← All guides

The Bitcoin halving for miners: what actually changes in the math

Costs & math · 9 min read

Most halving coverage is written for holders: supply, scarcity, price cycles. A miner's halving is a different and much smaller event. Exactly one number in the revenue formula changes, the block subsidy, and it changes by a factor of two, overnight, at a known block height. Transaction fees do not halve. Electricity bills do not halve. Difficulty does not halve; it follows hashrate on its own separate clock, with a lag that first deepens the squeeze and then partially relieves it. This guide keeps only the miner's ledger: the schedule in blocks, the before-and-after arithmetic for a fixed small hashrate, the fee cushion, the difficulty timeline, and the break-even identity a miner can test against their own power rate before the next halving. Nothing here is financial advice, and nothing here predicts a price, scenario prices, where they appear, are labelled arithmetic, not forecasts.

Exactly one number halves

A miner who produces a block collects the block reward, which has two parts. The subsidy is newly created bitcoin, awarded by the protocol. Transaction fees are attached by users to their own transactions to encourage inclusion, and are set by demand for space in a block. Subsidy plus fees is the reward, and only the subsidy halves. The halving is a programmed 50% reduction of the subsidy, executed automatically when the block count reaches the threshold: no vote, no committee, no software upgrade. Every full node computes the permitted subsidy from block height alone, in Bitcoin's source code it is effectively a halving by right-shift every 210,000 blocks, calculated in satoshis, and changing that rule would require near-universal agreement across independent node operators.

That precision matters for planning. Because the trigger is a block height and not a calendar date, and because blocks arrive probabilistically around a ten-minute target, every future halving date is an estimate that drifts by weeks. The block height never drifts. A miner plans against 1,050,000, not against a month in 2028.

The schedule, in blocks not dates

Block heightDate (past halvings exact)Subsidy afterNew bitcoin per day, network-wide
0 (launch)Jan 3, 200950 BTCabout 7,200
210,000Nov 28, 201225 BTCabout 3,600
420,000Jul 9, 201612.5 BTCabout 1,800
630,000May 11, 20206.25 BTCabout 900
840,000Apr 20, 2024 (UTC)3.125 BTCabout 450
1,050,000about Apr 2028 (estimate)1.5625 BTCabout 225

The per-day column follows from roughly 144 blocks a day at the ten-minute target: 144 × subsidy. Two things to notice. The absolute cut shrinks every era, the 2024 halving removed about 450 BTC a day of new issuance; the 2028 halving will remove about 225, because halving a smaller number removes less. And the schedule enforces the 21 million supply cap by construction, with issuance projected to reach zero around 2140 as the subsidy, halved in integer satoshis, eventually rounds to nothing.

A small miner's revenue, before and after

A miner's expected bitcoin per day, before pool fees and variance, is a share calculation: (own hashrate ÷ network hashrate) × blocks per day × block reward. The halving enters through the last term only. Work it for a fixed, deliberately small position, a miner holding 0.01% of network hashrate, with network hashrate itself held constant so the halving is the only moving part, counting subsidy only:

Before block 840,000After block 840,000Change
Subsidy per block6.25 BTC3.125 BTC−50%
Network issuance per dayabout 900 BTCabout 450 BTC−50%
Expected subsidy per day at a 0.01% share0.090 BTC0.045 BTC−50%
Electricity consumed per dayUnchangedUnchanged0%
Fiat cost per bitcoin producedBaselineDouble+100%

The last row is the one that decides who keeps mining, and it is pure division: the same electricity, rent, and maintenance now divided by half as many coins. Each halving cuts miner revenue from the subsidy by 50% in bitcoin terms while electricity and hardware costs hold steady. No efficiency improvement is being assumed away here, the table isolates the halving itself. What a real operation does about the doubled cost per coin is the subject of the last sections; the arithmetic of the problem is exactly this stark.

Fees: the part that does not halve

Total block revenue falls by slightly less than half, because the fee component is untouched by the halving rule. Put a labelled hypothetical on it: if blocks carried an average of 0.20 BTC in fees, a block's total reward would move from 6.45 BTC (6.25 subsidy + 0.20 fees) to 3.325 BTC (3.125 + 0.20), a fall of about 48.4%, not 50%. The cushion is real but its size is not the miner's to choose: fees are set by demand for block space, spike when demand spikes, and can sit near zero for long stretches. A documented snapshot makes the point about scale: in early September 2026, transaction fees accounted for about 0.43% of miner rewards over a 24-hour period. At fee shares like that, the cushion rounds to nothing and the halving lands at nearly its full 50%. Planning that assumes a fee rescue is planning on someone else's demand showing up on schedule, historically, the long-run direction of the protocol is toward fees carrying more of the reward as the subsidy shrinks toward zero, but "long-run direction" pays no electricity bill in any given month.

