How bitcoin mining works: from hash to block reward

Bitcoin mining is the process by which specialized computers (ASICs) assemble a new block from pending transactions and hash its header billions of times per second until the result falls below a target number. The first miner to find such a hash adds the block to the chain and collects the reward: newly created coins (3.125 BTC per block since April 2024) plus every transaction fee in that block. That is how bitcoin mining works, and it has worked this way since 2009: the same process issues new coins and makes the transaction history expensive to rewrite. On November 1, 2026, block 974,157 paid its miner 3.141 BTC, of which fees were just 0.016 BTC (mempool.space).

What a miner actually does

The popular image of miners "solving complex equations" is misleading. There are no equations, only trial and error. A miner does three things:

1. Builds a candidate block. It picks unconfirmed transactions from the mempool (usually those paying the most fee per byte), places a special transaction called the coinbase first, which pays the reward to the miner itself, and compresses all transactions into a Merkle tree with a single root hash.
2. Forms the block header: 80 bytes containing the version, the previous block's hash, the Merkle root, a timestamp, the encoded target (bits) and a free field called the nonce (Bitcoin.org Developer Guide).
3. Hashes the header with double SHA-256, changing the nonce until the output is below the target. Satoshi Nakamoto's white paper describes proof-of-work in one sentence: it is implemented "by incrementing a nonce in the block until a value is found that gives the block's hash the required zero bits" (bitcoin.org/bitcoin.pdf, section 4).

The found block is broadcast to the network. Other nodes verify it in a fraction of a second: are the signatures valid, is anything double-spent, is the hash really below the target. Verifying is cheap, finding is expensive. That asymmetry is the whole point of proof-of-work: to rewrite history you would have to redo all the work. For the structure of the chain itself, see our explainer on what blockchain is.

Hash, nonce and target: what the "work" looks like

A block hash is a 256-bit number. The target is another number the hash must not exceed. The lower the target, the fewer valid hashes exist and the more attempts are needed on average. Visually, a valid hash shows a long run of leading zeros: block 974,157's hash begins with 20 hexadecimal zeros, 000000000000000000001d9f….

The key detail: the nonce is only 4 bytes, about 4 billion values, "about 4 gigahashes" in the Developer Guide's words. A modern ASIC burns through them in a fraction of a second. So once the nonce is exhausted, the mining software changes the extra nonce inside the coinbase transaction. That changes the coinbase hash, which changes the Merkle root, which makes the header new again with a fresh 4 billion nonces to try. The search runs on two levels at once: the ASIC grinds nonces while the controller tweaks the coinbase and timestamp.

How many attempts does it take? That is what difficulty measures. On November 1, 2026, difficulty stood at 148.36 T and the estimated network hashrate was about 1,069 EH/s, roughly 10²¹ hashes every second (mempool.space, difficulty-adjustment; our Numbers block). How the network recalculates difficulty and why hashrate can only be estimated is covered in our piece on hashrate and difficulty.

Where the reward comes from: subsidy plus fees

A miner's reward has two parts. The first is the block subsidy, bitcoin that did not exist before this block. The white paper puts it this way: "By convention, the first transaction in a block is a special transaction that starts a new coin owned by the creator of the block," and compares it to gold mining, "analogous to gold miners expending resources to add gold to circulation" (bitcoin.org/bitcoin.pdf, section 6). The second part is transaction fees: for every transaction, the difference between inputs and outputs, which the miner is entitled to claim (Developer Guide, Block Chain).

The subsidy halves every 210,000 blocks, the event known as the halving. Since April 2024 it has been 3.125 BTC; by our data the next halving is expected around April 11, 2028, with 75,843 blocks remaining as of November 1, 2026. We cover the history of halvings and their price impact in a separate article.

Two rules about the reward matter in practice:

• Coinbase outputs cannot be spent for 100 blocks. This protects against the case where a block is orphaned and the reward disappears with it (Developer Guide).
• Fees are a small slice of revenue today. Over the 144 blocks ending at 974,157 (roughly one day), miners earned 452.03 BTC, of which fees were 2.03 BTC, about 0.45% (mempool.space, reward-stats). The recommended fee for fast confirmation that evening was 4 sat/vB. Why fees sometimes spike into the hundreds of sat/vB is explained in our guide to the mempool and fees.

Why a block arrives every 10 minutes

The network cannot see how many machines are running, but it can see how fast blocks arrive. Every 2,016 blocks it compares the actual elapsed time with the ideal 1,209,600 seconds (two weeks) and adjusts difficulty: up if blocks came too fast, down if too slow, with a cap on each adjustment (Developer Guide).

A live example: on October 29, 2026, at block 973,728, difficulty rose 5.54%. Over the following week the network still found 1,055 blocks instead of the "scheduled" 1,008, a block every 9.55 minutes on average, so miners were still ahead of the clock. The next retarget was expected around November 12 with a small decrease of roughly 0.5⁠–⁠0.7% (mempool.space).

The takeaway for miners: however much hashrate joins the network, roughly the same number of coins is issued per day, about 450 BTC at a 3.125 BTC subsidy. New machines do not grow the pie. They split it among more participants.

Solo or pool: how the reward is shared

At a network hashrate of 1,069 EH/s, one 270 TH/s machine is 0.000025% of the network. It can find a block, but the expected wait is measured in years. A solo miner with about 230 TH/s nonetheless found block 943,411 on April 2, 2026 through solo.ckpool.org and collected 3.139 BTC, about $210,000; CKpool developer Con Kolivas put the odds for a machine of that size at roughly 1 in 28,000 per day (mempool.space, CoinDesk). That is a lottery, not a business.

