Bitcoin halving: how it works and what it changes

You have probably seen the word halving associated with Bitcoin, almost always accompanied by price predictions. This article separates the two. First, the rule: what the halving is, where it is written and when it happened. Then, the implications: what changes mechanically and what is market interpretation.

If you are not yet clear on what blocks, miners and transactions are, it is worth reading the guide to what Bitcoin is first.

What the halving is

On the Bitcoin network, new blocks of transactions are produced by miners: participants who spend energy and equipment to solve a computational problem called proof of work. Whoever manages to add a valid block receives a reward.

That reward has two parts:

• Block subsidy. New bitcoins, created at that moment. This is how every unit in circulation entered the system.
• Transaction fees. Amounts paid by users whose transactions were included in the block.

The halving is the cut in the subsidy. It does not affect fees or bitcoins that already exist. It only affects how many new bitcoins each block can create from then on.

Where the rule is written

The document that introduced Bitcoin, signed with the pseudonym Satoshi Nakamoto, describes the general idea in the “Incentive” section. By convention, the first transaction in each block creates new coins for whoever produced the block. The text adds that once a “predetermined number of coins” has entered circulation, the incentive can transition entirely to transaction fees (Bitcoin whitepaper).

Note a detail: the whitepaper does not mention the number 21 million or the 210,000-block interval. Those values are in the software.

In Bitcoin Core, the network’s reference implementation, the function that calculates the subsidy is short. It starts at 50 bitcoins and divides that amount by two at each halving interval. The code comment says the subsidy “is cut in half every 210,000 blocks”, which happens “approximately every 4 years” (validation.cpp, Bitcoin Core). The interval appears as a main network parameter: nSubsidyHalvingInterval = 210000 (chainparams.cpp, Bitcoin Core).

Why “every four years” is approximate

The halving is not set by date. It is set by block height, that is, by the block’s position in the chain (block 210,000, block 420,000 and so on).

The protocol targets one block every 10 minutes. Every 2,016 blocks, it recalibrates the proof-of-work difficulty to keep that pace; Bitcoin Core’s parameters specify a target of two weeks per period and 10 minutes per block (chainparams.cpp). Because blocks come out a little faster or slower as mining capacity changes, the date of each halving shifts.

By simple arithmetic, 210,000 blocks times 10 minutes is about 1,458 days, a little under four years.

The halvings that have already happened

As of the date of this article, there have been four cuts. The dates below are the block timestamps, in UTC, checked by the editorial team on the mempool.space explorer and in the Bitcoin Wiki table:

Block	Date (UTC)	Subsidy before	Subsidy after
210,000	November 28, 2012	50 BTC	25 BTC
420,000	July 9, 2016	25 BTC	12.5 BTC
630,000	May 11, 2020	12.5 BTC	6.25 BTC
840,000	April 20, 2024	6.25 BTC	3.125 BTC

The times are those recorded in the blocks themselves and serve as a date reference. Block 840,000, for example, has a timestamp of 00:09 UTC on April 20, 2024, which in Brasília time was still the night of April 19 (mempool.space, block 840,000).

The next halving

The fifth cut happens at block 1,050,000, when the subsidy falls to 1.5625 BTC. On October 8, 2026, the chain was close to block 970,500 (mempool.space). That leaves about 79,500 blocks to go.

By the editorial team’s calculation, at 10 minutes per block this comes to about 550 days, pointing to sometime in the first half of 2028. It is an estimate: the actual date depends on the pace of blocks until then. Be wary of countdowns with an exact time set long before the event.

How long the halving continues

The subsidy is calculated in satoshis, Bitcoin’s smallest unit (one bitcoin has 100 million satoshis). Because the division uses whole numbers, there comes a point where half of 1 satoshi becomes zero.

According to the Bitcoin Wiki, the last block with a non-zero subsidy comes before block 6,930,000, and issuance should end around 2140; the page itself notes that predicting the exact year is difficult (Bitcoin Wiki). The code also forces the subsidy to zero after 64 halvings, as a technical safeguard (validation.cpp).

The sum of all subsidies is slightly below 21 million bitcoins: about 20,999,999.9769 BTC, according to the same wiki page. The number 21 million that appears in the code, the MAX_MONEY constant, is not the total supply. The comment in Bitcoin Core says it is a sanity check, that the actual supply “is lower than 21,000,000 BTC for various reasons” and that changing this constant could cause a network split (amount.h, Bitcoin Core).

