Skip to content
GOLEMREACH
GolemreachThe x402 market, in one honest number

Measured · dated · agent-readable

The x402 market, in one honest number.

Every guide tells a creator to “just go per-call.” None of them show the settlement data. Here it is: over 30 days, 523 x402 services settled $197,149 — and 86% of that was one service. Read the number and its four implications before you stand your endpoint up.

The number, measured

FiatDock audit, measured 2026-08-12 · prior 30 days · independent
$197,149
Settled across 523 x402 services in 30 days
Total for the whole market, all of it
86%
From a single service (BlockRun, $170,244)
The other 522 split ~$27k
66
Services earned more than $10 all month
Out of 523 listings measured
$153
The service with the widest buyer base earned
1,460 distinct paying wallets — the most reach, not the most money

Source: FiatDock, “How to monetize an MCP server” — fiatdock.com/monetize-mcp-server.html (published 13 Aug 2026, numbers measured 12 Aug 2026). FiatDock is the independent marketplace that performed the audit; it reports this even though it does not flatter the model.

Read those together

The two numbers that cut against the hype

The service with the widest buyer base — 1,460 distinct paying wallets in a month — earned $153. The service that earned the most reached far fewer wallets and charged far more per call.

What this means: reach does not equal revenue, and revenue concentrates in one outlier. Cheap, broad coverage attracts many paying wallets but that many wallets still translate to very little money when the price is a fraction of a cent. A high per-call price does not conjure buyers who were not there.

The payment model is not the problem. A no-human-attached buyer can finish an x402 transaction that no subscription or invoicing flow allows. The problem is volume: $0.01 × 50 calls/day ≈ $15/month, which UsageBox — a metering vendor — independently calls “barely worth metering.”

In the same measured window, FiatDock’s own first-party tools took $0.14 from 2 paying wallets, and its third-party marketplace had no external buyer at all. Even the marketplace that did the audit is not exempt.

What it means for YOUR endpoint

Four concrete implications before you write a 402 handler
  • Per-call is for high-cost calls, not lookups. A cheap lookup at $0.01 needs 50 calls/day just to hit ~$15/month. If your tool does something genuinely expensive — a model call, a paid upstream API — price to cover it and expect fewer buyers, not more revenue. The observed real-world settling band is $0.002–$0.03.
  • Reach is a distribution win, not a revenue win. The widest-reach service still earned $153/month. Listing where agents can discover you builds mindshare, but do not budget on the resulting wallet.
  • Directories pay $0, marketplaces take 15–20%. Catalog placement is not a payment rail. If you want the money to come back to you, the findable-but-you-settle path (your own x402 endpoint + the 402 as the authority) matters more than the directory’s badge.
  • Uptime literally is the earnings. A paid call that your server does not answer settles nothing and is not a delayed payment — it is no payment. Per-call concentrates your whole exposure on your own availability.
We priced our own rails against this. Golemreach’s Agentmeter runs at $0.01/call / $0.01 per OG render with prepaid credit packs — priced as cheap convenience and distribution, not as a revenue engine. This page is the measured reality check we would hand any creator; it is why we do not promise per-call income.

Paid reports are paused

The research above remains available. Do not send payment for a report. For an existing order, contact us privately. Our current offer is a fixed-scope workflow repair.

Method & honest limits