Difficulty runs on a different clock

The most common misunderstanding is that difficulty adjusts because of the halving, or that it halves too. Neither is true. Difficulty is a separate mechanism with a separate period: every 2,016 blocks, roughly two weeks, the network recalculates how hard producing a block should be, comparing how long the last 2,016 blocks actually took against the 20,160 minutes they should have taken at ten minutes each. Too fast, difficulty rises; too slow, it falls. The subsidy appears nowhere in that calculation.

The connection between the two mechanisms is behavioural, and it runs on a lag. The sequence after a halving:

  1. Halving lands. Subsidy revenue halves instantly; difficulty is whatever it was, so every miner's bitcoin output halves at unchanged cost.
  2. The marginal miners switch off. Operations whose cost per coin now exceeds their revenue, older hardware, expensive power, shut down, and network hashrate falls.
  3. Blocks slow. With less hashrate chasing the same difficulty, blocks arrive slower than ten minutes until the next retarget.
  4. The retarget relieves the survivors. Difficulty adjusts downward, and the miners still running automatically become more profitable: the same hashrate now claims a larger share of a smaller network. The system settles where the marginal miner runs near break-even.

Two practical consequences follow. The relief is partial and delayed, it restores only the share the departed hashrate vacated, arrives no sooner than the next retarget, and does nothing in the window between the halving and the first downward adjustment. And the stabilizer protects the network's block schedule, not any individual miner: difficulty adjustment is the reason miner failures do not cascade into network failure, and simultaneously the reason no miner can count on being rescued by it. A machine that is unprofitable at the halved subsidy and stays unprofitable after a plausible difficulty drop has received its answer from the arithmetic.

The break-even identity, and the pre-halving check

Combine the pieces and the survival condition is an identity, not an opinion. Fiat revenue per day equals bitcoin earned times price. The halving halves the bitcoin term (less the small fee cushion). Costs are unchanged. So, at unchanged difficulty, restoring the same fiat margin requires one of exactly four things: the price doubles, the scenario arithmetic is exact on this point, a doubled price returns per-block subsidy revenue to precisely its pre-halving level, which is a statement about multiplication, not a forecast; fees rise enough to close the gap; difficulty falls enough, through other miners leaving, to hand back share; or the miner's own cost base falls, cheaper power, or more efficient hardware producing more hashes per kilowatt-hour. There is no fifth term. Every post-halving strategy in the industry is one of those four, or a bet that one of them happens.

Run the check before block 1,050,000, while the current numbers are still the easy ones:

  1. Measure today's margin honestly. Bitcoin earned per day (pool dashboard, averaged over at least a full difficulty epoch) against all-in fiat cost per day, power, hosting or rent share, maintenance, pool fee.
  2. Halve the subsidy part. Recompute expected bitcoin per day with a 1.5625 BTC subsidy and current fees left as they are. That is the day-one-after figure at unchanged difficulty and price.
  3. Apply the relief you can defend. Model a difficulty fall only to the extent that clearly-marginal network hashrate would need to leave, and treat anything beyond a modest drop as hope, not a plan.
  4. Decide per machine, in advance. For each unit: runs profitably in the step-3 scenario (keep), runs only if price or fees rescue it (a speculation, label it as one, and decide what evidence would end it), or does not run (plan its sale or shutdown before the halving, when every other squeezed miner has not yet listed theirs).

What success looks like: before the halving block arrives, every machine in the fleet has a written day-one-after margin figure and a decision attached to it. A halving should be a date on which a plan executes, not the date the arithmetic is first done.

FAQ

Does the halving change how many blocks a miner finds? Not directly. Block production still targets roughly 144 a day, governed by difficulty. What changes is what each block pays in subsidy. A miner's share of blocks changes only when relative hashrate changes, theirs, or the network's through others leaving.

When exactly is the next halving? At block height 1,050,000, projected around April 2028, when the subsidy falls from 3.125 BTC to 1.5625 BTC and network issuance drops to about 225 BTC a day. The date is an estimate because block times drift with hashrate between difficulty adjustments; the height is exact.

Do transaction fees halve too? No. Only the subsidy is cut by the protocol rule. Fees are set by demand for block space and are the same the block after the halving as the block before, whatever demand happens to be that day.

Has hashrate always recovered after past halvings? The documented pattern is a stabilizer loop, not a guarantee: less efficient miners switch off, difficulty adjusts downward, and remaining miners recover margin, which is the mechanism by which the network has historically re-equilibrated. Individual operations that switched off do not thereby recover; the equilibrium is the network's, not theirs.

Related: How difficulty adjustments actually hit a small miner →