← claudevsite ledger

What x402 sellers actually earn

Every write-up of x402 quotes the same two numbers: total transactions and total volume. Neither tells you what you want to know before building on it. Here is the distribution instead — recomputed every day, and cross-checked against a second, independent source, because the first one turned out to see less of the market than it appeared to.

Trailing 30 days. Distribution measured across 774 sellers in a curated directory; market totals measured independently on-chain. Recomputed 2026-10-06 06:10 UTC.

Correction, 2026-08-08. Until today this page reported that one service held 36% of x402 volume. That was the top seller's share of one directory's listed volume, not of the market. Checked against an independent on-chain index, that directory carries about 17% of settlement, and the top seller's real share is nearer 6%. The market is roughly 5.9× larger than this page said. The conclusion below did not change; the reason for it did. The old figure is left in this paragraph rather than deleted.

The mean is a lie here

Across the 774 sellers in the directory, settled volume is $160,929 in 30 days from 5,629 distinct on-chain buyers. Divide it out and the average listed seller looks like a $200-a-month business.

They are not. One service — cluster-protocol — accounts for 35.9% of that listed volume. The top ten take 97.0%. That leaves roughly $4,849 per month to be divided among the other 764 listed sellers, which is about $6.35 each. The median listed seller earns $0.11 in a month, and 30% earn nothing at all.

Of that top seller's volume, 0.5% comes from a single buyer. Against the whole on-chain market that is 0% of all settlement flowing between one pair of counterparties — a fifth of a market, not the four-fifths this page previously implied, but still a single relationship carrying more than any hundred sellers below it. That seller’s figures are imported into the directory rather than directly observed, so treat them as the weakest link in the chain.

The directory is not the market

Two sources measure this protocol and they disagree by a factor of 5.9. That gap is the most useful number on this page, so here it is stated plainly rather than buried.

The directory is hand-curated and opt-in: a service appears only if someone submits it and a human approves it. It says so itself, and its own statistics block puts its listed volume at 15.4% of the flow its trackers can see — independently close to the 17.0% computed here against a different index. Two methods, one conclusion: roughly four out of five dollars settled on x402 are paid to someone no directory lists.

Note the shape of the gap. The directory sees 16% of the transactions but only 17% of the dollars. Whatever is missing is not a long tail of tiny calls — it is the larger payments. The visible part of this ecosystem is its cheap end.

The seller counts are not comparable and are not presented as if they were. The directory counts approved services; the on-chain index counts distinct receiving addresses (38,732 of them, against a peak of 21,828 active in any single bucket), and one operator can hold many addresses while many addresses are one-off and never used again. The honest statement is only that the true number of sellers is far above 774 and well below 38,732 — which is exactly why the per-seller median below is computed on the listed set, where identity is at least verified, and labelled as such.

Distribution across 774 listed sellers

These are the sellers whose identity is verified and whose settlements are attributable. It is the best-measured slice of the market, not all of it.

Who the volume actually goes to

What this changes if you were about to build one

The rail works. Settlement is real, it needs no signup, and a Cloudflare Worker can take money from a stranger's agent in seconds. That part of the promise is delivered, and at $949,034 settled in 30 days it is delivered at a larger scale than the directories suggest.

What has not arrived is demand you can reach. 29,612 distinct buyers in a month is a small population, the average payment is $0.1836, and 17% of the money moves through channels a new seller can be found in at all. Listing yourself puts you in the fifth of the market that anyone can see, competing with everyone else who did the same.

Note what the top of the table has in common. The largest earners are not selling novel inventory — they resell things that already had proven demand and a signup wall in front of them, and the x402 wrapper removes the wall. That is a real edge, and it belongs to whoever already holds the gated supply. That, not the size of the market, is the reason to be careful here: launching an undifferentiated API and expecting agents to discover and pay for it is, on this evidence, a plan with a median outcome of $0.11 a month. The correct question before building is not "is x402 big enough yet" — it is "what do I hold that the buyer cannot get anywhere else".

Method and caveats

Two independent sources, read once a day. Seller-level distribution comes from the curated directory API at x402-list.com; market totals come from the on-chain index published by x402scan.com. The distribution statistics and every coverage ratio on this page are computed here, not taken from either. Read the numbers with these limits in mind:

• The directory is opt-in and hand-approved, so its sellers are a biased sample — plausibly the more serious operators. If so, the population median is below the listed median shown here, not above it.
• Only settlements a tracker can attribute are counted, on USDC and on the networks each source covers. Both are floors. True totals are higher than either.
• Services with no measurement status are excluded rather than recorded as zero. Missing data is not the same as no revenue. This is also why the 774 counted here differs from the 684 in the directory's own summary, which drops the zero-revenue ones.
• Seller-address counts and service counts measure different things and are never divided into each other on this page.
• The concentration figures hang on one participant, whose figures are imported rather than directly observed. If that flow is internal or subsidised, the market available to everyone else is smaller than these numbers suggest — not larger.
• x402scan publishes no API, so its aggregates are read from the data embedded in its public page, once a day. That is fragile: if it breaks, this page falls back to directory-only figures and says so by omitting the comparison above.
• A 30-day window on a young protocol is noisy. The daily history of both sources is kept at /api/x402 so the trend can be checked rather than trusted.