How Proof-of-Work Mining Revenue Actually Works
Bitcoin mining is a lottery in which your number of tickets equals your share of the global hashrate. Every ~10 minutes one miner (or pool) wins the right to append a block and collect the block subsidy plus the transaction fees inside it. Your expected share of blocks is simply your hashrate divided by network hashrate, so expected BTC per day = (your hashrate / network hashrate) × 144 blocks × the 3.125 BTC subsidy, plus a fee component that fluctuates with on-chain congestion.
Work the numbers for a single 100 TH/s machine against a 600 EH/s network. The share is 100e12 / 600e18 ≈ 1.667e-7, or about one block in six million. Multiply by 144 blocks per day and the 3.125 BTC subsidy and you expect roughly 0.000075 BTC per day. At $60,000 per BTC that is about $4.50 of gross revenue per day, before electricity, pool fees, and hardware costs.
Two things follow immediately: no input in this calculation is under your control except your own hashrate, and the two biggest ones — network hashrate and price — move constantly.
// Expected daily revenue: 100 TH/s vs a 600 EH/s network
const myHashrate = 100e12; // 100 TH/s in H/s
const netHashrate = 600e18; // 600 EH/s in H/s
const share = myHashrate / netHashrate; // 1.667e-7
const btcPerDay = share * 144 * 3.125; // 0.000075 BTC
const usdPerDay = btcPerDay * 60000; // $4.50 grossDifficulty: The Thermostat That Erodes Your Earnings
Difficulty is the protocol's thermostat. Every 2,016 blocks — roughly two weeks — Bitcoin compares how long those blocks actually took against the 10-minutes-per-block target and rescales difficulty so blocks keep arriving at the same pace no matter how much hardware joins. The direct consequence for a miner is that revenue per TH/s falls in proportion to network hashrate growth: if global hashrate rises 20% and your rig stays the same, your expected BTC per day drops by roughly one-sixth (1 − 1/1.2 ≈ 16.7%).
This is the single most common modeling error. A projection that holds today's difficulty flat for 12 or 24 months will badly overstate earnings, because network hashrate has historically grown by double-digit percentages per year as newer, more efficient machines ship. A more honest approach is to run scenarios: flat difficulty as the optimistic bound, plus 2–5% average monthly growth as base and pessimistic cases, and check whether the machine still pays for itself in the worst one.
Electricity, Efficiency, and the Break-Even Price
Electricity is the cost that decides whether a miner lives or dies. Daily power cost = watts ÷ 1,000 × 24 × price per kWh. An S19-class machine drawing about 3,250 W at 100 TH/s consumes 78 kWh per day. At a retail rate of $0.10/kWh that is $7.80 per day — more than the ~$4.50 of daily revenue from the previous example, a loss of about $3.30 per machine per day. The identical machine at an industrial rate of $0.05/kWh costs $3.90 per day and ekes out roughly $0.60 of daily margin. Nothing about the hardware changed; only the power contract did.
The metric that captures this is efficiency in joules per terahash: watts divided by TH/s. The example machine runs 3,250 / 100 = 32.5 J/TH, typical of the S19 generation; newer S21-class machines reach roughly 17.5 J/TH, meaning nearly half the electricity per unit of work. Each hardware generation resets the survival bar for everyone.
A useful sanity check is the break-even electricity price: daily revenue ÷ daily kWh. Here, $4.50 / 78 kWh ≈ $0.058/kWh — pay more than that for power and this rig loses money under these assumptions.
// Power cost and break-even electricity price (S19-class rig)
const watts = 3250, terahash = 100;
const kwhPerDay = (watts / 1000) * 24; // 78 kWh/day
const costRetail = kwhPerDay * 0.10; // $7.80/day -> loss vs $4.50 revenue
const costIndustrial = kwhPerDay * 0.05; // $3.90/day -> ~$0.60/day margin
const efficiency = watts / terahash; // 32.5 J/TH
const breakEvenKwh = 4.50 / kwhPerDay; // ~$0.058 per kWhHalving Economics: Revenue Cut in Half Overnight
Every 210,000 blocks — about four years — the block subsidy is cut in half: 50 → 25 → 12.5 → 6.25 → 3.125 BTC, with the next step to 1.5625 BTC already on the schedule. From a miner's perspective this is an overnight event: revenue per TH/s halves the moment the halving block is found, unless the BTC price rises or transaction fee revenue expands enough to compensate. Neither is guaranteed, and fees have historically covered only a small fraction of the subsidy outside of short congestion spikes.
Halvings are when marginal miners capitulate. Operators paying retail electricity who were barely profitable become clearly unprofitable and switch off; difficulty then adjusts downward somewhat, redistributing revenue to survivors with cheaper power. When you evaluate a machine with a claimed 12–36 month payback, check the calendar: if a halving falls inside that window, the projection must survive a 50% revenue cut mid-stream. Most optimistic vendor payback claims quietly assume it will not happen.
Pools, Payout Schemes, and Variance
With 100 TH/s your expected block find rate is about 0.000024 blocks per day — one block every ~114 years on average. The expected value is identical to the pool case, but the variance is unlivable: you would most likely earn nothing for decades and then 3.125 BTC at once. Pools exist to trade that variance away: members combine hashrate, and payouts are smoothed into small regular payments proportional to submitted shares.
Payout schemes matter. FPPS (full pay per share) pays you the expected value of both subsidy and transaction fees for every share, giving the smoothest income; the pool carries the luck risk and charges the highest fee, commonly 2–4%. PPS is similar but typically covers the subsidy only. PPLNS pays out of what the pool actually mined over a recent window, so your income tracks pool luck — lower fees, more variance, and loyalty matters, since hopping between pools forfeits pending shares.
The fee comes straight off the top: at 2% FPPS, the $4.50/day example becomes $4.41 before electricity. Always enter net-of-fee revenue in profitability projections.
// Solo-mining variance vs pool fee impact (100 TH/s)
const blocksPerDay = 1.667e-7 * 144; // 0.000024 blocks/day
const daysPerBlock = 1 / blocksPerDay; // ~41,667 days
const yearsPerBlock = daysPerBlock / 365; // ~114 years between blocks
const netRevenue = 4.50 * (1 - 0.02); // $4.41/day after 2% FPPS feeTotal Cost of Ownership, Common Mistakes, and Taxes
A machine that mines $0.60 of daily margin is not making money if it cost $3,000 and will be obsolete in three years. Total cost of ownership includes hardware depreciation (ASICs lose resale value fast — often most of it within a single generation cycle), cooling and ventilation, hosting or rack fees, and downtime. Uptime cuts asymmetrically: 5% downtime removes 5% of revenue but none of your fixed costs, so thin margins get hit disproportionately. Firmware tuning adds another lever: underclocking lowers J/TH and can turn a loss into a small profit at high power prices, while overclocking does the reverse.
The recurring mistakes are consistent: projecting with today's difficulty and today's price fixed for years, forgetting pool fees and downtime, ignoring the collapse in ASIC resale value, and treating vendor payback claims as guarantees. Run pessimistic scenarios first.
Taxes deserve their own line: in many jurisdictions mined coins are taxable as income at fair market value when received, and again on any gain or loss when sold. Rules differ widely by country — consult a qualified tax professional. Everything on this page, calculator included, is educational content only, not financial advice; mining involves a substantial risk of loss.