Bitcoin Halving Explained: Schedule, Math and 2028 Date

The Bitcoin halving is a scheduled event, written into Bitcoin's code, that cuts the reward paid to miners for producing a block by 50%. It happens every 210,000 blocks, roughly every four years, and it governs the pace at which new bitcoin enters circulation until issuance eventually stops.

In the two years since the April 2024 halving, mining revenue per unit of computing power has fallen sharply, and the industry has responded by redirecting billions of dollars of power capacity toward artificial intelligence workloads. Meanwhile the transaction fees meant to eventually replace the block subsidy have stayed near record lows, accounting for well under 1% of what miners earn.

This guide covers how the halving works at the protocol level, the full schedule from 2012 to 2140, what the 2024 halving did to mining economics, and the open question of how Bitcoin gets paid for once the subsidy fades.

Use the self-custody Bitcoin.com Wallet app, trusted by millions to safely and easily buy, sell, trade, manage, and use Bitcoin and the most popular cryptocurrencies.

Key Takeaways

• The halving cuts the block subsidy in half every 210,000 blocks, roughly every four years. The subsidy is the newly created bitcoin paid to whoever mines a block.
• Four halvings have occurred: 2012, 2016, 2020 and 2024. The current subsidy is 3.125 BTC per block, or roughly 450 new BTC per day.
• The next halving is expected around April 2028 at block 1,050,000, when the subsidy falls to 1.5625 BTC.
• Bitcoin's annual issuance rate is now under 1%, roughly half the rate at which new gold enters the market.
• The halving does not change the 21 million supply cap. It changes the speed of arriving there. Issuance ends around 2140.
• Mining economics tightened considerably. Hashprice, the revenue a miner earns per unit of computing power, has traded between roughly $29 and $40 per petahash per day during 2026, against about $70 in 2020.
• Miners diversified into data centers. More than $70 billion in AI and high-performance computing contracts have been signed across the sector, and some operators have switched off hashing capacity to serve AI clients.
• Transaction fees are not replacing the subsidy. In early September 2026 fees accounted for around 0.43% of miner rewards over a 24-hour period.

What Is the Bitcoin Halving?

The Bitcoin halving is the programmed 50% reduction of the block subsidy, occurring once every 210,000 blocks.

A block is a bundle of confirmed transactions added to Bitcoin's ledger. Miners compete to produce them, and the network targets one new block roughly every ten minutes.

The block subsidy is the newly created bitcoin awarded to whoever produces a block. It is separate from transaction fees, which users attach to their own transactions to encourage inclusion. Subsidy plus fees together make up the block reward. Only the subsidy halves. Fees are set by demand for space in a block.

When Bitcoin launched in January 2009, the subsidy was 50 BTC per block. It has halved four times since, and now stands at 3.125 BTC.

The design comes from Satoshi Nakamoto and appears in the original Bitcoin whitepaper. Nakamoto compared the steady creation of new coins to gold miners expending resources to add supply to circulation, and described an endpoint where miner incentives could shift entirely to transaction fees, leaving the system free of monetary inflation. The halving carries Bitcoin from the first state to the second.

The rule lives in Bitcoin's source code, maintained publicly at the Bitcoin Core repository. It executes as arithmetic when the block count reaches the right number, and changing it would require near-universal agreement across thousands of independent node operators.

How Does the Bitcoin Halving Work?

The halving executes automatically once the block count reaches the threshold, with no vote or software upgrade involved.

1. Miners produce blocks: They compete by running a hashing function called SHA-256 repeatedly, searching for a value that satisfies the network's current target.
2. Every node counts blocks: The subsidy is calculated purely from block height, meaning a block's position in the chain.
3. At each 210,000-block interval, the subsidy halves: Block 210,000 cut it from 50 to 25 BTC. Block 840,000 cut it from 6.25 to 3.125 BTC.
4. The span between halvings is called an epoch: Bitcoin has been in its fifth epoch since April 2024.

Because the subsidy derives from block height and not from a calendar date, halvings do not land on fixed days. They arrive when the blocks arrive, which is why every future halving date is an estimate.

