Fee allocation and the security budget

BSAHI — where Bitcoin's fee revenue must eventually go. Produced: 2026-09-17 · Instrument: tools/research/fee_allocation.py

The two claims on one revenue stream

Bitcoin's transaction fees are the only long-run revenue. Today almost none of the competing claims is actually paid by fees — the block subsidy pays the security bill:

Rows 2 and 3 are the SAME cost at different horizons: L_net = N·C·T, so row 2 is row 3 multiplied by T. Their coverage percentages therefore differ by exactly T (30.2% x 10 = 301.8%) — they are one fact, not two. Reading the pair as "node operation is covered but permanence is not" is an error: the model has no separate permanence term, permanence IS the multi-year node cost. What the pair actually says is that fees fund roughly the first 3.0 of the 10-year commitment.

All rows share ONE fee basis — the measured 30-day mean ($1,411/block). The SCCR is a separate reading on its own live block window; quoting the two together without naming the window is a category error.

Actual fee revenue: ~$1,415/block (measured; frozen series) — against a $198,549/block production cost and a $4,676/block modeled storage cost (N=26,586). The subsidy ($212,517/block) is what keeps the security budget met today.

Miners are fee recipients; validators are cost bearers. That asymmetry is the externality. Miners are a few dozen pools running nodes; the ~26,586 reachable (and the unobservable non-listening) nodes are the ones carrying the storage and validation burden, and they receive nothing.

The halving clock

Subsidy halves every ~210,000 blocks. Only this arithmetic is exact:

When fees must take over

The subsidy alone stops covering the current network-wide energy cost of production in:

At the current price that crossover is 2028 — after which fees must fund security, and, if the storage externality is to be internalized, roughly $204,177/block more on top.

Temporal expansion — the fee requirement per halving epoch

Price paths are scenarios (grade C); the halving is arithmetic (grade A).

Specified price path (aggressive appreciation)

The structural invariant: when the price doubles each epoch, the USD subsidy is constant — the halving cancels the appreciation. On this path the nominal security protection plateaus at ~$117,188/block and does not grow again. The security-budget cliff is therefore a protocol invariant, not a pricing problem: no amount of fiat appreciation on a doubling path raises the subsidy above the plateau, while the production deficit and the storage claim are set by costs and by schedule.

Flat-price path (no appreciation)

Under a flat price the crossover arrives sooner and hardens, because the subsidy and the protection fall together.

Doubling-each-epoch from today

Machine-readable: data/fee_allocation.json → temporal.paths (per path, per epoch).

Grades

Caveats

• The multi-decade table is a scenario, not a forecast: only the subsidy halving is deterministic. Prices, energy prices, ASIC efficiency and the price of block space all move.
• N is provisional (an address sample, not a node count) — see the Verification Population Observatory. Re-basing N moves the storage claim linearly.
• The 241k/218k "node" figures are addresses, not nodes, and are not used here.
• Fees are network-wide revenue; node costs are network-wide costs. Single-node figures are not compared to network fees.

Why this matters

The SCCR asks whether fees cover the storage externality. The security budget asks whether fees cover hashing. Both are claims on the same fee stream, and the subsidy is currently masking both. As the subsidy falls, the two claims compete — and that is a research question worth owning.

Data: `/data/fee_allocation.json` · Related: SCCR sensitivity · Production cost · Evidence Matrix

← All research · ← Back to Learn