Proof of Work Explained: Why Bitcoin Needs Miners

Proof of work makes rewriting Bitcoin's history cost real energy and real hardware. That physical cost is exactly why Bitcoin still needs miners.

By Michal Beno, CEO, OneMiners11 October 2026

Imagine Alice has one digital coin. On Monday morning she sends it to Bob for a bike. One second later she sends the same coin to Carol for a laptop. If a coin is just a file, nothing physical stops her: files can be copied. This is the double-spend problem, spending the same money twice.

A bank fixes this easily. It keeps one master ledger, sees both payments, accepts the first and rejects the second. Bitcoin has no bank. Thousands of independent computers, called nodes, each keep a copy of the ledger, and some hear about Bob's payment first while others hear about Carol's.

Signatures already prove Alice owns the coin. The hard part is ordering: which payment came first? A simple vote fails too, because on an open network one person can create a million fake identities for free, a Sybil attack.

Proof of work answers both problems at once. Instead of one identity, one vote, it is one unit of computing work, one vote. Work cannot be faked, because it costs real hardware and power. The chain with the most accumulated work is accepted as the true history, and whichever of Alice's payments is recorded in it first is the one that counts. Bob gets his coin. Carol's payment is simply rejected.

The idea is older than Bitcoin. In 1997 cryptographer Adam Back described Hashcash, a scheme that asked email senders to attach a small proof of computing effort. Satoshi Nakamoto's 2008 Bitcoin white paper cited it directly.

The work itself is guessing. A hash is a digital fingerprint: feed any data into the SHA-256 function and out comes a fixed-length number. Change one letter and the fingerprint changes completely. Miners keep changing a small field in the block, called the nonce, and hashing again, until the fingerprint comes out below a target number set by the network. The only way to win more often is to guess more often. Hashrate is simply guesses per second.

Every Bitcoin block contains the fingerprint of the block before it. Suppose Alice later tries to undo her payment to Bob, which sits in block 100. Change it and block 100's fingerprint changes, which breaks the link to block 101, 102 and every block since. To make the edit stick, she would have to redo the proof of work for block 100 and for every block after it, then overtake the honest network, which keeps adding new blocks the whole time.

That is why exchanges and merchants wait for confirmations. Each new block stacked on top of a payment is another layer of work an attacker would need to redo. The white paper worked out the odds: an attacker with less than half the network's power falls further behind with every block, so the chance of catching up shrinks rapidly. The chart below uses the white paper's own formula.

Six confirmations, about an hour, is the common rule of thumb for large payments. Against an attacker with 10% of the hashrate, the chance of reversing a payment after six blocks is about 0.02%. Against a bigger attacker you simply wait longer. The white paper lists how many blocks it takes to push the odds below one in a thousand.

A 51% attack is when one party controls more than half of a network's hashrate. It could reverse its own recent payments or block others from confirming. It could not take coins from other people's wallets or print bitcoin, because every node rejects invalid transactions.

So what would it take? Bitcoin runs at about 970 EH/s, which is 970 million TH/s. Here is a worked example. Divide that by the hashrate of one machine and multiply by its price. With the Antminer S21 Pro at 245 TH/s and $2,299 each, merely matching the network needs about 4.0 million units, roughly $9.1 billion of hardware, before buildings, cooling or grid connections. And to win, an attacker needs more than a tie.

Hardware just to match Bitcoin's hashrate, by model

Bitcoin miner	Units needed	Hardware	Power
Antminer S21 200T	4.85 million	$4.4B	17.1 GW
Antminer S21 Pro 245T	3.96 million	$9.1B	14.6 GW
Antminer S21 XP Hyd 473T	2.05 million	$12.7B	11.6 GW
Antminer S23 Hyd 580T	1.67 million	$20.6B	9.2 GW

OneMiners calculation at 970 EH/s and OneMiners retail prices, 5 to 6 October 2026. Matching is not enough to attack; an attacker needs more than half.

