The security budget and what happens as the subsidy falls

A network's security budget is the flow of value paid to whoever produces blocks, per unit of time. It is the practical answer to how much an attacker would have to outspend, and on chains with a capped supply it is designed to shrink in native units as the block subsidy falls. What replaces the subsidy is transaction fees, and the honest statement about that transition is that its adequacy is unknown rather than assured.

Defining the budget without hand-waving

The budget has two observable components and one partially hidden one. New issuance is fixed by the protocol's emission schedule and can be computed exactly in native units. Transaction fees are recorded on chain and can be summed, though on chains where part of the fee is destroyed the recorded total is not what producers receive. MEV is income from ordering, only some of which appears in public data, so any budget figure understates the total by an unknown margin.

Denomination is the second definitional trap. Measured in native units, the budget is a schedule and is almost perfectly predictable. Measured in another currency, it moves with price and can change by a large factor without a single protocol parameter changing. Both are legitimate views and they answer different questions: the native figure describes what the protocol promises to pay, the converted figure describes what that promise buys in the market for electricity or capital. When a security budget is quoted without saying which is meant, the number is close to uninterpretable.

The subsidy falls by design

Bitcoin's halving cuts the per-block subsidy in half every 210,000 blocks, roughly every four years, and the sequence terminates at a maximum supply. This is not a market outcome or a policy decision that can be revisited casually; it is the rule the network was launched with and the reason its issuance is described as disinflationary. Supply against maximum shows how much of the eventual total already exists, and annual issuance rate shows the current pace of dilution.

The arithmetic consequence is straightforward. If the subsidy halves repeatedly and fees do not grow, the flow paid to producers declines in native units, and with it the cost of assembling attacking capacity. Whether the converted value declines depends on price, which is outside the scope of any factual description and is not something this site forecasts.

Fees as the replacement, and why the substitution is not smooth

A fee market is a market for scarce block space. When demand is high, users bid up inclusion and fee income rises; when demand is low, income falls, and it falls without floor. That variance is qualitatively different from a scheduled subsidy: it is a demand-driven flow that can collapse for extended periods, and security systems generally prefer predictable funding.

Fee mechanism design also determines who gets the money. Under EIP-1559, introduced to Ethereum in 2021, the base fee adjusts algorithmically to target a given block fullness and is burned, while only the priority fee reaches the producer. This makes fees a better price signal and reduces supply growth, and it also means that a period of heavy usage does not fund security in proportion to what users pay. Burn and fees should therefore be read as separate lines rather than added together. Token burn benefits holders through reduced supply; it does not pay validators.

Proof of stake changes the funding source, not the arithmetic

Under proof of stake the budget still has to come from somewhere, and the somewhere is again issuance plus fees. The difference is what the payment secures: capital that is bonded and slashable rather than machines that are running. This changes the attack calculus — an attacker must acquire stake, and a successful attack destroys it — but the funding question is unchanged. Issuance dilutes holders; fees are paid by users; burns offset issuance. Net issuance reports the balance of the first and third, and real inflation rate nets supply growth against destruction so that the two are not double counted.

Net issuance near zero or below is often described as a supply property, which it is. It is not a statement about the security budget, because the burned portion never reached producers. Keeping those two readings separate is the single most common correction needed when this data is discussed.

What a security budget does not tell anyone

It is not a valuation input, and no defensible ratio converts it into one. The relevant security question is comparative: the cost of mounting an attack against the value that a successful attack could capture, where the second quantity includes deposits at exchanges, positions in lending markets, and bridged assets, and is genuinely hard to measure. A network with a large budget securing an even larger set of extractable positions can be worse placed than a smaller one, which is why absolute budget figures rank networks poorly.

It is also not linear in safety. Doubling the flow does not halve the risk of a 51 percent attack, because the constraint may be capacity availability, hardware supply, or the willingness of counterparties to accept the proceeds. And a chain whose budget is nearly all subsidy is in a different structural position from one whose budget is nearly all fees, even at the same total. Annualized fees against issuance shows that composition directly.

Composition, not level, is the reading that survives scrutiny. See supply for emission schedules, fees for the demand side, and methodology for how burned fees and MEV are treated in every figure named here.