A developer went looking for the demand behind x402 — not the headlines — and pulled the full Coinbase CDP Bazaar discovery catalog. The result, with raw data and scripts published so anyone can check: the entire independent x402 seller economy grosses $11,748 a month, across 14,128 registered services. When re-measured 16 days later, the catalog grew 8% while the money fell 13%.
Strip out the artificial volume and what's left is a market smaller than one senior engineer's salary. That is the honest starting point for anyone building on agent payments.
The methodology is unusually clean: the CDP Bazaar exposes, per service, a 30-day call count and a unique-payer count. Count them all and the economy becomes countable.
| Metric | 7 Aug 2026 | 23 Aug 2026 |
|---|---|---|
| Registered discoverable services | 14,128 | 15,309 (+8.4%) |
| 30-day GMV, all sellers combined | $11,748 | $10,200 (−13.2%) |
| Services with ≥100 paid calls / 30d | 287 (2.0%) | — |
| Services with "real demand"* | 32 | 16 (−50%) |
*real demand = ≥20 unique payers AND ≥10 calls per payer. Sixteen services, out of fifteen thousand. That is 0.1% of the catalog carrying the actual economy.
Public numbers in the hundreds of millions are real, and they're measuring something different:
For someone building a service, the relevant number isn't the headline — it's the $11,748.
The most useful line in the whole analysis compares two shapes of service:
641 unique payers → 1,018 total calls (1.6 calls per payer) 422 payers → 55,372 calls (131 calls per payer)
The first service has six hundred and forty-one different agents that found it, paid once, and never came back. That's not a customer base — that's a tasting menu. The second is a business. Repeat usage is the only metric worth watching; revenue will flatter you, repeat usage won't.
Only 32 services clear the bar of "≥20 payers and ≥10 calls per payer." Twenty-two services have 300+ payers but fewer than 2 calls each — pure tourism.
The few services with real demand almost all do the same thing: resell premium APIs that an agent can't otherwise buy — People Data Labs, full-enrichment, search, scraping — on a per-call basis. The underlying business is arbitrage on account-creation friction: an autonomous agent has a wallet but no credit card, no legal entity, no ability to sign a SaaS contract, so it pays per-call for access that normally requires an account.
That's the one model demonstrably working — and it's also, in many cases, a breach of the upstream provider's terms. The author ruled it out for that reason.
Probing 350 registered endpoints directly:
| Result | Share |
|---|---|
| Returns HTTP 402 correctly | 73.4% |
| 405 (POST-only, probably fine) | 9.1% |
| 404 (dead route) | 7.4% |
| DNS / TLS failure (gone) | 4.3% |
| 200 OK with no paywall (accidentally free) | 2.9% |
| 500 | 1.1% |
About 13% are hard-broken, and another 3% are giving their service away by mistake. A catalog where a tenth of the entries are dead is a catalog where discovery and verification are the actual problem — not payment.
This is exactly why the discovery layer matters. When 15,000 services compete for $10,200 a month, the scarce resource isn't another payment rail — it's a way to find the few services with real repeat demand, verify they're actually live, and try them before paying. That's the catalog minia2a runs: every service is live-probed, and the honest numbers are the point, not a secret to be spun.
The agent-payment market is small and real. Find the services with actual repeat demand — and verify they're live — before you pay a cent.
Browse the catalog →