MPR-001REV D SCALE 1:1 · UTCSHEET 1 OF 1
DATA UNDER REVISION
FIG.01 — MACHINE-NATIVE PAYMENTS

How much do AI agents
actually spend?

DWG NO.MPR-001 · REV D
UNITUSDC · BASE MAINNET
PROTOCOLx402 / EIP-3009
INDEXED THROUGH-
LIVE FEED — FACILITATOR-CONFIRMED SETTLEMENTS · UTC-
( 01 )

Overview

LATEST INDEXED DAY · EXACT COUNT
( 02 )

Daily Settlements & Facilitator Share

FIG.02 — ALL EIP-3009 (EXACT) VS FACILITATOR-CONFIRMED (EST)
FIG.02
⚠ EIP-3009 fires for non-x402 gasless USDC too; facilitator share isolates confirmed x402. Pre-9/9 farm-distorted.
( 03 )

Estimated Settled Volume

FIG.03 — USD / DAY · SAMPLED EST · GATEWAY DEPOSITS EXCLUDED
FIG.03
( 04 )

Payer Economy

FIG.04 — WHO PAYS · TWO CALIBERS

ALL EIP-3009 (EXACT COUNT)

-
one-time payers-
repeat payers-
median ticket (sample)-

FACILITATOR-CONFIRMED X402 (SAMPLED)

-
one-time payers-
repeat payers-
median ticket-
p99 ticket-
TICKET SIZE DISTRIBUTION (FACILITATOR-CONFIRMED, LOG10 USD)-
( 05 )

Top Payees

FIG.05 — EST VOLUME SINCE AUG 27 · PATH CLASSIFIED
#ADDRESSIDENTITYPATHEST VOLSAMPLED
( 06 )

Research Note: the Sep 23 spike

NOTE.01 — 268,618 IN ONE DAY, 2.5× THE PREVIOUS DAY
-
SETTLEMENTS 9/23 (VS 9/22)
-
TOP PAYER SHARE OF DAY (EST)
-
FACILITATOR-CONFIRMED SHARE
RANKPAYERSAMPLED TXSEST DAY COUNT
⚠ Burst-settlement farm pattern, not organic commerce. Top drivers: unidentified.
( 07 )

Method & Data

APPENDIX — CALIBER, PROVENANCE, REPRODUCIBILITY
CALIBER — WHAT COUNTS AS AN X402 SETTLEMENT

Fingerprint: AuthorizationUsed events on Base USDC 0x8335…2913 (EIP-3009). This fires for every gasless USDC authorization spend — including non-x402 uses.

Two calibers shown side by side: ALL EIP-3009 (exact event count) and FACILITATOR-CONFIRMED — settlements whose submitting address is in the x402scan facilitator registry (51 Base addresses; Coinbase ×40, FluxA, Ultravioleta DAO, Primer, Polymer, Meridian; snapshot Sep 27 2026).

Excluded from x402 volume: Circle-Gateway-type infrastructure deposits (they also use ERC-3009 for gasless USDC onboarding). Out-of-scope: Hyperliquid and other venues — this radar indexes Base mainnet only.

SAMPLING & ESTIMATION

Counts and payer addresses are exact (full event replay, 2.6M+ events, no gaps, deduplicated by block range). Amounts and payee identities come from a stratified sample: 1,200 txs/day, decoded from tx input or receipt logs; daily volume = sampled sum × (day events / sampled events), single-ticket decode errors >$1M dropped (35 records).

Addresses we could not verify are marked unidentified — never guessed. Path badges: registered = submitting facilitator in registry; gateway = Circle Gateway-type infra; unregistered = active settlement path outside the registry.

KNOWN DISTORTIONS

Before Sep 8: farming/airdrop campaign inflated raw counts to 1M+/day (visible in x402scan aggregates). Sep 23 spike: burst-settlement farm pattern (see Note.01). Facilitator-confirmed share is the steadier signal on both.

DOWNLOAD & REPRODUCE

All figures recompute from public Base RPC endpoints. Rebuilt hourly from the chain; og card regenerates with each build.