The subsidy also does not shrink forever. Bitcoin's smallest unit is the satoshi, one hundred-millionth of a bitcoin, and the subsidy is stored as a whole number of satoshis. Eventually halving that number leaves less than one satoshi, and integer arithmetic rounds it to zero. Issuance stops, at zero, around the year 2140.

The Bitcoin Halving Schedule and Dates

Each halving cuts the block subsidy by 50%. Partial coins represent fractional rewards. The 2028 figure is projected, since halvings are triggered by block count rather than by calendar date.

The bitcoin halving dates below are exact for events that have happened and estimated for 2028, since halvings are triggered by block count rather than by calendar.

The fourth halving occurred on April 20, 2024 UTC, which fell on April 19 in US time zones, which is why sources differ. Issuance rates are measured at the moment of each halving and drift lower until the next one. Data as of September 18, 2026.

The absolute cuts shrink each time. The first halving removed about 3,600 BTC per day from new supply. The 2024 halving removed about 450. The 2028 halving will remove about 225. Each event is, by design, half as significant as the one before it.

This is what makes bitcoin disinflationary: new supply keeps arriving, but at a rate that falls on a fixed schedule. Bitcoin's issuance rate now sits below gold's. New bitcoin enters circulation at roughly 0.8% per year. Using World Gold Council data on above-ground stock and annual production, new gold enters at roughly 1.6% per year. Since April 2024, bitcoin's supply has grown more slowly than gold's, which is the threshold the halving schedule was built to cross.

Gridlines double at each step, so every halving is an identical drop. Bitcoin's issuance rate fell below 1% at the April 2024 halving and now sits near 0.82%, roughly half the rate at which new gold enters the market.

More than 95% of all bitcoin that will ever exist has already been mined. The rest trickles out over the next century.

What the 2024 Halving Did to Mining Economics

The revenue cut lands on real businesses immediately and in full, which makes mining the clearest place to observe a halving's effects.

Mining revenue is best tracked through hashprice, the revenue a miner earns per unit of computing power per day, usually quoted in dollars per petahash per second per day. Hashprice folds four variables into one number: the bitcoin price, the block subsidy, total network competition, and fee income. It is the standard measure of whether mining is viable.

Hashprice, hashrate and difficulty figures are from the September 6, 2026 retarget at block 965,664.

Hashprice spent much of 2026 between roughly $29 and $33, then climbed 22% in early September as bitcoin's price recovered, reaching $39.63. Miners hold a fixed cost base in local currency against a revenue stream that swings with bitcoin, so a 22% move in a fortnight changes which machines are worth running.

Efficiency became the deciding variable. Mining hardware, known as ASICs, or application-specific integrated circuits, is rated in joules per terahash, or J/TH, meaning energy consumed per unit of work. Lower is better. Break-even electricity prices vary sharply by hardware generation:

Approximate figures at mid-2026 hashprice levels.

An operator running older hardware needs electricity at roughly half the price of one running new machines to reach the same break-even. Electricity can account for up to 90% of operating costs, so that gap determines who stays online. Falling hardware prices, from roughly $20 to roughly $10 per terahash since 2020, have partly offset the hashprice decline for operators able to keep upgrading.

Reported production costs moved close to or above market price during 2026. Analyst estimates put full production cost for listed miners near $74,300 per coin in January 2026, with electricity alone near $46,400 per coin in March. With bitcoin trading in the $60,000s for much of the first half of the year, one widely cited estimate had listed miners losing roughly $19,000 per coin produced in March. These are estimates, they vary enormously between operators, and low-cost private miners sit in a very different position from listed companies carrying debt and depreciation. They still describe a sector under sustained pressure.

Block production concentrated. Thin margins push miners toward large mining pools, which smooth payouts by sharing rewards across many participants. By mid-2026 the Nakamoto coefficient for pools sat at 3, meaning three pools accounted for more than half of all blocks over a trailing month, with the largest near 27%. Protocol work such as Stratum V2 aims to move block template construction from the pool back to the individual miner, which would reduce coordination risk without changing who owns the hardware.

How Difficulty Adjustment Absorbs the Shock

Bitcoin has a built-in stabilizer that prevents a halving, or any sudden loss of mining capacity, from breaking the network.

Every 2,016 blocks, roughly every two weeks, the network recalculates how hard it is to produce a block. If blocks are arriving too quickly, difficulty rises. Too slowly, and it falls. The target never changes: one block roughly every ten minutes, regardless of how much computing power is connected.

