Hash rate

Nobody actually counts hashes. The figure is inferred from how quickly blocks arrive relative to the published difficulty, and it is quoted in terahashes (a trillion per second) or exahashes (a million terahashes), with the whole network running on the order of a thousand exahashes. A high hash rate says an attack would be expensive; it says nothing about price.

How it works

Hash rate is an estimate rather than a meter reading, and knowing how the estimate is built explains why the charts jump around.

Only two ingredients are available to anyone watching: the current difficulty, which the protocol publishes, and the timestamps of the blocks that have been found. Multiply difficulty by 4,294,967,296 to get the average hashes needed per block, then divide by the average seconds between blocks. Because block discovery is a random process, short windows are noisy. A lucky hour can imply a fifty percent surge that never happened, which is why every chart worth reading uses a seven day average.

Units climb in thousands. A terahash is a trillion hashes per second, a petahash is a thousand terahashes, an exahash is a thousand petahashes, and a zettahash is a thousand exahashes. One current machine produces roughly 200 terahashes per second, so the seven day network average that first touched one zettahash on September 2, 2025 is the output of about five million such machines running at once.

The number moves with money and with geography. It collapsed by roughly half between May and July 2021, when China expelled its mining industry and the machines went into shipping containers, and it climbed back over the following months as those containers were plugged in across the United States, Kazakhstan, and Paraguay. It falls when revenue per unit of work makes older hardware pointless and rises when new fleets are energized.

Why this matters when you buy bitcoin

Hash rate matters to a buyer in exactly one way, as the cost of attacking the ledger your coins are recorded in, and it is routinely sold to you as something else entirely.

Start with the security reading. Rewriting recent history means out hashing everyone else, and at the roughly 850 exahashes per second the network has been running lately, matching the honest chain would take something like 4.3 million machines of 200 terahashes each, drawing close to 15 gigawatts. Cambridge independently estimates network demand at 16.09 gigawatts, so the two figures agree. That is why serious discussion of a 51 percent attack involves nation states rather than hackers, and why your coins are far safer from reversal than they are from a phishing page.

The sales pitch is where it goes wrong. Cloud mining and hash rate rental contracts market a slice of that same number as passive income: pay up front, receive daily payouts, own no hardware. The arithmetic rarely survives contact with a spreadsheet, because the operator is selling you the revenue a real miner earns while keeping the machine, the power contract, and your deposit. Guaranteed daily returns on rented hash rate have been one of the most durable scam formats in this industry.

The third misuse is the price argument. Hash rate at a record high gets reported as a bullish signal, but hash rate follows revenue with a lag of many months, because machines are ordered when mining looks profitable and arrive long after that mood has passed. A new high is a picture of what miners believed last year.

Turning difficulty into a hash rate estimate

The hash rate estimate is arithmetic anyone can reproduce, and doing it once removes the mystery from every hash rate chart you will ever see.

Take the difficulty the network is currently enforcing. A block needs, on average, difficulty times 4,294,967,296 attempts, since that many tries is what a difficulty of 1 demands. At a difficulty of 100 trillion that is 429 billion trillion attempts per block. Divide by the 600 seconds a block is supposed to take and the answer is 716 exahashes per second. Run the same sum with the difficulty above 120 trillion the network has carried recently and it lands near 860 exahashes per second, which is what the public trackers report. If real blocks are arriving faster than 600 seconds apart, true hash rate is above the estimate, and the next difficulty retarget will confirm it.

Hash rate vs mining difficulty

Hash rate is what miners supply; difficulty is how the protocol answers back. Hash rate can double overnight when a large fleet is switched on, and difficulty will not notice until the next retarget, up to two weeks later. In the meantime blocks simply arrive faster than ten minutes. Over months the two track each other closely, which is exactly why difficulty can be used to estimate hash rate, but on any given day one is the cause and the other is a lagging correction.

Hash rate vs hashprice

Hash rate is a quantity of work; hashprice is what that work earns. Hash rate is counted in hashes per second and reveals nothing about revenue. Hashprice is quoted in dollars per petahash per day and is a miner's income statement compressed into one number. The two usually move in opposite directions, because when hash rate climbs the same block rewards are divided among more machines. A record hash rate is generally bad news for miners, which is close to the opposite of how it tends to be reported.