All numbers below are chain-derived via eth_getLogs against public RPCs. Nothing is self-reported, and the caveats section is not fine print — read it.
There's a lot of talk about the "agent economy" on Base right now, usually backed by big aggregate numbers. We run a paid-API platform that settles over x402, so we wanted ground truth, not vibes. We pointed an indexer at the settlement primitive underneath x402 — USDC's EIP-3009 AuthorizationUsed event — and counted everything for ~19 hours.
Method
- Contract: Base USDC
0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913 - Event:
AuthorizationUsed(address indexed authorizer, bytes32 indexed nonce)— emitted on everytransferWithAuthorization/receiveWithAuthorization(EIP-3009), the gasless-transfer mechanism x402 facilitators use to settle - Window: blocks 47,215,825 → 47,249,826 (34,003 blocks ≈ 18.9 hours at ~2s/block, June 11–12, 2026)
- Scan: incremental
eth_getLogsevery 30 minutes over public RPCs; per-event transaction receipts sampled (150 per window) for recipient/amount data
Headline numbers
| Metric | Value |
|---|---|
AuthorizationUsed events |
167,725 |
| Average rate | ~8,879/hour (~2.5/second) |
| Share from the single busiest wallet | 58.9% (98,777 events) |
| Share from top 5 payers | 68.3% |
| Share from top 10 payers | 72.0% |
| Share from top 500 tracked payers | 92.4% |
The busiest payer, 0x2b4e…9037, is a plain EOA (we checked eth_getCode — no contract). It fired 98,777 authorizations across the entire 19-hour window — a steady ~5,229/hour. That's relayer behavior, not a human. We found no public label for it; if you know what it is, comment below.
Activity distribution (top 500 tracked payers)
| Events per payer | Payers |
|---|---|
| over 1,000 | 11 |
| 101–1,000 | 48 |
| 11–100 | 401 |
| 2–10 | 40 |
So the shape is: a handful of industrial-scale relayers, then a long middle of wallets doing tens of payments in 19 hours.
The recipient side is where it gets interesting
We sampled transaction receipts to see where the money lands (448 distinct recipients in our sample):
| Recipient | Sampled payments | Total USDC | Avg per payment |
|---|---|---|---|
0xe903…1abf |
3,217 | $53.03 | $0.0165 |
0xa9dd…2315 |
609 | $1.22 | $0.0020 |
0x6a5e…c666 |
510 | $0.52 | $0.0010 |
0x6839…1a2b |
213 | $19.58 | $0.0919 |
0x7ce7…14ff |
182 | $346.25 | $1.9025 |
The top sampled recipient is being paid ~1.6 cents at a time, thousands of times. Numbers two and three are receiving sub-cent payments — $0.002 and ~$0.001 averages. Whatever you think of the agent-economy narrative, genuine machine-scale micropayment streams exist on Base today. We also saw the opposite end: single transfers of $3,400+ riding the same primitive (consistent with exchange/payment-processor gasless withdrawals).
Caveats — read these before quoting the numbers
-
EIP-3009 ≠ x402.
AuthorizationUsedfires for any gasless USDC flow: Coinbase Pay, exchange withdrawals, payroll relayers, x402 facilitators. These numbers are an upper bound for x402 settlement on Base USDC, not a measurement of it. - ~19 hours is a short window. This may not be representative of weekly or monthly patterns. We'll keep counting.
- The payer table is capped at the top 500 by event count. The long tail of one-off payers is undercounted; 7.6% of events are outside the tracked set, and the true distinct-payer count is unknown to us.
- Recipient amounts come from sampled receipts (150 per 30-min window), not an exhaustive join. Treat recipient totals as indicative, not complete.
- We could not identify any of the top addresses. All three we checked are EOAs with no public labels.
Reproduce it
from web3 import Web3
w3 = Web3(Web3.HTTPProvider("https://mainnet.base.org"))
USDC = "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
topic = "0x" + w3.keccak(text="AuthorizationUsed(address,bytes32)").hex().lstrip("0x")
head = w3.eth.block_number
logs = w3.eth.get_logs({"address": USDC, "topics": [topic],
"fromBlock": head - 1800, "toBlock": head}) # ~1 hour
print(len(logs), "gasless USDC payments in the last hour")
Who's counting
We're LogicNodes — deterministic API tools priced per-call over x402. In the spirit of this post: our own external revenue to date is $0.00, published live at logicnodes.io/transparency with the same chain-derived methodology (machine-readable, live ecosystem stats). We'd rather count honestly and earn the first dollar in public.
— Conner (DENNEYTRADINGCO LLC, AZ)
Top comments (0)