The network saw 18 retargets between January and early September: eight increases totaling about 33% and ten decreases totaling about 45%, for a net decline near 12% across the period.

February 2026 ran the full sequence in a few weeks:

• Network hashrate had reached record highs above 1 zettahash per second, meaning one sextillion hashing attempts per second
• Difficulty climbed to roughly 144 trillion, a 15% jump in a single adjustment, the steepest since 2021
• A severe winter storm then hit Texas, where a large share of North American mining sits. Operators shut down to protect the grid, a practice known as curtailment, and hashrate fell to roughly 826 EH/s, the sharpest drawdown since China banned mining in 2021
• With less computing power connected, blocks slowed, and the following adjustment cut difficulty by double digits
• Profitability for the miners still running improved immediately, and hashrate recovered

Over the full year, difficulty fell from its October 2025 record near 156 trillion to roughly 126 trillion by late August 2026, a decline of about 19%, before ticking back up to 127.45 trillion on September 6.

That feedback loop is why individual miner failures do not cascade into network failure. When the least efficient operators switch off, the remaining ones become more profitable automatically. The system settles at an equilibrium where the marginal miner runs near break-even.

Difficulty adjustment protects the network's block schedule. It does not protect individual miners, and it does not protect the dollar value of the security budget.

The Halving Turned Miners into Data Center Companies

Facing a subsidy cut, record competition and rising power prices, mining companies found that their most valuable asset was access to large quantities of energized electrical power, along with the buildings, cooling and grid connections around it. Artificial intelligence workloads need exactly that, and AI customers pay in dollars on multi-year contracts.

The scale of the shift, drawing on industry and analyst estimates:

• Disclosed AI and high-performance computing contracts across listed miners rose from roughly $43 billion in late 2024 to more than $70 billion by early 2026
• Analysts project AI and HPC could reach around 70% of revenue for miners that execute their contracts by the end of 2026, up from roughly 30% in late 2025
• Companies with announced AI or HPC strategies include IREN, Core Scientific, TeraWulf, Riot, MARA, Bitfarms, Hut 8, Cipher Mining, Bitdeer and Hive
• Some funded the transition by selling bitcoin. Core Scientific sold 1,992 BTC, roughly $175 million, in March 2026

The trade-off became concrete on September 1, 2026, when Hyperscale Data switched off every bitcoin miner at its Michigan site to free 20 megawatts of capacity for an AI cloud customer. Hashing capacity was removed from the network in favor of contracted compute revenue.

Not every announced contract has produced revenue. A credible pivot requires energized power, buildings suitable for high-density computing, liquid cooling, fiber connectivity, a customer with real credit quality, and a balance sheet able to deliver on schedule. Visible HPC revenue on an income statement separates a genuine transition from a press release, and many companies have only the latter so far.

The shift also creates tension for Bitcoin itself. Miners selling bitcoin to fund construction adds supply to the market, and power redirected from hashing to AI reduces the computing capacity securing the network. Neither has proved damaging so far, with hashrate holding near record levels through 2026 and briefly touching 1,001 EH/s in early September. Both cut against the assumption that miners are structurally aligned holders of bitcoin.

The Security Budget Question

Halvings raise a long-term question that has no settled answer.

Bitcoin's security budget is the total value paid to miners for securing the network, consisting of the subsidy plus fees. A larger budget makes attacks more expensive, since anyone attempting a 51% attack, meaning seizing majority control of mining power to rewrite recent transactions, must out-spend every honest miner combined.

The protocol's design assumes fees gradually take over as the subsidy shrinks. Two years into the current epoch, that transition has not begun:

• The subsidy accounts for roughly 98% to 99% of miner revenue
• Transaction fees accounted for around 0.43% of miner rewards over a 24-hour period in early September 2026
• Typical next-block fee rates through 2026 ran near 2 satoshis per virtual byte, with 1 sat/vB acting as a floor
• At current issuance and price, the subsidy is worth roughly $34 million per day, or close to $12.5 billion per year

Fees are not rising to meet the subsidy. They are sitting near the minimum, with the mempool, the queue of unconfirmed transactions waiting for inclusion, rarely congested enough to force competitive bidding.

