Is Bitcoin’s Fee Market Ready to Replace the Subsidy? A First-Principles Look

Every halving cycle, the same question resurfaces, usually stated as a settled fact in one direction or the other: will Bitcoin’s transaction fees eventually replace the block subsidy as miner revenue, funding the security budget once issuance approaches zero? We think this deserves a properly first-principles look, because both the confident “yes, obviously” and the confident “no, this is a fatal flaw” camps are making claims the current evidence does not fully support.

Why this question exists at all

Miners are paid two ways: the block subsidy (newly issued bitcoin, currently a fraction of what it was at launch and halving again roughly every four years) and transaction fees (paid by users for inclusion in a block). The subsidy is on a fixed, publicly known schedule converging toward zero over the coming century, with the overwhelming majority of it already issued well before then. As the subsidy shrinks, fees necessarily become a larger share of whatever miner revenue remains — the open question is whether that share becomes large enough, in absolute terms, to fund a security budget adequate to protect a network of Bitcoin’s value.

What the historical data actually shows

Looking at the actual history of miner revenue composition, transaction fees have typically made up a small share of total revenue — often in the low single digits as a percentage — with the block subsidy doing the overwhelming majority of the work. This share is not static, however: several episodes have seen it spike sharply. The 2017 congestion period, various 2021 demand surges, and the 2023–24 inscription-driven fee spikes all pushed the fee share into double digits, occasionally briefly exceeding the subsidy itself during the most acute moments.

The pattern across every one of these episodes is the same: a sharp rise, followed by reversion toward the prior low baseline once the specific driving demand (congestion, a particular on-chain activity fad) subsided. No episode observed so far has established a new, durable, higher floor for sustained fee revenue.

The two forces pulling in opposite directions

There are two structural forces at work, and they point different ways. On one side, layer-2 scaling — principally the Lightning Network — is explicitly designed to move routine, small-value payments off the base chain, settling many payments through occasional on-chain channel transactions instead. If Lightning succeeds at its stated purpose, it should reduce the number of routine payment transactions competing for base-layer block space, which works against fee growth even as the broader system succeeds on its own terms.

On the other side, base-layer demand is not limited to payments. Data-carrying transactions (inscriptions and similar use cases), large batched settlements from exchanges and layer-2 systems, and use cases that do not yet exist could all drive demand for scarce block space upward independent of anything resembling “more people making payments.” The 2023–24 fee spikes were driven substantially by exactly this kind of non-payment demand, which nobody had specifically forecast in advance.

What would actually resolve this question

This is not resolvable by argument alone — it requires data that does not yet exist, because the subsidy has never fallen anywhere near the level where this dynamic would be tested for real. The honest position is to track fee-share-of-total-miner-revenue at every halving as a genuine, ongoing data point, without treating any single reading — high or low — as proof of where the multi-decade trend ends up.

It is also worth separating two questions that often get merged: whether fee revenue rises enough in absolute dollar terms, and whether that’s enough security relative to what is actually needed to protect the network’s value at that future point. A smaller absolute security budget could be entirely adequate if the cost of attacking the network scales sensibly with what is being protected. We cannot resolve that jointly with the fee question, because both depend on the same unknowable future price and demand path.

Where we land

Unproven, in either direction, on the evidence available today. This is one of the genuinely open long-run questions in Bitcoin’s design, and treating it as settled — in the “obviously fine” direction or the “fatal flaw” direction — is asserting more confidence than four halvings of historical data can support. We will keep tracking the fee-share figure at every future halving, and we will keep saying so if the picture stays unresolved.