What is the Bitcoin halving? The issuance schedule, miner economics and the 2028 cut explained

Who this is for: Allocators, treasurers, founders, policy staff and serious retail readers who want to know what the halving does mechanically, what it has and has not done to price and hashrate, and how to read a miner’s unit economics.

Every four years or so, the Bitcoin network cuts the pay of the people who secure it in half, on a schedule written into the software in 2009 and never changed. The result is the most predictable supply curve of any major asset, and a recurring stress test for the mining industry that has to live on the other side of each cut.

It matters now for three reasons. The fifth halving is about eighteen months away. The miners that absorbed the 2024 cut have spent 2025 and 2026 redeploying power and land toward AI data centres, with lease commitments that dwarf what they earn from bitcoin. And the debate about Bitcoin’s “security budget”, what pays for security once the subsidy approaches zero, has moved from mailing lists into institutional research.

This guide covers the schedule, the four halvings so far and the date of the fifth, what happened to price and hashrate around each, miner economics and the pivot to AI hosting, the fee market and security budget, the cap and lost coins, and how institutions model supply, ending with a worked example at October 2026 figures.

The halving by the numbers

What the halving is and why Bitcoin has one

Bitcoin pays miners for each block they add to the chain. That block reward has two parts: the block subsidy, which is newly created bitcoin, and the transaction fees attached to the transactions in the block. The halving concerns only the subsidy. The protocol started it at 50 BTC per block in January 2009 and halves it every 210,000 blocks. At the target pace of one block every ten minutes, 210,000 blocks take just under four years.

The schedule is geometric. Each era adds half as many coins as the one before: 10.5 million in the first, 5.25 million in the second, 2.625 million in the third and so on. Summing that series gives 21 million, which is why the cap exists; it is a consequence of the halving rule and the starting subsidy, not a separate rule. Because the subsidy is stored in whole satoshis and rounded down, the final total is fractionally under 21 million, and the last satoshi of subsidy is expected around the year 2140.

The 2008 white paper says only that once a predetermined number of coins have entered circulation, the incentive can transition entirely to transaction fees. The design rewards the people securing the network while it is small, and lets anyone compute the exact supply at any future block height.

Difficulty adjustment keeps the clock honest

The halving is defined in blocks, not calendar time. Every 2,016 blocks the network re-targets difficulty so average block time moves back toward ten minutes; when hashrate grows, blocks arrive slightly fast until the next adjustment, which is why each halving has come a little ahead of a naive four-year projection. Average block time was 10 minutes 4 seconds in the week to September 28, 2026, per Hashrate Index.

All four halvings so far, and the fifth

Prices at the first three halvings are from Datawallet’s 2026 halving review; the 2024 price and all hashrate figures are approximations from exchange and explorer data, given to show scale.

Estimating the 2028 date

The chain tip was block 970,041 on October 6, 2026 (mempool.space), leaving 79,959 blocks to block 1,050,000. At exactly ten minutes per block that is 555 days, landing on April 13, 2028; at the 10 minute 4 second average of late September 2026 it is April 17, 2028. A working estimate is the second or third week of April 2028, earlier if hashrate outruns difficulty and later if it stalls.

Issuance on each side

Today the network issues 3.125 BTC about 144 times a day, roughly 450 BTC daily or 164,000 BTC a year. Mined supply at block 970,041 was about 20.09 million coins (19,687,500 at block 840,000 plus 130,041 blocks at 3.125 BTC), so annual issuance is about 0.8% of supply. After April 2028 it becomes about 225 BTC a day and 0.4% a year.

Price and hashrate around each halving: record and caveats

The popular narrative holds that each halving starts a bull market that peaks 12 to 18 months later. The pattern is in the data, and it is confounded.

The price record

After the 2012 halving at about $12, bitcoin crossed $1,000 within a year. After the 2016 halving at about $650, it reached about $20,000 by December 2017, roughly 18 months later. After the 2020 halving at about $8,800, it rose above $68,000 in November 2021, again about 18 months later (all per Datawallet, 2026). After the 2024 halving, the record came in early October 2025, roughly 18 months on, at about $126,000 according to Datawallet, with Hashrate Index’s 2025 review putting the high at $124,485.