The bill keeps running after purchase. Even at a very cheap $0.048 per kWh, the S21 Pro version would burn through roughly $17 million of power every day. And a visible attack would crash confidence in the coin, and with it the value of hardware that is useful only for SHA-256 mining. For anyone holding that much hashrate, the most rewarding option is usually to mine honestly.

Small networks show what happens when that cost is low. In August 2020, Ethereum Classic, a much smaller proof-of-work chain, was hit by three 51% attacks in one month, according to The Block. The lesson is not that proof of work fails. It is that security scales with the work behind a chain, and Bitcoin has by far the most.

Proof of stake is the main alternative. Instead of spending energy, participants called validators lock up, or stake, the network's own coins as collateral, and the protocol picks who proposes the next block in proportion to that stake. Validators who cheat can lose part of their deposit, a penalty called slashing. Ethereum moved from proof of work to proof of stake on 15 September 2022 in an upgrade called the Merge, which cut its energy use by about 99.95%, according to ethereum.org.

That energy saving is real, and proof of stake works for the networks that choose it. The real difference is where security comes from. In proof of work, the cost of attacking sits outside the system: chips that must be manufactured and power that must be bought. In proof of stake, the cost is coins inside the system, and whoever holds the most stake has the most say over future blocks.

Proof of work would not stay stable if blocks simply sped up whenever more hashrate joined. Bitcoin keeps its rhythm with the difficulty adjustment: every 2,016 blocks, about two weeks, the target is reset so blocks keep arriving roughly every ten minutes. Today's difficulty is about 132.7 T, with the next change estimated at around +2.4%.

For security, difficulty matters in one specific way: there is no discount for attackers. Every block on the honest chain was produced at the difficulty of its time, so overtaking that chain means doing at least as much work again. As more hashrate joins, difficulty rises and the cost of an attack rises with it, automatically, without anyone voting on it.

Proof of work ties Bitcoin's security to things that exist in the physical world: chips, buildings and electricity. That cost cannot be printed or voted away, and anyone with a Bitcoin miner and a power connection can take part without asking permission. Miners are not overhead on top of Bitcoin. They are its security budget, paid every block with 3.125 BTC plus fees.

Energy use is the honest trade-off, and the best answer is where mining happens: where power is plentiful and otherwise underused. Many OneMiners hosting sites run on renewable power, including Norway (100% hydro), Paraguay (Itaipu hydro), Ethiopia (hydro), Canada (hydro) and Brazil (hydro).

For anyone thinking about taking part, the same physics decides profitability: hardware efficiency and power price. An Antminer S21 (200 TH/s, 17.6 J/TH, $907) earns about $7.96 a day at today's numbers before fees, and costs about $4.06 a day to run at $0.048 per kWh, the OneMiners rate at the Nigeria site.

Antminer S21 200T$907Hashrate200 TH/sEfficiency17.60 J/THPower draw3,520 WBreak-even power$0.094/kWhNet/day at 4.80c$3.90View on OneMiners

Antminer S21 Pro 245T$2,299Hashrate245 TH/sEfficiency15.00 J/THPower draw3,675 WBreak-even power$0.111/kWhNet/day at 4.80c$5.52View on OneMiners

Antminer S23 Hyd 580T$12,299Hashrate580 TH/sEfficiency9.50 J/THPower draw5,510 WBreak-even power$0.175/kWhNet/day at 4.80c$16.74View on OneMiners

Three ways into proof of work, from entry level to the most efficient hydro unit. Live OneMiners prices, 6 October 2026; net per day at $0.048 per kWh, at today's numbers, before pool fees.

Frequently asked questions

Michal Beno

CEO of OneMiners, the global Bitcoin mining hosting and hardware platform with 15 published hosting locations, 7-year fixed electricity contracts and a 7-year hardware warranty. He writes about the economics of industrial Bitcoin mining.

Informational only, not financial advice. Figures are a snapshot from public market data and news reports as of 5 October 2026 and can change quickly. Crypto prices can fall as well as rise. No return is offered or implied. Do your own research.