What Is Bitcoin Mining? How It Works, Costs and Hardware

A miner takes the block it wants to publish, adds a changing number called a nonce, and runs the pair through the SHA-256 hash function. The output is unpredictable, so the only way to find a hash below the target is to try quintillions of combinations per second. This deliberate waste of computation is what makes rewriting bitcoin history economically impossible: an attacker would have to out-spend the entire honest network in electricity and hardware.

Because the answer is hard to find but instant to verify, every other node on the network can check a submitted block in milliseconds. That asymmetry — expensive to produce, cheap to validate — is the core of proof of work and the reason bitcoin needs no trusted referee.

Hashrate measures how many hash attempts a machine makes per second, quoted in terahash (TH/s) for modern units. Your expected earnings are simply your hashrate divided by the total network hashrate, multiplied by the coins issued in that period. Nothing else in the formula is under your control.

Roughly every two weeks the network retargets difficulty so blocks keep arriving about every ten minutes. When more hashrate joins, difficulty rises and every existing machine earns slightly less bitcoin per day. Any profitability model that assumes today's difficulty forever is wrong; assume a steady climb and buy hardware efficient enough to survive it.

Each block pays a fixed subsidy that halves roughly every four years, plus the fees users attached to their transactions. Fee income spikes during congestion and can briefly rival the subsidy, but it is volatile and should never anchor a purchase decision. Model revenue on the subsidy and treat fees as upside.

Almost nobody mines alone. Solo mining a full-size ASIC means going years between wins on average, so miners join pools that combine hashrate and pay out proportionally, minus a one to three percent fee. Payouts are smooth and predictable; the trade-off is that you never hit a whole block yourself.

A modern full-size miner draws between three and five and a half kilowatts continuously — 72 to 132 kilowatt-hours a day. At twelve cents per kilowatt-hour that is roughly nine to sixteen dollars of electricity every day, per machine, forever. Over a three-year life the power bill routinely exceeds the purchase price several times over.

This is why efficiency in joules per terahash matters more than headline hashrate. A machine at 15 J/TH produces the same work for roughly half the electricity of a 30 J/TH unit. Above about ten cents per kilowatt-hour, only current-generation hardware makes sense; below six cents, cheap previous-generation machines often pay back fastest.

Full-size air-cooled ASICs such as the Antminer S19 and S21 families or MicroBT's WhatsMiner M50 and M60 series are the workhorses of the industry. They need a dedicated 240 volt circuit, real airflow and somewhere that tolerates 72 to 80 decibels of fan noise — in practice a garage, outbuilding, warehouse or hosted facility.

If the machine must live near people, hydro-cooled units and low-power home miners in the one to two kilowatt band are the sensible options. If you have no suitable site at all, hosting places your hardware in an industrial facility at a fixed per-kilowatt-hour rate, which is usually cheaper than residential power anyway.

Work in this order: confirm your delivered electricity rate, confirm the electrical capacity and location available, then pick hardware that fits both. Buying a miner before checking your circuit and your noise tolerance is the most common and most expensive beginner mistake we see.

Start with one machine, learn the firmware, pool configuration and thermal behaviour, then scale. Our team quotes delivered pricing per terahash and ships worldwide from Hong Kong, including to the USA, Canada and Europe, so you can compare real numbers rather than datasheet promises.