Why three and a half data points prove little

Four halvings are not a sample. The 2012 and 2016 moves happened when bitcoin’s market value was a tiny fraction of today’s and the subsidy cut removed a far larger share of daily sell pressure. The 2020 halving coincided with the largest monetary expansion in modern history, which lifted nearly every risk asset. The 2024 cycle then broke the template twice: bitcoin set a record in March 2024, before the halving, for the first time, and the drawdown from the October 2025 peak was about 53% as of September 2026, versus 77% to 85% in prior bear markets, per the analysis CCS covered. On-chain analyst Willy Woo’s conclusion was that the halving cycle is giving way to a six-to-eight-year rhythm driven by debt and liquidity. Whether or not that holds, the ETF era added a demand channel earlier cycles lacked, and anyone modelling 2028 should treat the halving as one input among several.

The hashrate record

Hashrate follows margin: after each halving the least efficient machines turn off, difficulty adjusts down, and survivors earn more per machine until new hardware and higher prices pull hashrate back up. After 2024 the recovery was fast: the network entered what Hashrate Index called “the zettahash era” in September 2025, crossing 1,000 EH/s, and Datawallet put the peak near 1.16 ZH/s in October 2025. The pullback since has been about price, not the halving: bitcoin was down 9.6% year to date at $79,020 on September 14, 2026, and difficulty had fallen 5% to 127.17 trillion on July 11, 2026 at block 957,600, the fourteenth adjustment of the year. By September 28, 2026 hashrate was back to 952 EH/s and difficulty to 132.76 trillion.

Miner economics: hashprice, difficulty and margins

The most useful number for understanding the halving’s effect on miners is hashprice: revenue per unit of hashrate per day, quoted in USD per petahash per second per day. It is total block rewards (subsidy plus fees) times bitcoin price, divided by network hashrate. The subsidy cut halves the numerator; rising hashrate raises the denominator; price and fees move the numerator either way.

On September 28, 2026, hashprice was $39.87 per PH/s per day with bitcoin at $83,479 and hashrate at 952 EH/s (Hashrate Index). In BTC terms it was 0.0004767 BTC per PH/s per day and flat week on week, which says the USD move was all price. Hashprice had been $31.10 on July 11, 2026, when trade reports put it 37.2% below its October 2025 peak. The forward market implied an average of $38.34 over the following six months as of late September 2026.

Energy hashprice decides who survives

Because electricity is an ASIC’s main operating cost, it helps to restate revenue per megawatt-hour consumed. In the week to September 14, 2026, Hashrate Index put fleets under 14 J/TH at $133 per MWh, 14 to 19 J/TH at $97, 19 to 25 J/TH at $73 and 25 to 38 J/TH at $50. A miner paying $50 per MWh (5 cents per kWh) with 2021-era hardware was at breakeven before overhead; current hardware on the same contract had a wide margin. The halving does not kill mining; it kills the oldest hardware and the worst power contracts.

CCS’s April 2026 analysis of mining profitability put the practical breakeven near 4.5 cents per kWh, energy at about half of mining revenue, hashprice about 45% below its pre-halving peak, and industrial operators on 2 to 4 cent contracts as the ones still comfortably profitable.

The hardware treadmill

Hashrate Index’s 2025 review grouped ASICs into old generation at 25 to 38 J/TH, which saw steep price declines, mid generation at 19 to 25 J/TH, and machines under 19 J/TH, which held demand, and projected frontier efficiency near 7 J/TH in 2026. Each halving pulls the viability line forward by roughly one hardware generation.

How the shock absorber works

1. The subsidy halves; USD hashprice drops by close to half unless price or fees move.
2. Miners whose energy cost exceeds their energy hashprice switch off.
3. Block times lengthen until the next 2,016-block retarget lowers difficulty.
4. Survivors earn more BTC per machine; if price rises, curtailed machines return.
5. New hardware ships, hashrate passes its old level, and the cycle resets at a lower subsidy.

The 2024 to 2026 pivot to AI and HPC hosting

The fourth halving coincided with the generative AI build-out. Miners owned what AI developers needed and could not get quickly: energised sites with grid interconnection, transformers, cooling and permits. In June 2024 Core Scientific signed a 12-year agreement to host GPUs for CoreWeave; JPMorgan noted the 14 miners it tracked gained about $4 billion, or 22%, in market value afterwards and valued ready power at $105 million to $140 million per megawatt (The Block, June 24, 2024).

The major deals, checked against filings and company statements

Further leases involving IREN, Hut 8 and TeraWulf were reported by trade outlets through mid-2026 but had not been checked against filings at the time of writing, and only a fraction of contracted capacity had been energised.

What the pivot means for the halving

The marginal buyers of ASICs are in several cases winding mining down, which slows hashrate growth and supports hashprice for those who remain; listed miners are no longer a clean bitcoin proxy; and hosting contracts pay dollars over long terms but concentrate counterparty risk in a few AI firms and their guarantors. Hashrate Index’s 2025 review put it plainly: the AI pivots won and pure-play miners suffered. Policy adds cost too; CCS reported on March 31, 2026 on Washington’s push to cut Chinese-made machines out of US mining supply chains.

