The 2028 Bitcoin halving: what it actually does to your miner

Every miner knows the halving is coming. Far fewer have worked out what it does to the specific machine sitting in front of them, and the answer is harsher than the usual "revenue halves" summary suggests.

Here is the number that matters, and almost nobody leads with it: the halving does not just halve your income — it halves the electricity price at which your machine breaks even. A current-generation miner that works today at 12.2 cents per kilowatt-hour will need power at roughly 6.1 cents to survive the same day after the halving. Most hosting contracts are not written at 6 cents. Most home tariffs are not within a factor of five of it.

This page works through the arithmetic properly, with the real network numbers, and is specific about which machines survive and which do not. Checked 12 September 2026.

When it actually happens

The date moves because block intervals vary. The network is currently running fast — about 146 blocks a day against the nominal 144 — which pulls the date slightly earlier. Nothing you can do changes it, but if you are financing hardware over 24 months, note that the halving lands inside that window.

Why "revenue halves" understates it

Miners earn two things: the block subsidy and transaction fees. Only the subsidy halves. So the honest question is how much of your income is fees — and right now, almost none of it.

At 0.6%, fees are a rounding error. When the subsidy drops to 1.5625 BTC, the total block reward falls from 3.14435 to about 1.58185 BTC — a 49.7% cut. Effectively the full 50%.

People sometimes argue fees will rise to fill the gap. They might, eventually — but the fee share would need to roughly double as a proportion of a halved subsidy just to keep your income flat, and nothing in the current data suggests that is happening. CoinShares' own reporting puts average fees at about 0.018 BTC per block, consistently under 1%. Plan for the 50% and treat any fee recovery as a bonus.

The break-even shift, machine by machine

This is the section worth reading twice. Break-even electricity price is simply daily revenue divided by daily kilowatt-hours. Halve the revenue and you halve the break-even.

Figures below assume Bitcoin near $77,700 and today's difficulty, so they isolate the halving effect rather than mixing in price guesses.

Read the right-hand column against what power actually costs. Riot Platforms — a gigawatt-scale operator with grid-services revenue behind it — achieved a net 3.6 cents per kWh in Q2 2026. That is roughly the level an S19k Pro will need in April 2028. An operator paying industrial rates of 5 to 6 cents keeps only the most efficient machines above water. Retail hosting at 6.5 to 9 cents, which is the normal US range, puts everything except the 9–11 J/TH flagships underwater on day one.

And every machine below about 13 J/TH today is already close to the line. Our hosting rate benchmark has the current numbers.

The part people forget: difficulty does not wait

The table above holds difficulty constant, which is generous. It will not be.

CoinShares models the network at 1.8 ZH/s by the end of 2026 and 2 ZH/s by March 2027, against roughly 0.92 ZH/s today. If that plays out, your share of the network halves again before the subsidy halves. Two independent 50% cuts, compounding, inside about 19 months.

That is not a forecast we are making — it is theirs, and forecasts are often wrong. But the direction is not in doubt: difficulty has risen roughly fivefold since 2021 and has never fallen for long. Any plan that assumes today's revenue per terahash persists to 2028 is wrong by construction.

What it costs to produce a bitcoin, before and after

Energy per bitcoin depends only on fleet efficiency — and the halving doubles it, because you do the same work for half the coin.

Now set that against what the listed sector already reports. Riot's fully-costed cost to mine in Q2 2026 was $90,631 per BTC against a production value of $71,667 — 126.5%. One of the largest miners in the world was already producing at a loss on a fully-costed basis, before the halving. CoinShares put the sector's cash cost near $80,000 and all-in near $96,100.

The implication is uncomfortable and worth stating plainly: at current prices, the 2028 halving is not survivable for a large part of the existing fleet. Either Bitcoin's price roughly doubles, or a meaningful share of hashrate switches off. Historically it has been some of both — which is why difficulty tends to stall or dip in the months after a halving, handing the survivors a partial reprieve.

For the full method behind these figures, see our guide to how long it takes to mine one bitcoin.

What actually survives

Three things decide it, in this order.

1. Your electricity price, by a mile. The gap between 3.6 cents and 9 cents is larger than the gap between the best and worst machine on the market. A 17 J/TH miner at 3 cents beats a 9.5 J/TH miner at 9 cents, comfortably, before and after the halving. If you only fix one variable, fix this one.

2. Efficiency, second. It matters — but it buys you a few cents of headroom, not a different category of outcome. Moving from 17 J/TH to 9.5 J/TH roughly doubles your break-even rate, which is the same order of effect as the halving itself. That is why the flagship machines command double the price per terahash: they are buying you one halving's worth of survival.

3. Hardware age, third. A machine bought today will be four years old in 2028 — around the end of the realistic working life we describe in our guide on how long ASIC miners last. If your payback runs past April 2028 on today's numbers, it will not be met.

What to do about it now

Price your payback to complete before April 2028. That is the single most useful rule. Nineteen months. If a machine does not repay its purchase price inside that window at your real electricity rate, you are betting on a higher Bitcoin price to bail you out, and you should know that is the bet you are making.

Do not buy old, cheap hardware on the strength of today's break-even. The 25–38 J/TH tier trades around $1 per terahash for a reason. Those machines break even near 3.5 cents after the halving, which is below what the largest industrial operators pay. Cheap per terahash is not cheap if it is scrap in 19 months.

Fix your power cost, not your hardware. If you are mining at domestic rates, no machine in this article — current or future — is profitable for you after April 2028, and most are not profitable now. The variable to change is the kilowatt-hour price. Our hosting locations exist for that, and the home versus hosted comparison runs both sides with numbers.

If you are choosing a machine today, the efficiency premium on the 9–11 J/TH tier is easier to justify than it looks, precisely because of the halving — see our guide to what is shipping in 2026–2027 for what that costs per terahash right now.

Run your own rate through the profitability model, then halve the revenue figure and look again. If it still works, you have a plan. If it does not, better to know in 2026 than in 2028.

Sources: mempool.space API for block height, difficulty, subsidy and fee statistics; Riot Platforms and MARA Holdings Q2 2026 SEC filings; CoinShares Bitcoin Mining Report Q1 2026 for sector cost and hashrate projections; Luxor Hashrate Index for hashprice. Network data as of September 2026. Halving date estimates vary by source because block intervals are not fixed. Break-even figures hold Bitcoin price and difficulty constant in order to isolate the halving effect; both will move. Mining income is never guaranteed.