than to rigs going dark.
Whether the plateau is bearish depends on what you think hashrate is for. As a security budget, 900+ EH/s remains staggering — hundreds of times the level that secured the network five years ago, with the
difficulty adjustment automatically rebalancing block production as machines come and go. Some researchers argue the shift is quietly healthy: US public miners had grown to over 40% of global hashrate, and their pivot toward AI spreads the map back out toward newer entrants in Paraguay, Ethiopia, and Oman. Concentration, not size, was always the sharper risk.
The forecasts that still call for growth all carry the same asterisk. CoinShares models roughly 1.8 ZH/s by year-end — conditional on bitcoin recovering toward $100,000, a price that would lift hashprice enough to justify a new fleet cycle. Until then, the
mining industry's discipline is doing what it always does at the bottom of a margin cycle: the least efficient machines power down, difficulty follows them, and the miners left standing earn more per hash. The engine isn't stalling. It's idling — waiting for a reason to rev.
BITCOIN ALMANACK ANALYSIS
2026: the year the line went flat
Network hashrate, monthly average estimate (EH/s)
1,000 — THE ZETTAHASH LINE
JANFEBMARAPRMAYJUNJUL
Monthly averages approximate, from public network estimates; July month-to-date · Chart: Bitcoin Almanack
WHY IT MATTERS
Hashrate is the most-cited "network health" stat in Bitcoin, and it's about to print its first flat year in recent memory. Knowing why — margins, summer curtailment, and the AI pivot — inoculates you against both the doom headlines ("security collapsing!") and the hopium ("hashrate always leads price!"). Neither survives contact with the difficulty adjustment.
What to watch next
1.August's difficulty adjustments. If hashrate holds near 915 EH/s, difficulty ticks back up and July's margin relief starts eroding — the cycle's usual rhythm.
2.Q2 earnings from public miners. American Bitcoin reports August 3. Watch the ratio of capex going to ASICs versus AI retrofits — it's the cleanest read on how much hashrate growth is being permanently diverted.
3.The end of 4CP season. Texas curtailment incentives fade after September; whatever hashrate doesn't come back then wasn't curtailed — it was retired or repurposed.