The fee market and the security budget debate

Bitcoin’s long-run plan is that transaction fees replace the subsidy. The question is whether fees can grow enough, fast enough, to keep the network expensive to attack as the subsidy heads toward zero.

Where fees stand

In the week to September 28, 2026, miners collected about 3,112 BTC in rewards, of which fees averaged 0.0217 BTC per block, about 0.70% or roughly $1.83 million for the week (Hashrate Index); the week to September 14 was 0.59%. Subsidy still pays for over 99% of security.

The spikes: Ordinals and Runes

Fees spike when a new use case competes for block space. The halving block itself is the best example: block 840,000 on April 20, 2024 was mined by ViaBTC, contained 3,050 transactions and paid 37.6256 BTC in fees, about $3.25 million at the time, against a 3.125 BTC subsidy (blockchain.com). The driver was the launch of the Runes token protocol at that block, after a year in which Ordinals inscriptions had already lifted fees. The fee market can clear at very high levels; it also reverts fast, since fee share fell back below 1% within months.

The argument

Paul Sztorc laid out the security budget worry in February 2019. While the subsidy is large, the network is secure almost by accident; as it shrinks, security depends on fee revenue that competition from cheaper chains and off-chain rails tends to suppress. He then put the annual security budget near $2.6 billion with fees contributing about $70 million. Counter-arguments: a higher price can offset the cut for several more halvings; demand for block space from inscriptions, tokens and layer-two settlement may grow; and a small tail emission could be considered, a proposal that remains deeply unpopular because it would break the 21 million cap.

The arithmetic: at 3.125 BTC and $83,479 the subsidy pays miners about $37.6 million a day; in 2028 the same price gives $18.8 million, and 2032 halves it again.

The 21 million cap and lost coins

The cap is enforced by every node that validates blocks: a block claiming more subsidy than the schedule allows is simply rejected. Changing it would need near-universal agreement from node operators, which is why every institutional model treats it as fixed.

Mined supply is not spendable supply. Some coins are provably unspendable, such as the genesis block’s reward, and a larger amount sits in wallets whose keys are presumed lost. The roughly 1 million coins mined in the first year and attributed to Satoshi have never moved. Estimates of permanently lost coins vary widely because dormancy is not proof of loss. The practical effect is that the real float is some millions of coins below the 20.09 million mined, which makes any movement of long-dormant coins a watched signal.

How institutions model supply

For an ETF issuer, treasury company or fund, the halving schedule is the one input in a bitcoin model with no uncertainty. Models therefore build around three quantities: scheduled issuance, which is fixed; free float, which is mined supply less long-dormant and institutionally locked coins; and demand flows, the only thing left to forecast.

The demand side changed after January 2024, when US spot bitcoin ETFs launched. By September 2026, per the analysis CCS cited, those ETFs held about 1.3 million BTC, around 6% of supply, and public company treasuries held more than 1 million BTC, for a combined 12% of circulating coins. Daily issuance of 450 BTC, about $13.6 billion a year at $83,000, is small against holdings of that size, which is the core of the argument that halvings matter less for price than they once did; after April 2028 the figure halves. CCS explains the funds in How spot bitcoin ETFs work, tracks their holdings on the ETF tracker, covers the corporate buyers in Bitcoin treasury companies explained, and covers how those coins are held in Crypto custody explained.

How we got here: a timeline

Worked example: one miner, before and after the 2024 halving

How to evaluate halving claims and miners: a checklist

• Is the claim about price or about supply? The halving is a certainty about supply and nothing more; any argument that moves from “issuance halves” to “price doubles” has to account for demand, liquidity and the ETF channel.
• What is the fleet efficiency in J/TH? It sets energy hashprice: in September 2026 a sub-14 J/TH fleet earned $133 per MWh against $50 for a 25 to 38 J/TH fleet. A miner that does not disclose it is hiding its most important number.
• What is the all-in power cost and how is it contracted? Fixed-price long-term power under 4 cents per kWh survives most scenarios; spot or curtailment-dependent power does not.
• Does the plan depend on fee growth? Fee share was 0.6% to 0.7% in September 2026. A model that needs fees at 10% of rewards by 2028 is a bet on a new use case, not a base case.
• Is the AI hosting revenue contracted, and with whom? Check counterparty, term, backstop and start date. Rent on Cipher’s AWS lease starts in August 2026; much of the sector’s contracted capacity is still under construction.
• Does the company hold or sell its mined bitcoin? A miner holding treasury bitcoin is a leveraged bet on price; one selling daily is a spread business on energy.

