Theses›Software & applications›Agentic payment rails

Agentic payment rails

A standalone neutral netting/clearing layer for agent micro-payments does NOT survive as an independent business; netting is absorbed by the rails and card networks by 2027 (contested: the original bet has inverted)

  • Shut
  • $29B by 2035

The call, a draftCard networks, stablecoin issuers and payment protocols are absorbing netting for agent payments, and the demand it would serve is shrinking. A standalone clearing layer is losing, and the window reads structural.

D1D2R1R2R3R4C1C2foragainst D1D2R1R2R3R4C1C2foragainst
Fig. 1The balance. Drivers and catalysts against restraints, each block sized by its weight; catalysts hang dashed because they have not happened yet, sized by their odds. It tips toward the restraints.
  1. D1Forecasts put agentic commerce at $3-5T by 2030, and netting would cut settlement counts by 80-95% if that flow ran on per-payment settlement.
  2. D2Each incumbent nets only inside its own network, so netting across competing rails is a seat no single participant can easily fill.
  3. R1Mastercard launched Agent Pay for Machines on 10 Jun 2026 with more than 30 co-builders, offering guaranteed settlement across cards, accounts and stablecoins.
  4. R2x402 volume fell 85% from April to July 2026 to $238k a month, while the average payment fell to about 4c.
  5. R3Clearing is a licensed activity, and the likely absorbers, Circle and the card networks, already hold the licences.
  6. R4A fee of 5-10 basis points on a plausible $720M a year of real agent flow by 2030 comes to under $1M.
  7. C1Organic agent payment volume reaches $1bn a year, with less than a quarter of it wash or test traffic.31 Dec 2026
  8. C2A dedicated agent micropayment netting company reports material live clearing volume from third parties.31 Dec 2027
Rests onthe beliefs underneath
Supported by · confidence
What would prove itdated tests, open and resolved
20272028today12345predictionswatch 20272028today12345predictionswatch
Fig. 2Each ring is a dated test: dashed while open, struck through when overdue, ticked or crossed once resolved.
  1. 1Agent-initiated stablecoin payment volume stays largely non-organic (>25% wash/test) and sub-$1bn annualised real volume through 2026.Dec 2026 · open
  2. 2Whether organic agent payment volume stays under $1bn a year and more than 25% wash.Dec 2026 · open
  3. 3An incumbent rail/issuer/network ships native agent-payment clearing/netting/guaranteed-settlement, absorbing the standalone-clearing opportunity.Dec 2027 · open
  4. 4Whether Agent Pay for Machines reaches general availability with broad participant coverage.Dec 2027 · open
  5. 5Any agent netting company reporting third-party clearing volume.Dec 2027 · open
Expressionslisted, private, technologies
Private · 10 private companies, 6 of them in the US.

AEON · Cycles · Catena Labs · Skyfire · Nevermined · Payman · Dhali · Floe Labs · SolvaPay · Nuvanté

Technologies
What changedthe record, newest first
  • 11 Aug 2026A netting layer I have been watching turns out to make its money from a cut of the yield on pooled collateral, not from a fee per transaction. That changes the question. The business then scales with float rather than volume, and it is also the bit most likely to make a regulator call you a custodian.
  • 11 Aug 2026Spent a morning actually measuring agent payments instead of assuming. Every rail I can measure is flat or shrinking, and the one number going up is how many agents get REGISTERED, which is free and proves nothing. Agents are being created faster than ever and paying for less. Worth noticing that the metric everyone quotes is the cheap one.
  • 6 Aug 2026A funded, audited clearing startup for machine payments put its contract on mainnet in May. Three months later it has moved seventeen dollars, all of it from its own team. The plumbing works; nobody has turned the tap. When you can check a claim on-chain, check it, and it is worth asking for the address.
3 earlier
  • 31 Jul 2026The whole x402 protocol has moved $15M in 14 months at an average transaction size of a fraction of a cent. The interesting problem is not who owns the clearing layer, it is what a clearing fee on a sub-cent payment can possibly be worth.
  • 18 Jun 2026The neutral cross-rail clearer's best defence against the incumbents holds up: each big network only nets inside its own walls, so the seat that nets across all of them is one no single competitor can take. And the first hard demand signal is showing up outside agent commerce, at the infrastructure layer (a Cloudflare-stablecoin LOI), not in the wash-traded agent volume.
  • 13 Jun 2026Everyone agrees agents will pay. Almost no independent analyst thinks the value sits in a neutral 'clearing layer' — they say the rail, the issuer, and identity. The agent-payments 'boom' is also still ~half wash-trading. Worth owning the listed rails before betting on the middle.
Reading
Show the reading