What Is Hashrate? How Bitcoin's Difficulty and Hash Price Really Work

Bitcoin's hashrate is the most quoted number in mining and the least examined. It gets reported like a scoreboard: up is good, down is bad, records are bullish. The reality is stranger and more interesting. Nobody measures hashrate directly. It is a statistical inference drawn from how fast blocks are arriving and how hard the network has made them to find, which means every figure you see is an estimate with an error bar, and the number that the protocol actually enforces is a different one entirely.

That distinction matters whether you are a miner judging an equipment purchase, an investor reading a headline, or simply someone trying to understand what secures the coins you own. This is a guide to what hashrate is, how bitcoin difficulty works, what hash price means for the people doing the work, and what the 2026 numbers say about the network.

What hashrate actually measures

A miner repeatedly feeds a block header through SHA-256, changing one field each time, hoping to land on an output below a target set by the protocol. Hashrate is the speed of that guessing: hashes per second, aggregated across every machine on the network. Right now the whole network is producing somewhere around 1.08 zettahashes per second, or 1,080 exahashes, which is roughly 1.08 sextillion guesses every second.

You cannot count those guesses. Miners do not report them and there is no register of machines. What you can count is blocks. If the network finds blocks faster than the ten-minute target, more computing power is almost certainly at work, so the standard estimate divides the work a block is expected to require by the time it took. In practice that is difficulty multiplied by 2 to the power of 32, divided by the average block interval. Plug in today's difficulty of 140.58 trillion and an average interval of about 564 seconds and you get roughly 1.07 zettahashes per second, which is where the published figure comes from.

The consequence is that short-window hashrate numbers are noisy. A single lucky miner can make a one-hour reading jump by double digits. Serious dashboards use seven-day rolling averages for that reason, and different providers still disagree by a few percent because they weight the calculation differently. Treat the figure as a good estimate, not an instrument reading.

Difficulty is the number the protocol actually enforces

Difficulty is the part that is real. It is a consensus rule, written into every node's validation logic, that raises or lowers the target a block hash must beat. When difficulty rises, the same hardware finds fewer blocks. When it falls, the same hardware finds more.

Bitcoin adjusts it every 2,016 blocks, roughly two weeks, by comparing the time those blocks actually took against the 20,160 minutes they were supposed to take. If the network ran 10% fast, difficulty rises about 10%. One detail that rarely makes it into explainers: Bitcoin Core clamps any single adjustment to a factor of four in either direction, so a genuine hashrate collapse cannot produce an absurd one-step correction. The clamp exists as a safety rail against edge cases in the difficulty calculation, and in practice it almost never binds.

For a miner, difficulty is the cost of admission and the reason mining is a treadmill. Every machine on the network is competing for the same 450 or so new bitcoin a day, and difficulty rises to absorb whatever hardware arrives. The block subsidy halves every 210,000 blocks, as our guide to the halving explains, while difficulty keeps climbing between halvings. The gap between those two curves is mining economics in one sentence.

Reading the units: megahash to zettahash

Hashrate is denominated in powers of a thousand: kilohash, megahash, gigahash, terahash, petahash, exahash, zettahash. The order matters because the numbers become meaningless fast. At difficulty 1, the network was doing about 7 million hashes a second; that is what the software implies at the very first blocks in January 2009. Today's 1.08 zettahashes is around 150 trillion times that figure, and most of the increase has come in the last decade.

The unit ladder also explains a common confusion. A modern ASIC is rated in terahashes: a machine advertising 200 TH/s is about one five-millionth of today's network. That is not a reason to avoid mining, because nobody mines alone; it is a reason to understand that individual hardware is priced against network difficulty, and that a machine bought at a good price can become uneconomic within a year if difficulty keeps rising faster than the bitcoin price.

What the numbers say this week

As of 28 October 2026, the network's seven-day average has just climbed back above 1 zettahash per second for the first time since 3 March, according to BlocksBridge Consulting's miner newsletter. Difficulty sits at a record 140.58 trillion after rising 5.92% at the 16 October adjustment, one of the larger moves of the year, and the current period is tracking toward another increase of roughly 6.5% with about 250 blocks left, because blocks have been arriving in 564 seconds on average instead of 600.

That combination is worth sitting with. Difficulty is at an all-time high, the next adjustment is likely to push it higher, and the price of bitcoin is around $80,500, well below its 2025 peak. Miners are therefore being paid less per coin while being asked to do more work per coin. The 2026 hashrate chart is the story of that squeeze: a peak somewhere between 1.3 and 1.44 zettahashes late last year, a slide through the spring, a trough near 726 exahashes in July, and a recovery back above a zettahash now.

Hash price: what a petahash actually earns

Hashrate tells you how much work is happening. Hash price tells you what that work pays. It is the daily revenue attributable to a unit of computing power, usually quoted per petahash per second, and it folds together the block subsidy, transaction fees and the bitcoin price, divided by the total hashrate competing for them.

The arithmetic is not complicated. At $80,500 a coin, the 450 BTC issued each day is worth about $36.2 million. Spread across roughly 1.08 million petahashes of network capacity, that is about $33.50 per petahash per day before fees, and fees are close to negligible at the moment. Industry indices quote a slightly higher figure, around $36, because they use their own hashrate and fee inputs; hashrateindex.com had it at $36.04 on 22 October, down from $40.42 on 25 September.

