Bitcoin mining, costs and energy use explained

Bitcoin miners use specialized computers to compete to add blocks of transactions to Bitcoin’s shared ledger. They make repeated cryptographic guesses, consuming substantial electricity. Estimates of mining’s total energy use and cost per bitcoin vary with the date and assumptions used, as well as which expenses are included.

Efficient machines use less electricity per unit of computing work, but the network’s total use can still rise if miners add enough machines to increase its total computing power.

How Bitcoin miners add blocks

Miners assemble pending transactions into a candidate block and repeatedly calculate a hash, a short digital result produced from the block’s data. They change parts of the candidate and try again until they find a hash that meets Bitcoin’s target. Specialized computers called application-specific integrated circuits, or ASICs, perform these calculations.

Finding a qualifying hash takes repeated work, making it costly to propose a block. Full nodes, other network participants, verify the block and its transactions against Bitcoin’s rules with comparatively little work.

The protocol adjusts mining difficulty to keep blocks arriving about every ten minutes on average. That interval is an average across the network, not a promise that each block will take exactly ten minutes. Mining helps order transactions and makes changing past blocks expensive, while full nodes enforce the rules that miners must follow.

How Bitcoin miners earn revenue

A miner that adds a valid block can receive two kinds of payment. The block subsidy is newly issued bitcoin, and transaction fees come from users whose transactions the block includes. The subsidy follows a schedule that cuts it in half at each halving. Fees vary with demand for limited block space.

This makes mining a competitive business. Miners share the available block rewards according to their computing power, so a miner’s expected earnings depend partly on how much hashrate, or computing power, the entire network brings to the competition. Bitcoin’s price also affects the value of rewards. Mining continues to rely on fees as the subsidy declines.

Why Bitcoin energy estimates vary

There is no precise, directly metered global total for Bitcoin’s electricity use. Researchers estimate demand from network computing power and likely equipment efficiency. They also make assumptions about how miners operate. The worldwide mix of machines is not fully known, and researchers lack complete data on electricity prices and utilization rates. Different methods therefore produce different results.

The Cambridge Centre for Alternative Finance estimated annualized electricity use at 138.2 terawatt-hours (TWh) on June 30, 2024, or about 0.54% of global electricity use. It estimated 183 TWh annualized by the end of 2024. These figures describe the estimated demand at those points in time, projected across a year. They are not direct measurements of electricity used during complete calendar years.

Cambridge’s estimates drew on a survey of mining firms that represented 48% of global hashrate when the data was collected, alongside network and industry data. Its Bitcoin Electricity Consumption Index also gives a best estimate and theoretical bounds. Those ranges reflect uncertainty about the equipment in use, how efficiently it runs, and how miners respond to changes in profitability. When comparing figures, check the date and whether each one is an annualized estimate or a measured total.

What Bitcoin mining costs

Electricity is a major operating expense, but it is only part of the bill. A miner also has to account for ASIC purchases and replacement, along with facility costs, cooling, and maintenance. Financing, labor, and fees paid to mining pools add to the bill. A cost per bitcoin that includes only electricity cannot be compared directly with one that includes hardware and overhead.

Cambridge’s 2025 industry report found that survey respondents had a median electricity cost of $45 per megawatt-hour (MWh) and reported an all-in electricity cost of $55.50 per MWh. Electricity made up more than 80% of respondents’ cash-based operating expenses. Those figures describe the surveyed operators and the report’s data period; they are not a current tariff or a universal cost to produce one bitcoin.

The distinction also appears in company reporting. TeraWulf reported an average energy cost of $53,609 per bitcoin in 2025, based on a reported facility energy cost of $0.060 per kilowatt-hour. That is one company’s measure under its own operating and accounting conditions. It does not establish the cost for the whole Bitcoin network.

How to estimate a mining machine’s power

To estimate power draw, multiply a machine’s rated efficiency, expressed in joules per terahash (J/TH), by its hashrate:

Power in watts = efficiency in J/TH × hashrate in TH/s

For example, a machine’s estimated power draw depends on its rated efficiency and hashrate. If it ran continuously, its daily electricity use would depend on that power draw. Electricity cost would then depend on the amount used and the rate charged.

That calculation estimates a machine’s electricity use, not its cost per bitcoin or its profitability. Cooling and downtime are excluded, along with facility charges and hardware costs. The calculation also leaves out other expenses. Estimating revenue requires the network’s hashrate and difficulty, plus the block subsidy and transaction fees. It also requires uptime and any pool fees. Actual power use can differ from a machine’s specification under operating conditions.

Why efficiency gains do not guarantee lower energy use

Efficient ASICs use less electricity for each unit of computing work, but total demand depends on how much work miners perform across the network. Cambridge estimated that the weighted average efficiency of mining hardware improved from about 28.2 J/TH in June 2024 to 23.7 J/TH by December. Over the same period, network hashrate rose from 558.7 to 795.7 exahashes per second, and estimated annualized electricity use increased.

That pattern shows why a single efficiency figure cannot describe Bitcoin’s total energy use. Efficiency measures the power needed for computing work; network demand reflects both efficiency and the total amount of work being done.

Energy per bitcoin needs similar care. Dividing estimated network electricity use by the number of bitcoins mined during a period gives an average allocation. It does not show that a particular bitcoin, or an individual transaction, directly caused that amount of electricity use. Mining activity responds to reward values, competition, and equipment economics, rather than simply to transaction counts.

Where electricity affects the grid

The environmental effect of mining depends on where and when miners use electricity, and which power sources respond to that demand. Cambridge’s 2025 survey estimated that 47.6% of surveyed mining electricity came from fossil fuels and 52.4% from sources it classified as “sustainable.” That second category included 42.6% renewables and 9.8% nuclear. The report also estimated associated network-wide emissions.

These are survey-based estimates, not a complete measurement of every mine. The label “sustainable” also includes nuclear power, so it does not mean that all of that electricity was renewable. A mine’s reported energy mix alone does not show how much emissions its demand caused; that depends on where and when it operates and which power plants increase output to meet the extra demand.

Mining can also create local grid concerns even when a national electricity share looks modest. The U.S. Energy Information Administration estimated that cryptocurrency mining could account for 0.6% to 2.3% of U.S. electricity use. Its Bitcoin-specific estimate for 2023 ranged from 25 to 91 TWh and relied partly on an assumed U.S. share of global activity. The agency noted that miners can be difficult to identify among electricity customers and that equipment can move. National estimates therefore do not describe the impact of every site or region.

Mining machines can reduce power use or shut down, which gives some operators the option to respond to grid conditions. That flexibility has value only when operators reduce power use at useful times and take part in suitable market or utility arrangements. A Texas electricity-market study found that modeled demand response, in which miners reduce electricity use in response to grid conditions, reduced mining’s emissions effect. The findings do not mean every mining operation supports the grid or lowers emissions.