By the editorial team’s calculation, applying the subsidy rule up to block 970,500, the maximum issued so far is close to 20.1 million bitcoins, about 96% of the total. It is a theoretical ceiling: some of these bitcoins may be lost forever, for example when someone loses their access keys, and that number cannot be observed precisely.

What actually changes: miners

The direct and verifiable consequence of the halving is this: in the block after the cut, the miner receives half the new bitcoins they received before. Fee revenue does not change because of the halving.

This has three mechanical consequences.

Revenue per block in bitcoin falls. If a miner’s energy and equipment costs do not change and the subsidy is halved, their margin shrinks. A January 2020 report from the Swiss bank AMINA (formerly SEBA) describes exactly this squeeze: since the subsidy made up almost all of miners’ revenue, they would be under pressure unless the dollar price rose (AMINA Research, 2020).

The network rebalances through difficulty. If miners switch off machines, blocks come out more slowly until the next adjustment. Then difficulty drops and the pace returns to around 10 minutes. This mechanism exists for any change in capacity, not just the halving.

Fees become more important over time. With each cut, fees come to represent a potentially larger share of the reward. In individual blocks, they can already exceed the subsidy: block 840,000 itself paid about 37.6 BTC in fees, against a 3.125 BTC subsidy (mempool.space). It was an atypical block and should not be read as an average.

The long-term question: security paid only by fees

Bitcoin’s security depends on miners having an incentive to follow the rules. When the subsidy approaches zero, that incentive will come almost entirely from fees.

A study by Princeton researchers, presented at the ACM CCS conference in 2016, argues that this is not neutral. With rewards coming only from fees, the value of each block varies widely, and it may become worthwhile to redo a fee-“rich” block to take those fees for oneself. The authors point to “worrying” implications for future security, based on theoretical analysis and simulation (Carlsten et al., 2016).

It is an open debate, not a prediction of failure. The point for you is to know that the halving has a long-term structural consequence beyond the price discussion.

What is narrative: halving and price

Here we leave the realm of rules and enter that of interpretation. None of the ideas below is a protocol rule.

The scarcity narrative. The most common argument is that if new supply falls and demand holds steady, the price would tend to rise. AMINA’s 2020 report observed that previous Bitcoin halvings were neutral or positive for the price in periods when mining capacity remained stable.

The efficiency objection. The same report presents the counterpoint: the issuance schedule has been public from the start. Under the efficient market hypothesis, information known to everyone would already be reflected in the price. The authors also point out that, with most of the supply already issued, the supply shock tends to be smaller with each cut, and that without analyzing demand, any conclusion about the effect of supply is incomplete.

Statistical evidence is limited. There have been only four halvings, each in a different economic context. An October 2025 paper posted on arXiv, not yet peer-reviewed, tried to measure the causal effect using the synthetic control method. The author found signs of a positive effect of the 2024 halving on price about three months later, but did not obtain a statistically significant and robust result for the 2020 halving (Virtonen, 2025).

In the editorial team’s assessment, the honest picture is this: there are plausible narratives pointing in different directions, few events to test them and results that do not repeat consistently. Patterns observed in past halvings guarantee nothing about future ones.

Risks you need to consider

If the halving sparks your interest in buying or holding bitcoin, consider the risks that do not depend on it:

• Volatility. The price can swing sharply in a short time, up or down, including around widely expected events.
• Total loss. You can lose the entire amount invested, through a price drop, an operational error or the loss of the keys that give access to the bitcoins.
• Custody. Leaving bitcoins on an exchange means depending on its solvency and security. Holding them yourself shifts responsibility for the keys to you.
• Opportunistic scams. Heavily publicized dates attract offers promising to multiply your money “by taking advantage of the halving”. A promise of guaranteed gains is a red flag, with or without a halving.
• Rule changes. Changing the schedule would require the network to adopt software with different rules, which is difficult and could split the chain. There is no sign of this, but the rule depends on the consensus of participants, not on a law of physics.

Key takeaways

• The halving is a software rule: every 210,000 blocks, the subsidy is cut in half. There have been four cuts, and the current subsidy is 3.125 BTC per block.
• The next one should occur at block 1,050,000, around the first half of 2028 by the editorial team’s estimate. The exact date depends on the pace of blocks.
• The mechanical effect falls on miners and, in the long run, on the network’s security model, which will rely more and more on fees.
• The link with price is narrative, not rule. The evidence is scarce and mixed, and past patterns do not guarantee future results.

To compare with another network that has neither a fixed supply cap nor a halving, see what Ethereum is and how it works.