How Bitcoin Mining Works: Hash Rate, Difficulty Adjustment and Miner Economics

What a miner actually does

A block header is 80 bytes. According to the Bitcoin developer reference, it contains six fields:

A block is valid only if the hash of its header is at or below the target, which the reference describes as a 256-bit unsigned integer. The miner cannot compute which nonce will work. It has to try one, hash the header, compare the result with the target and try another. The reference says the nonce is the value miners change "to produce a hash less than or equal to the target threshold." The reference writes the header hash as SHA256(SHA256()), so each attempt is a double SHA-256.

A lower target means fewer hashes qualify, so more attempts are needed on average. That is all "difficulty" means.

• the block version (4 bytes);
• the hash of the previous block's header (32 bytes);
• the Merkle root hash of the block's transactions (32 bytes);
• a timestamp (4 bytes);
• nBits, a compact encoding of the target (4 bytes);
• a nonce (4 bytes).

Difficulty and hash rate

The Bitcoin Wiki defines difficulty as "a measure of how difficult it is to find a hash below a given target." Its simplified rule of thumb is that the expected number of hashes needed to find a block is difficulty × 2^32. Dividing by the 600-second target block time gives the network hash rate implied by a given difficulty: difficulty × 2^32 / 600\.

Using the difficulty reported by Hashrate Index for September 2026, 127.45 trillion:

Hashrate Index reported a 7-day average hash rate of 943 EH/s for the week to 14 September 2026 and an average block time of about 9 minutes 30 seconds over the previous 24 hours. Those two facts fit together: blocks were coming faster than ten minutes because more hash power was running than the current difficulty assumed. That is what the adjustment is for.

• Expected hashes per block: 127.45 × 10^12 × 2^32 ≈ 5.5 × 10^23 (our calculation).
• Implied hash rate at exactly 10-minute blocks: about 912 EH/s (our calculation). One EH/s is 10^18 hashes per second.

How difficulty adjusts

The rules, as described by learnmeabitcoin:

Three examples (our calculation, using a 14-day expected period):

The September 2026 data point from Hashrate Index shows the same logic. With blocks averaging about 9.5 minutes, the estimated next adjustment was \+5.26%. The ratio 600 / 570 seconds is 1.0526, so the estimate is consistent with the formula. Hashrate Index labels that figure as an estimate from Luxor that may change.

The adjustment has a delay built in. Hash rate changes between adjustments are not reflected until the next one, so block times drift while an epoch is under way, as the 9.5-minute figure shows.

• The target adjusts every 2016 blocks, which is roughly two weeks at ten minutes per block (20,160 minutes).
• The new target equals the old target multiplied by the ratio of actual time to expected time. The expected time is 2016 × 10 × 60 seconds.
• A single adjustment is limited by a factor of 4, either ×0.25 or ×4.
• The calculation actually spans 2015 block intervals rather than 2016, an off-by-one quirk that the page says still exists. It is small, so the examples below ignore it.

Where miners get paid

A miner who finds a valid block collects the block subsidy plus the transaction fees in the block. The subsidy halves every 210,000 blocks, which is about every four years. The fourth halving took place at block height 840,000 in April 2024 and cut the subsidy from 6.25 BTC to 3.125 BTC, according to The Block. At 144 blocks per day, that is about 450 BTC of new issuance daily (our calculation, and The Block cites the same figure).

Fees are a small part of revenue at the moment. For the week ending 14 September 2026, Hashrate Index reports that fees were 0.59% of block rewards, about 19 BTC out of roughly 3,253 BTC. That is one week, and fee share varies a lot over time. In that week, nearly all miner revenue came from the subsidy.

Hashprice: the miner's revenue per unit of hash power

Because miners differ in size, revenue is usually quoted per unit of hash power. Hashrate Index calls this hashprice: expected revenue per petahash per second (PH/s) per day. As of 14 September 2026 it stood at $39.25 per PH/s per day (7-day average $38.60). It moves with the bitcoin price, the network hash rate, the subsidy and fees.

A worked example: one machine

All machine numbers below are hypothetical. Hashprice and network figures are the September 2026 data above.

The machine: 200 TH/s (0.2 PH/s) at an efficiency of 17.5 joules per terahash (J/TH).

Power: 200 TH/s × 17.5 J/TH \= 3,500 W \= 3.5 kW. Over a day that is 84 kWh.

Revenue: 0.2 PH/s × $39.25 \= $7.85 per day, before any pool fee, maintenance, hosting charge or tax.

Profit at different electricity prices (our calculation):

Break-even electricity price: $7.85 / 84 kWh \= about $0.0935 per kWh. Above that price, the machine loses money on electricity alone.

Cross-check: Hashrate Index also publishes revenue per megawatt-hour by fleet efficiency. For machines at 14–19 J/TH it reports $97/MWh, which is $0.097/kWh. That is close to our $0.0935, which is a useful sanity check on the arithmetic.

Sensitivity: If hashprice fell 20%, the break-even price would fall to about $0.075/kWh (our calculation). Efficiency matters just as much. At the same hashprice, the same Hashrate Index data show $50/MWh for 25–38 J/TH machines, so a less efficient machine is under water at electricity prices where a modern one still earns a margin. Hashrate Index is a product of a mining company (Luxor), so treat its figures as industry data, not an independent audit.

Why solo mining is a lottery

Your chance of finding any given block is your share of the network's hash power. With 200 TH/s against 943 EH/s, the share is about 2.1 × 10^-7. At 144 blocks a day, the expected number of blocks found is about 3.1 × 10^-5 per day, one block per roughly 32,700 days, or about 90 years (our calculation). The expected value is the hashprice figure above. The variance is huge, and that gap between a steady expected income and a very unpredictable actual income is the problem mining pools are designed to smooth by splitting rewards among members. (That is our reading of why pools exist, not a claim from the sources.)

The feedback loop

When mining is profitable, more machines come online, the hash rate rises, and the next adjustment raises difficulty. A higher difficulty means each unit of hash power finds a smaller share of the same fixed reward, so hashprice falls. When it is unprofitable, the least efficient machines switch off, hash rate falls, and difficulty comes down at the next adjustment. This is general economic logic and not a sourced forecast, but it is also how the Cambridge Centre for Alternative Finance models the network. Its electricity index treats miners as rational operators who run hardware only while revenue covers the electricity cost (price per joule × efficiency).

What this means for energy estimates

Nobody can measure Bitcoin's total electricity use directly, so it has to be estimated. The Cambridge index (CBECI) estimates it from the hash rate, with assumptions about hardware efficiency and electricity cost. It uses a default of $0.05/kWh, a data-centre overhead factor (PUE) of 1.01 for its lower bound, 1.10 for its best guess and 1.20 for its upper bound, and it publishes lower-bound, best-guess and upper-bound estimates. Cambridge states that results depend heavily on the electricity price assumption and that the exact hardware mix is unknown. Treat any single energy figure for Bitcoin as an estimate within a range.

What to take from this

This article is general information, not financial advice. The machine in the worked example is hypothetical, and network data are as of mid-September 2026\.

1. Difficulty follows hash rate, not price. Price affects difficulty only by changing how many miners find it worth running.
2. Block time drifts between adjustments. A faster-than-ten-minute average means hash power rose within the epoch.
3. Efficiency and electricity price decide who survives. A margin that looks fine at $0.05/kWh can disappear at $0.10.
4. Hashprice is a snapshot. It changes daily with price, hash rate and fees, so recalculate with current data.
5. Fees are small today. Roughly 0.6% of rewards in the week we checked, so revenue still depends mostly on the subsidy and the bitcoin price.