The bitcoin halving, explained through 210,000 blocks

The halving is a single rule in the program code of bitcoin: every 210,000 blocks the reward credited to a miner for a block found is cut in half. It has stood unchanged since 2009, it takes effect automatically, and the cap of just under 21 million units follows from it.

What exactly happens at a halving

New bitcoin arise only as a block reward. It started at 50 BTC per block and falls by half every 210,000 blocks. Because a block appears on average about every 10 minutes, so roughly 144 a day, one stage lasts about four years. Nobody has to do anything for it: every full node knows the formula and rejects a block whose reward comes out too high.

The calculation behind it is short. 210,000 blocks times 50 BTC gives 10,500,000 BTC in the first stage. The second stage delivers half of that, 5,250,000 BTC, the third 2,625,000 BTC, and so on. The sum of that series approaches 21,000,000; because of the rounding in the code, the final figure that can be reached by calculation is 20,999,999.9769 BTC. If computing power stayed constant, the last fraction would fall around the year 2140. Why the network pays out the reward at all is explained by the whitepaper in its section on incentives: it brings new units into circulation and at the same time pays the computers that secure the network.

The counting is not done in bitcoin but in the smallest unit. One bitcoin consists of 100,000,000 satoshi, so the first reward was 5,000,000,000 satoshi. Every halving divides that whole number, and because remainders drop away in the process, the total ends up marginally below 21 million. After 33 halvings the reward has reached zero by calculation, and from then on transaction fees alone carry the income of miners.

The four halvings so far

Four halvings have taken place so far. The table shows the date, the block height, the reward and the quantity that still arises per day afterwards, calculated with 144 blocks a day.

The column on the far right shows how early the greater part of the quantity had been handed out: after the second halving in 2016, 75 per cent of all units that can ever be reached had already been created, and after the fourth in 2024 around 93.75 per cent. Everything else is spread over 29 future halvings.

In practice the intervals were never exactly equal. Between the start on 3 January 2009 and the first halving on 28 November 2012 lay just under 3 years and 11 months, between the first and the second around 3 years and 7 months, between the second and the third about 3 years and 10 months, and between the third and the fourth just under 4 years. The reason lies in computing power: if it grows faster than the difficulty follows, blocks appear on average a little faster than in 10 minutes.

Why the next date can only be estimated

The block height is fixed, the date is not. Bitcoin counts blocks, not days. The difficulty of the computing task is reset every 2,016 blocks, in such a way that those 2,016 blocks ideally take exactly two weeks. Between two adjustments, though, computing power can rise or fall, and then blocks run faster or slower than the target of 10 minutes.

Over 210,000 blocks, small deviations add up. An average deviation of just 10 seconds per block gives a shift of around 24 days over a whole stage. That is why it makes sense to name the block height for a halving and a date only as an approximation. The halving countdown converts the remaining blocks into an estimate and shows the assumption it makes while doing so. How the difficulty adjustment works in detail is set out in the piece on how a blockchain works.

What the halving means for miners

For miners it is a cut on a single day: on 20 April 2024 daily new issuance fell from 900 to 450 BTC, while costs stayed the same. The University of Cambridge survey from April 2025 shows how tight the arithmetic is. The 49 companies surveyed, which on the reference date of 30 June 2024 together supplied just under 48 per cent of the derived network power, named a median electricity price of 45 US dollars per MWh and total costs of 55.5 US dollars per MWh; electricity accounted for more than 80 per cent of their running costs. The estimated annual consumption of the network on that date stood at around 138 TWh, and the derived network power at the end of 2024 at 796 EH/s.

Two adjustments follow. First, uneconomic machines are switched off; the companies surveyed expected that between 30 June and 31 December 2024 around 11.1 per cent of the network power of 30 June 2024, that is 61.8 EH/s, would go out of service. The study itself qualifies this by saying that a considerable share of that stood idle only temporarily, because machines were rebuilt, overhauled or set up again elsewhere. If power drops out, the difficulty falls at the next adjustment, and the rest earn more again per unit. Second, the source of income shifts slowly from the block reward to transaction fees. On that the study records that today's fee levels are nowhere near enough to carry the security of the network at its present level. That is the heart of the debate about the so called security budget.

What the halving is not

It is no guarantee for the price and no date on which something has to happen. There are 4 data points so far, and 4 data points carry no rule. The supply side also changes only slowly: after the fourth halving, 450 BTC a day were added. Measured against the stock that follows from the reward schedule, that is 19,687,500 at block 840,000 plus 450 a day since then, and so by now a good 20 million units, that comes to less than 0.003 per cent a day.

How unreliable expectations are is shown by a figure from the same survey. The companies surveyed, that is specialists with their own money at stake, put the bitcoin price at the end of 2024 at a median of 80,500 US dollars, with a stated range of 60,000 to 150,000 US dollars. It actually stood at 93,390 US dollars. On network power the same specialists were fairly accurate: 83 per cent of them expected a figure between 600 and 900 EH/s at the end of 2024, the median stood at 750 EH/s, and it actually came to 796 EH/s. Technical quantities can therefore be estimated better than prices. What price swings mean for investors is dealt with in the piece on the volatility of bitcoin.

How to work out the next date yourself

You need two figures for that: the current block height and the target height of 1,050,000. The difference divided by 144 gives the number of days at an average of exactly 10 minutes per block. If the block height stands at 966,000, for example, that is 84,000 blocks or around 583 days. Work out a range rather than a date: at 9.5 minutes per block it is about 554 days, at 10.5 minutes about 612 days.

Three things are worth keeping apart. First the mechanism: the halving is certain, its timing is not. Second the effect on supply: it is real, but small in relation to the stock. Third the effect on the price: it can neither be derived from 4 past events nor predicted. Anyone wanting to read up on the basics will find them in the piece on bitcoin.

Frequently asked questions

Sources