How miners earn revenue, and how the mix changes

The subsidy began at fifty bitcoin per block and halves every 210,000 blocks, which at the ten-minute target is roughly four years. The first epoch issued fifty; the second twenty-five; the third 12.5; the fourth 6.25; and the fifth 3.125. Each epoch issues 210,000 times its reward, so the sequence of epoch issuances is a geometric series converging on twenty-one million coins. The supply schedule page sets out the full sequence epoch by epoch.

The halving is the mechanism that makes the subsidy decline, and it is enforced by the same consensus rule that caps the coinbase. There is no discretion in it: the epoch boundary is a block height, and the permitted subsidy is a function of that height. A miner cannot choose to claim the previous epoch's reward, and no coordination among miners can restore it. The halving history page covers the events themselves.

Because the subsidy is denominated in bitcoin and paid in bitcoin, its value to a miner depends on the market price, which the protocol does not know and cannot influence. A halving halves the quantity of bitcoin a miner receives per block; whether it halves the miner's revenue in currency terms depends entirely on what the price does afterwards. The protocol guarantees the quantity and says nothing about the value.

Block space is scarce by construction. The block size limit caps how much transaction data a block can carry, and the ten-minute target caps how often a block arrives, so the network's throughput is a fixed quantity that does not respond to demand. When more transactions want confirmation than the next block can hold, they compete for the available space, and the way they compete is by bidding up the fee.

The result is that fee revenue is not a stable stream. In quiet periods a block may be found with very little fee attached, because there was no queue to outbid. In busy periods the same block may carry fees that rival or exceed the subsidy, because a large backlog of transactions is bidding for a small amount of space. The variation is driven by demand for settlement, not by anything in the protocol, and it can change within hours.

Miners select transactions by fee rate rather than by absolute fee, because the constraint is space rather than count. A small transaction paying a high rate per byte is more attractive than a large one paying a larger total but a lower rate. This is why fee estimation is expressed in satoshis per virtual byte, and why a transaction's cost depends on its size as well as its value. The selection rule is what makes the fee market work: miners are maximising revenue per block, and the transactions that pay most per unit of scarce space are the ones that get in.

The long-run direction is unambiguous. The subsidy declines by half at every epoch boundary and will eventually round to zero, while fee revenue is bounded only by demand for block space. Whatever the current proportions, the subsidy's share of total miner revenue must fall over time, and the fee share must rise, because one side of the ratio is scheduled to disappear and the other is not.

The path between here and there is not smooth. Fee revenue depends on demand, and demand has historically been episodic: long stretches of low fees punctuated by periods of intense congestion. A halving that cuts the subsidy in half during a quiet fee period reduces miner revenue sharply; the same halving during a busy period may be barely noticeable in currency terms. The schedule is deterministic, but the revenue it produces is not, because half of the revenue is a market outcome.

This is the central open question in Bitcoin's long-term design, and it is worth stating honestly rather than resolving. The security budget is the total revenue miners receive, because that is what pays for the hash rate that makes rewriting the chain expensive. If fee revenue does not grow to replace the declining subsidy, the security budget falls in bitcoin terms, and whether that is adequate depends on the price and on how much hash rate is needed to deter an attack. Nobody knows the answer in advance, and anyone who claims to is guessing. The mining economics page works through what that budget has to cover, and the issuance and inflation page shows how the subsidy's contribution to supply growth has already declined.