Hash price is the number that decides whether a machine runs. An ASIC at 15 joules per terahash consumes 15 kilowatts per petahash, or 360 kilowatt-hours a day. At $33.50 of revenue per petahash per day, a miner paying five cents a kilowatt-hour spends $18 on power and keeps about $15.50 before hardware, staff and hosting. At ten cents, the same machine spends $36 and is losing money. That is the entire industry in two lines, and it explains why miners obsess over power contracts and why older hardware gets switched off first when hash price falls.

Why hashrate fell in 2026, and why that is not a red flag

The obvious explanation for a falling hashrate is that mining became unprofitable and machines were switched off. That happened, and the post-halving economics of 2024 and 2025 made it worse. But the bigger 2026 story is a reallocation of capital rather than a retreat from Bitcoin.

Publicly traded miners have been converting power contracts and data-centre shells into capacity for artificial intelligence and high-performance computing, where contracts are longer and less exposed to a volatile hash price. BlocksBridge reported a 23% decline in the realised hashrate of publicly traded miners alongside a 52% quarter-on-quarter increase in their AI and HPC revenue, and estimated the displaced capacity at roughly $1.5 billion of mining hardware. Some of that equipment is being sold secondhand to smaller operators with cheaper power, which is why the network hashrate recovered even as listed miners shrank.

None of this is a verdict on Bitcoin's security, and it should not be read as one. The interesting thing about proof-of-work is that it is permissionless: capacity leaving one set of operators is capacity available to another, and the difference between proof of work and proof of stake is precisely that security here is bought with energy in the open market rather than assigned to a fixed validator set.

Concentration matters as much as size

A single aggregate number hides who is producing the blocks. Over the last 1,080 blocks, roughly one week, Foundry USA produced 26.6%, AntPool 20.4% and F2Pool 17.1% of blocks, according to mempool.space data. Three pools therefore accounted for about 64% of new blocks, with ViaBTC at 8.2%, Luxor at 5.4% and a long tail of smaller pools splitting the rest.

Pool share is not the same as control. A pool is a coordinator: individual miners point their machines at it and can redirect them, and most large pools do not custody the hashrate. But the concentration is real, it has grown over the past several years, and it is the most plausible route to the kind of coordination that hashrate alone cannot measure. It is worth watching alongside the headline figure.

What it would cost to attack Bitcoin

Hashrate is usually introduced as a security metric, so it is fair to ask what it buys. A 51% attack requires enough computing power to outrun the honest chain long enough to rewrite recent blocks, which in practice means acquiring roughly half the network's capacity from scratch. There is no market where you can buy 540 exahashes, which is the point, but you can estimate the bill.

Assume 15 joules per terahash and $18 per terahash for current-generation hardware, both reasonable 2026 figures. Half of 1.08 zettahashes is 540 exahashes, or 540 million terahashes, which implies about $9.7 billion of equipment and 8.1 gigawatts of continuous draw. Running that for a day at five cents a kilowatt-hour costs about $9.7 million in electricity alone, before the cost of facilities, cooling or the fact that such a purchase would take months and move the market. The estimate is rough and the assumptions are contestable, but the order of magnitude is the point: rewriting Bitcoin's recent history is a capital project, not a hack.

Notice what the attack does not do. It does not let an attacker steal coins from wallets or forge signatures. It lets them double-spend their own coins and censor transactions for as long as they hold the majority, which is why exchange confirmation policies matter more than most users realise. You can watch confirmations accumulate on any block explorer.

The energy question, briefly

Hashrate is a proxy for electricity consumption, which is why the two debates are always tangled. Cambridge's Centre for Alternative Finance, in its 2025 industry report, put Bitcoin's annual electricity use at roughly 138 terawatt-hours, about half a percent of global consumption, with a sustainable mix of 52.4% from renewables and nuclear. At today's hashrate those totals have most likely grown, and any honest estimate should say so rather than quote a stale figure. The arithmetic is straightforward: 1.08 zettahashes at 15 to 20 joules per terahash implies somewhere between 16 and 22 gigawatts of continuous load.

How to check the numbers yourself

The useful sources are few and free. Mempool.space shows current hashrate, difficulty and the estimated next adjustment. Blockchain explorers show individual blocks, their timestamps and which pool found them. Difficulty history is public consensus data, so nobody can shade it. What you should be sceptical of is any single-day hashrate reading presented as a trend, any all-time-high claim without a stated averaging window, and any argument that treats hashrate as a price forecast.

The bottom line

Hashrate is a measure of how much capital and energy is currently committed to finding Bitcoin blocks, inferred rather than observed, and useful mainly as a trend. Difficulty is the enforced version of the same fact, adjusted every two weeks so that blocks keep arriving every ten minutes whatever happens to the hardware. Hash price is what that work pays, and it is falling even as difficulty sets records, which is why miners are selling machines and renting out data centres.

None of that predicts the price. It does tell you something more durable: the network's security budget is large, it is expensive to attack, and it is now contested by an industry with a credible alternative use for its electricity. If you want the underlying mechanics of block discovery, how Bitcoin mining works covers the hardware and the mempool side; this page is about the scoreboard.