Bitcoin miners run specialized computers that make trillions of guesses per second until one produces a block hash below a network target. The winner earns new bitcoin plus transaction fees. Is it still profitable? As of October 2026, mainly for operators with cheap power. A 200 TH/s machine breaks even near 9 cents per kilowatt-hour, and at typical US home rates it loses money every day, before you even count its purchase price.
What is the hash puzzle?
Mining is a guessing contest in which the only strategy is to guess faster. Each Bitcoin block has an 80-byte header containing the previous block's hash, a summary of the block's transactions, a timestamp, the current target and a number called the nonce. A miner runs that header through the SHA-256 hash function and checks the result. A block is valid only if its header hash doesn't exceed the target.
The output is effectively random, so the miner changes the nonce and tries again. When all the nonce values run out, it changes the block's first transaction, which alters the summary and opens a fresh set of guesses. Finding an answer takes enormous effort, but checking one takes a single hash. This is the "work" in proof of work.
Miners measure speed in hashes per second. One terahash (TH) is a trillion hashes, one petahash (PH) is 1,000 TH, and one exahash (EH) is 1,000 PH. At the difficulty set on October 16, 2026, about 140.58 trillion, a block takes about 604 sextillion hashes on average. That implies a network speed of about 1,000 EH/s.
How does the difficulty adjustment work?
Every 2,016 blocks, the network compares how long those blocks took with a two-week target and resets the difficulty to match. Changes are capped at a 300% rise or a 75% fall. Two real examples from 2026:
- Up. The 2,016 blocks before October 16 took about 13 days and 5 hours. That's faster than 14 days, so difficulty rose 5.92%, from 132.72 trillion to 140.58 trillion.
- Down. On June 14, difficulty fell 10.09%, from 138.96 trillion to 124.93 trillion, because blocks had been arriving too slowly.
This has a consequence most newcomers miss: more mining power doesn't create more bitcoin. It only splits the same reward among more machines. Your machine's output can shrink even when nothing about it changes.
What does a miner earn?
A miner earns the block reward, made up of the subsidy and the transaction fees. The block's first transaction pays it out, and those coins can't be spent for 100 blocks.
- The subsidy has been 3.125 BTC per block since block 840,000 in April 2024, and it halves every 210,000 blocks. The next halving, at block 1,050,000, is expected around 2028.
- Fees are currently tiny. They averaged 0.0199 BTC per block in September 2026, or 0.63% of rewards, and had been below 1% for 15 straight months.
Fees can spike. The halving block itself paid about 37.6 BTC in fees, around 12 times its subsidy. Spikes like that are too rare to plan a business around.
Why do miners join pools?
Miners join pools because solo mining is a lottery with a very long wait. A 200 TH/s machine has about a 1 in 5 million chance of finding any given block at October's difficulty. That works out to about one block every 96 years on average.
A pool combines many miners and shares the rewards. Each miner submits shares, which are valid hashes against an easier target, and these prove how much work it did. Payout methods vary:
- FPPS pays a fixed amount per share, including an estimate of fees, and the pool carries the luck.
- PPLNS pays based on recent shares when the pool actually finds a block, so your income varies.
Listed pool fees mostly run from 0% to 2.5%. Pools also concentrate power over which transactions get included. Mempool.space data from October 2026 shows the top three, Foundry USA, AntPool and F2Pool, finding about 64% of blocks. A pool doesn't own all that hardware, and its miners can leave, but this is still worth watching.
What is hashprice?
Hashprice is the expected daily revenue from one unit of hash power, usually quoted in dollars per PH/s per day. It's the most useful number in mining, because it combines price, difficulty, subsidy and fees. It rises with the bitcoin price and fees, and falls as difficulty rises.
| Period | USD hashprice per PH/s per day |
|---|---|
| 21 months before the 2024 halving, average | $78.43 |
| May 2024 to March 2026, average | $48.52 |
| June 2026, record-low monthly average | $30.37 |
| June 6, 2026, record daily low | $27.74 |
| September 2026, average | $39.33 |
A worked profitability example
This example estimates daily results for one machine at mid-October 2026 conditions. Your numbers will differ. These are the assumptions:
- Machine: 200 TH/s at 3,500 watts, or 17.5 joules per terahash (Antminer S21 class).
- Difficulty: 140.58 trillion, which gives about 0.000143 blocks per PH/s per day.
- Reward: 3.145 BTC per block (3.125 subsidy plus 0.02 fees).
- Bitcoin price: $82,000, close to its level on October 20.
- Pool fee: 2%.
Revenue. 0.000143 blocks × 3.145 BTC gives 0.00045 BTC per PH/s per day. At $82,000, hashprice is about $36.90. A 0.2 PH/s machine earns $7.38, or $7.23 after the pool fee, which is 0.000088 BTC.