So nearly all hashrate runs through mining pools. A pool hands participants work with a much easier target: miners submit "shares," hashes that fall short of the network target but prove the work was done. When one of those hashes happens to beat the network target, the pool publishes the block. Rewards are split in proportion to shares, and income becomes frequent and predictable (Developer Guide). Payout schemes differ mainly in who carries the luck risk:

Braiins Pool, for example, computes FPPS as subsidy × (miner's shares ÷ network difficulty) plus average block fee × the same ratio, minus a 2.5% pool fee, with payouts once a day at 9:00 UTC (Braiins Academy). The pool switched to FPPS on December 12, 2023, explaining that when luck exceeds 100% the surplus goes into a reserve fund, which then covers payouts on unlucky days (Braiins).

Pools are noticeably concentrated. In the week to November 1, 2026, out of 1,055 blocks Foundry USA found 277 (26.3%), AntPool 215 (20.4%), F2Pool 176 (16.7%) and ViaBTC 98 (9.3%); the top three pools together produced 63% of blocks (mempool.space, Mining Pools). Keep in mind what a pool is: it does not own the machines, and participants can switch pools in minutes, but the pool does decide which transactions go into the block.

What a miner earns: hashprice and electricity

Miner revenue is best measured by hashprice, revenue per 1 PH/s per day. It is easy to compute from public data: in the day to November 1, 2026, the network paid 452.03 BTC; at a price near $77,980 (Binance, 6:00 PM UTC) that is about $35.2 million spread over 1,069 EH/s, or roughly $33 per PH/s per day. For comparison, Hashrate Index notes that in June 2024, when the Antminer S21 XP launched, hashprice was about $49 per PH/s per day (Hashrate Index).

The cost side is almost entirely electricity. Take the Antminer S21 XP: 270 TH/s at 3,645 W, or 13.5 J/TH (Hashrate Index):

1. Revenue: 0.27 PH/s × $33 ≈ $8.90 per day before pool fees.
2. Consumption: 3.645 kW × 24 h ≈ 87.5 kWh per day.
3. Break-even power price: $8.90 ÷ 87.5 ≈ $0.10 per kWh. At $0.05 the machine clears about $4.50 a day; at $0.12 it runs at a loss.

This is why mining geography looks the way it does: sites with cheap power win, whether that is stranded gas, surplus hydro and wind, or industrial tariffs. Home mining at residential rates is unprofitable in most countries at this step alone, before counting the hardware and cooling.

What it looks like on November 1, 2026

Bitcoin traded near $77,980 as of 6:00 PM UTC, which valued a single block at roughly $245,000. Current hashrate, difficulty and fee figures update in the Numbers block on our front page.

Energy and the critics

The most common objection to mining is electricity use. The Cambridge Centre for Alternative Finance's 2025 report estimated the network's annual consumption at 138 TWh (up 17% year on year), about 0.54% of global electricity; surveyed miners representing 48% of hashrate (268 EH/s) reported that 52.4% of their energy came from sustainable sources, and emissions were estimated at 39.8 MtCO₂e (Cambridge Judge Business School). The sustainable-energy share comes from a survey of miners themselves, which is its main limitation.

Economics pushes the industry toward cheap power, and cheap power is often surplus or far from consumers. But the net balance of costs and benefits is a live debate, not a settled fact, and an honest explainer should not pretend otherwise.

Common mistakes

• "Miners solve a hard math problem." No: they iterate nonces and compute hashes. There is no "solution" to compute cleverly, only brute force.
• "More machines means more bitcoin." Issuance is fixed at about 450 BTC per day at a 3.125 BTC subsidy. Rising hashrate raises difficulty, not the number of coins.
• "Fees are the miner's main income." On November 1, 2026 they were about 0.45% of revenue. The share climbs when the mempool is congested, but the subsidy still dominates.
• "The pool owns the hashrate." Pools coordinate, they do not own; miners switch freely. The concentration risk lies in transaction selection, not in "capturing" machines.
• "You can mine on a laptop or a GPU." Not bitcoin: CPUs and GPUs are millions of times slower than SHA-256 ASICs, and the electricity costs more than the coins earned.
• Forgetting pool fees and downtime. FPPS pools charge 2⁠–⁠2.5%, machines go offline for maintenance, and hashprice moves daily with price and difficulty.

Frequently asked questions

How long does it take to mine one bitcoin? A single miner does not "mine one bitcoin"; it earns fractions of the reward in proportion to its hashrate. At a hashprice near $33 per PH/s per day and a price near $77,980, a 270 TH/s machine earned about 0.000114 BTC per day, so one BTC would take roughly 24 years at constant conditions (calculated from November 1, 2026 data).

What happens when all 21 million coins are mined? The subsidy reaches zero around 2140, and miners will live on fees alone. The white paper allows for exactly that: "The incentive can also be funded with transaction fees." Whether fees will be enough to secure the network is an open question.

How is mining different from staking? In mining, the right to produce a block is bought with spent energy; in staking, with locked-up coins. Bitcoin uses proof-of-work only; we compare the two in our blockchain explainer.

Can I mine bitcoin in the cloud or with an app? Legitimate hosting services exist, but most "cloud mining" offers and mobile "tap-to-mine" apps are either resold hashrate at a markup or outright fraud. Check whether the service shows real machine data, a real pool and on-chain payouts.

This is not investment advice.