Risks and open questions

The first open question is the security budget. Fees under 1% of rewards in 2026 mean that, two more halvings out, network security depends on bitcoin’s price having roughly quadrupled or fee demand having grown by an order of magnitude. If neither happens, the debate about a tail emission moves from the fringe to the centre, and any such change would itself threaten the fixed-supply thesis that attracts institutional holders.

The second is hashrate concentration. If listed miners keep converting sites to data centres, security increasingly rests on private and offshore operators with less disclosure; Hashrate Index’s 2026 outlook expected hashrate to migrate out of the United States toward lower-cost jurisdictions, including Venezuela. A slowdown in AI capital spending would also hit pivoted miners from a direction unrelated to bitcoin.

The third is the cycle narrative itself. If much of the market still positions around an 18-month post-halving peak, that belief can be self-fulfilling for a while and sharply self-defeating when it breaks. The 2024 to 2026 cycle already deviated in timing and depth.

What to watch next

Glossary

Block subsidy

The newly created bitcoin paid to the miner of each block: 50 BTC in 2009, 3.125 BTC since April 2024.

Block reward

The subsidy plus the transaction fees in a block. Only the subsidy halves.

Halving

The scheduled 50% cut to the block subsidy every 210,000 blocks, about every four years.

Hashrate

Total computing power securing the network, in hashes per second; 952 EH/s in late September 2026.

Hashprice

Miner revenue per unit of hashrate per day, usually in USD per PH/s per day: total rewards times price divided by network hashrate.

Difficulty

The parameter setting how hard it is to find a valid block, re-targeted every 2,016 blocks to hold block time near ten minutes.

J/TH

Joules per terahash, a machine’s efficiency rating. Lower is better; current machines are under 19 J/TH.

Security budget

Total value paid to miners per period, which funds the cost an attacker would have to beat.

Why it matters

The halving is the clearest expression of what bitcoin is: a monetary system whose supply is a published formula rather than a committee decision. Its economic weight has shifted. In 2012 the subsidy cut removed a meaningful share of daily sell pressure from a thin market. In 2026, with issuance at 0.8% of supply and institutions holding more than a tenth of all coins, the halving matters less as a price catalyst and more as a stress test of two things: the mining industry’s ability to keep refreshing hardware on thinner margins, and the network’s ability to pay for its own security as the subsidy fades.

The 2024 to 2026 period answered the first question in an unexpected way, by turning the largest miners into AI landlords. The second remains open. For live data see CCS’s crypto prices page and research hub; for the operator’s view, CCS interviewed Marathon Digital’s Fred Thiel ahead of the 2024 halving in this June 2023 interview and NiceHash’s Filip Primec on hashrate marketplaces in this April 2026 interview.

Sources

1. mempool.space: Bitcoin chain tip height (970,041), October 6, 2026
2. Blockchain.com: Bitcoin block 840,000, April 20, 2024
3. SEC EDGAR: TeraWulf 8-K, Fluidstack lease and Google backstop, August 14, 2025
4. SEC EDGAR: Cipher Mining 8-K, third quarter 2025 results and AWS lease, November 3, 2025
5. Core Scientific: Core Scientific and AMD announce infrastructure partnership, July 28, 2026
6. Hashrate Index: Hashrate Index Roundup, September 28, 2026
7. Hashrate Index: Hashrate Index Roundup, September 14, 2026
8. Hashrate Index: 2025 Bitcoin Mining Year in Review, December 2025
9. TechInformed: Microsoft signs $9.7 billion IREN deal for Nvidia GB300 capacity, November 3, 2025
10. DCD: Hut 8 signs 245MW capacity deal with Fluidstack as part of multi-gigawatt partnership with Anthropic, December 17, 2025
11. DCD: CoreWeave Core Scientific deal ended, October 30, 2025
12. The Block: Investor interest surges for bitcoin miners following Core Scientific AI hosting deal, June 24, 2024
13. dWealth News: Bitcoin difficulty falls 5% to 127.17T in 14th reset of 2026, July 11, 2026
14. Datawallet: Live Bitcoin halving countdown (2028), accessed October 6, 2026
15. Paul Sztorc: Security budget in the long run, February 14, 2019
16. Crypto Coin Show: Is bitcoin mining still profitable in 2026?, April 13, 2026
17. Crypto Coin Show: Bitcoin’s 4-year cycle could be changing, Willy Woo reveals what could replace it, September 6, 2026

Disclosure: This guide is for education only and is not investment, legal or tax advice.