Power. 3.5 kW × 24 hours is 84 kWh a day, so the result depends on your rate:
| Electricity rate | Daily power cost | Daily profit | Power cost per BTC mined |
|---|---|---|---|
| US average residential, 18.16¢ | $15.25 | -$8.02 | About $172,900 |
| US average industrial, 9.64¢ | $8.10 | -$0.86 | About $91,800 |
| Average for industrial bitcoin miners, 4.8¢ | $4.03 | +$3.20 | About $45,700 |
The two US averages are August 2026 figures. The break-even rate here is about 8.6¢ per kWh. Newer machines raise it: Luxor put September's break-even power cost at about 13.7¢ for machines under 14 J/TH, against 9.9¢ for the 14 to 19 J/TH group. They also cost more to buy.
Hardware. Suppose the machine costs $3,000, an illustrative figure. At $3.20 a day it takes about 2.6 years to pay back, and that assumes hashprice stays flat. Rising difficulty would stretch that out, and at today's block pace the next halving is about 1.5 years away. It would cut the subsidy to 1.5625 BTC. Payback only works if price, fees or falling difficulty make up the gap.
What has changed since the 2024 halving?
Mining is now tighter and more dependent on price. Average hashprice fell 38% after the halving, and fees haven't filled the hole. Bitcoin's drop from a peak near $125,000 in October 2025 to an average near $70,000 in March 2026 helped push hashprice to record lows.
Three shifts stand out:
- Costs caught up with listed companies. CoinShares puts their weighted average cash cost at about $79,995 per bitcoin in Q4 2025. That is a company-wide cash cost, so it isn't comparable with the power-only table above.
- Power is being sold to AI. CoinShares counted over $70 billion in AI and high-performance computing deals signed by miners as of early 2026. Hashrate Index links hashrate declines in Canada and the UAE to power being redirected this way.
- Geography keeps moving. The US led at about 35.6% of hash power in Hashrate Index's Q4 2026 estimate, followed by Russia and China. Operators follow cheap power.
Long term, this feeds a question for Bitcoin: can fees replace the subsidy as it keeps shrinking? Nobody knows yet. Other networks have exited the problem entirely, as Ethereum did when it moved to proof of stake in 2022.
How should a hobbyist think about it?
Treat home mining as an expensive way to get bitcoin, or as a hobby, and don't count on it as income. A few checks make that concrete:
- Compare it with buying. At the US residential rate, the $15.25 power bill would buy about 0.000186 BTC at $82,000. The machine produces 0.000088 BTC. You'd be paying more than twice the market price in electricity alone. Buying directly, or through a spot bitcoin ETF, skips the hardware, though each route has its own risks and costs. Mined coins also need secure storage in a wallet.
- Count the heat honestly. Nearly all of a miner's electricity becomes heat. If you'd otherwise pay for electric resistance heating, part of the bill may offset that, but only when you need the heat. A heat pump is more efficient, so the offset is smaller.
- Check the practical limits. A listed S21 runs at about 75 dB, and one hardware seller's spec sheet calls for 220 to 277 volts through a 20-amp input. Plan on an electrician, and check the manufacturer's datasheet.
- Treat tiny miners as lottery tickets. A 1 TH/s device mining solo has about a 1 in 1 billion chance per block, or one block every 19,000 years on average. Someone may get lucky, but you shouldn't budget for it.
- Rerun the numbers with your rate. Use your real price per kWh, the machine's efficiency and the current hashprice. Then test a 20% rise in difficulty.
Mining income can also be taxable, and rules differ by country, so ask a tax professional if you earn any. This article explains the mechanics and isn't financial advice. For more, see our mining guides and market explainers.
Key takeaways
- Mining is a guessing race: miners hash block headers until one falls below the target.
- Difficulty resets every 2,016 blocks, so more hash power means a smaller share and not more bitcoin.
- Rewards are almost all subsidy, with fees below 1% in recent months.
- Hashprice sums up revenue, and it hit record lows in 2026 after the 2024 halving.
- In our example, a 200 TH/s machine breaks even near 8.6¢ per kWh, before hardware costs.
- At the US average home power rate, the machine in our example pays more than twice the market price of the bitcoin it produces, in electricity alone.
Sources
- Bitcoin developer guide: block chain
- Luxor Hashrate Lookback: September 2026
- Hashrate Index: Halfway Across the Halving
- Luxor documentation: hashprice
- Mempool.space: difficulty and hashrate data
- US Energy Information Administration: retail electricity prices
- CoinShares: Bitcoin mining report, Q1 2026
Get the weekly commit
New blockchain deep dives every week.