The Scale of the Gap

The fee level needed to replace the subsidy outright depends almost entirely on how many transactions share the load.

Approximate, at mid-September 2026 prices. The figure scales with bitcoin's price, since a higher price raises the dollar value of the subsidy being replaced.

At Bitcoin's current base-layer throughput, replacing the subsidy with fees would require transaction costs that price out ordinary use. At much higher throughput, whether through protocol changes or settlement layers built on top, the same total revenue comes from far more transactions at a few cents each. The security budget question is therefore closely tied to Bitcoin's capacity to process transactions.

Both Sides of the Argument

The disagreement turns on whether fee demand can grow enough to matter before the subsidy becomes negligible.

Those treating it as a real problem note that the subsidy halves on a fixed schedule while fee demand grows on no schedule at all, and that a shrinking dollar security budget eventually lowers the cost of attacking the network. A working paper circulated in May 2026 observes that the period since April 2024 provides the first genuine sample for testing whether fees can substitute for subsidy, and reads the early evidence as discouraging.

Those treating it as overstated, including several prominent Bitcoin developers and commentators, argue that the subsidy was never what secured Bitcoin, that attack costs depend on far more than the flow of miner revenue, and that more than a century remains for fee markets and scaling layers to develop. Some view the concern as an argument advanced in bad faith by promoters of competing systems.

Adoption Reshaped the Fee Market

A second-order effect connects directly to Bitcoin's success as an investment asset.

As bitcoin has been adopted primarily as a long-term store of value, held in custody and increasingly through investment products, its use for routine everyday payments on the base layer has declined. Fewer routine payments means less competition for block space, which means lower fees, which means a thinner fee market. The adoption pattern that strengthened bitcoin's investment case has weakened the revenue stream intended to secure it once the subsidy is gone.

The Next Bitcoin Halving in 2028

The fifth halving will occur at block 1,050,000, currently expected around April 2028, when the block subsidy falls from 3.125 BTC to 1.5625 BTC. Daily issuance drops from roughly 450 BTC to roughly 225, and the annual issuance rate falls to around 0.4%. As of early September 2026, with block height near 966,000, the network sits roughly 84,000 blocks away.

Based on the pattern of the previous four:

• Another margin compression event for miners, with half the subsidy revenue removed overnight and absorbed by whoever has the cheapest power and most efficient hardware
• A round of hardware retirement, followed by a downward difficulty adjustment that restores profitability for those still running
• Further consolidation, and probably further diversification into non-mining compute work
• A smaller supply effect than any previous halving, since 225 BTC per day is a small figure against daily trading volume

Financing conditions will matter more than in previous halvings, because much of the sector's AI and data center buildout is debt-funded. On September 16, 2026, the Federal Reserve raised its target range by a quarter point to 3.75% to 4.00%, its first increase since July 2023, with projections pointing to another before year end. Higher borrowing costs raise the hurdle for operators funding construction on credit. Policy statements and projections are published by the Federal Reserve.

What the Bitcoin Halving Does Not Do

• It does not change the 21 million supply cap. The cap is fixed. The halving controls the pace of reaching it.
• It does not reduce anyone's holdings. A halving changes nothing about your balance. Only the reward for producing new blocks changes.
• It does not directly change transaction fees. Fees are set by competition for block space, independently of the subsidy.
• It does not change block times. Difficulty adjustment handles that, and the ten-minute target is unaffected.
• It does not guarantee a price increase. The year following the 2024 halving was the first post-halving year in Bitcoin's history to close negative.
• It does not require you to do anything. There is no action to take, no upgrade, no migration.

Conclusion

The Bitcoin halving is the automatic 50% cut to the block subsidy that occurs every 210,000 blocks, and it is the mechanism carrying Bitcoin from an inflationary launch phase toward a fixed supply of 21 million coins. Four have occurred, the current subsidy is 3.125 BTC per block, and the next is expected around April 2028.

Two years after the most recent halving, its clearest effects have shown up in mining. Hashprice has fallen well below 2020 levels, the sector has redirected tens of billions of dollars of infrastructure toward artificial intelligence workloads, and transaction fees remain far too small to replace the subsidy that currently pays for Bitcoin's security.