What Is Hashrate?

On Christmas Eve 2022, Bitcoin looked as if it had lost more than a third of its computing power in three days. Glassnode's estimate of the network's hashrate fell from 252.98 exahashes per second (EH/s) on 21 December to 156.36 EH/s on 24 December, as Winter Storm Elliott swept across the United States and large miners in Texas and elsewhere powered down. By the following evening a second tracker, IntoTheBlock, had it back at 249.02 EH/s.

The machines had not gone anywhere, and nobody had counted a single hash either. The figure that collapsed and recovered was an inference drawn from how quickly blocks were arriving, and that is the most useful thing to know about hashrate before reading any chart of it. Educational guide, not financial advice.

Hashrate is the combined number of guesses per second that a proof-of-work network's miners make at the puzzle deciding who adds the next block. The guessing itself is covered in the mining explainer. This piece is about the number.

From one hash to a zettahash

One hash is one attempt: a candidate block's header put through Bitcoin's SHA-256 hashing and checked against the target. One attempt a second is 1 H/s. The prefixes then climb in steps of a thousand, through kilo, mega, giga, tera and peta to exa and zetta.

Do the arithmetic once and the scale lands. An exahash per second is a billion billion hashes, a 1 followed by 18 zeros. A zettahash per second (ZH/s) is 1,000 EH/s, which is a 1 followed by 21 zeros. A network running at 1.05 ZH/s is therefore trying about 1,050,000,000,000,000,000,000 hashes every second, and on average one of them, somewhere, wins every ten minutes.

Nobody counts the hashes

A winning block carries no tally of the failed attempts behind it, and no miner files a report, so no tracker ever sees inside a warehouse in Texas or anywhere else. What anyone can see is two numbers: the difficulty, which sets how many hashes a block takes on average, and the timestamps showing when blocks actually arrived.

Hashrate is worked backwards from those. The Bitcoin wiki's rule of thumb multiplies the difficulty by 4,294,967,296 (2 to the power of 32) and divides by the seconds per block. Plug in the difficulty on 2 November 2026, about 148.36 trillion, with the 600-second target, and out comes roughly 1,062 EH/s: quicker blocks push the estimate up, slower ones drag it down.

Block timing is random, and that is the snag. A lucky run of quick blocks looks like a fleet of new machines plugging in, while a slow patch looks like an exodus that never happened, so every tracker averages over a window, and the window it picks changes the answer.

At about 13:40 Singapore time on 2 November 2026, one site was giving several answers at once. mempool.space's one-week estimate, built from the last 1,008 blocks, read about 1,046 EH/s. Its three-day figure said 1,072 and its last-24-hours figure said 957, while its daily readings for the ten days to 1 November had swung between 964 and 1,228. Bitcoin Core's getnetworkhashps command uses the last 120 blocks by default, roughly 20 hours' worth. Two sites quoting different hashrates on the same morning are usually just averaging different stretches of time.

The two-week correction

Bitcoin's answer to all this wobble is the difficulty adjustment. Every 2,016 blocks, each node compares the time those blocks took with the ideal of 1,209,600 seconds, exactly two weeks at ten minutes a block, and resets the difficulty in proportion. Blocks that took 12 days push it up by about a sixth, while 16 days brings it down by an eighth.

There is a ceiling too. One adjustment can at most quadruple the difficulty or cut it to a quarter, which Bitcoin's developer documentation describes as a rise of up to 300% or a fall of up to 75%. No committee meets and nobody votes, because the rule runs in every node's software.

Take the latest cycle. The adjustment at the end of October 2026 raised difficulty by about 5.5%, and blocks since then have been running slightly slower than ten minutes. Read on 2 November, mempool.space projected a cut of about 1.8% at block 975,744, due around 13 November. Slow blocks lead to an easier target, and the ten-minute pace comes back.

Milestones on the climb

mempool.space keeps a daily estimate stretching back to January 2009, and the record shows how steep the climb has been. Its one-day figure first passed 1 EH/s in January 2016 and 100 EH/s in September 2019. Reaching 500 EH/s took four more years, until September 2023.

The first day above 1 ZH/s came on 4 April 2025. The seven-day average of the same series only cleared that line at the end of August 2025, almost five months later, which says a lot about how jumpy single days are.

What it says about security, and what it does not

Picture someone trying to undo a payment they made an hour ago. On Bitcoin they would have to mine a longer chain in secret, faster than everyone else combined, and keep going until their version overtook the public one. Satoshi Nakamoto's 2008 white paper rests the design on that race: as long as honest machines hold more computing power than any group cooperating to attack the network, the honest chain stays ahead. At around 1 ZH/s, the attacker's bill is more machines than the rest of the network put together, plus the power to run them for as long as the attack lasts.

So hashrate prices one kind of attack, and nothing else. A wallet whose seed phrase was phished last week loses its coins just as completely at 1 ZH/s as at 1 EH/s, and the number tells you nothing about the price of bitcoin.

The total also hides who directs it. mempool.space attributed 270 of the 1,022 blocks found in the week to 2 November 2026 to the Foundry USA pool and 205 to AntPool, which is 46.5% from two pools. A pool only coordinates the work, and miners can point their hardware at a different pool overnight. That share still belongs next to the headline figure whenever someone quotes it.

Chains with no hashrate at all

Hashrate exists only on proof-of-work chains such as Bitcoin. Ethereum had one until the Merge on 15 September 2022, when it switched to proof of stake and mining stopped producing its blocks. Validators who lock up ETH as collateral produce them now, and the network is secured by that stake.

Since then there has been nothing on Ethereum to count or estimate. Putting its security next to Bitcoin's in hashes compares something with nothing.