The headline number is doing a lot of work. In an interview with HackerNoon, Venom CEO Christopher Louis Tsu claimed that x402 — the payment protocol originally spun out of Coinbase’s developer stack — has already cleared 100 million agentic payments, and that the next billion blockchain users will not be human. It is a clean thesis: networks designed for machine customers, settling in stablecoins, at machine frequency. It is also a claim that deserves the skepticism the sector has lately earned. We wrote only days ago that blockchain intelligence firm TRM Labs found most x402 payment volume is not attributable to autonomous AI agents — which makes a nine-figure payments count a number worth interrogating rather than repeating.
What Is Actually Being Claimed
The HackerNoon piece is an interview, not an audited disclosure. Tsu frames x402’s 100 million agentic payments as evidence that machine-native commerce has moved from concept to volume, and argues that networks built for machine customers need different primitives than human ones: lower-friction authentication, programmatic payment negotiation, and settlement rails that assume no human is watching the transaction at all. None of that is controversial as architecture. What is missing is the denominator and the definition. “Agentic payments” is not a standardized term: a payment triggered by a script, a payment routed through an agent framework but initiated by a human, and a payment autonomously decided by a model can all be counted under the same label depending on who is counting.
Why the TRM Finding Still Hangs Over This
The friction with the 100 million figure is that it lands days after TRM Labs concluded that the majority of payment volume on x402 does not originate from autonomous agents. Both claims cannot be casually true unless the definition of “agentic” is doing something generous. The plausible reconciliation is boring: x402’s HTTP-native payment mechanism is genuinely easy to integrate, so it gets used for programmatic but human-directed payments — API monetization, automated top-ups, bot-mediated purchases — which are adjacent to agentic commerce without being it. That is still a real business. It is just not the same business as machine customers transacting without human intent behind each payment. Infrastructure narratives conflate the two because the second is worth more in a pitch.
The Machine-Customer Thesis on Its Own Merits
Strip out the number and Tsu’s argument is more durable. If agents become meaningful economic actors, they need wallets that are not Dodd-Frank-shaped, payment negotiation at protocol level, and per-transaction spending policies encoded rather than enforced by UI. Stablecoins are the obvious settlement asset because they are programmable, divisible to machine-scale amounts, and do not require an agent to pass a KYC flow designed for humans. This is the same territory that Coinbase Agent Payments, Ant International’s agent payment protocol, and Binance’s Agent OS are each approaching from a different direction — incumbent rails bolting agent support on, versus crypto-native rails assuming agents from the start. Venom is arguing for the latter camp.
What to Watch
The honest position is that agentic payment volume is currently unverifiable from the outside, and the parties with the strongest incentive to inflate it are the ones quoting it. Watch for third-party attribution: whether analytics firms can distinguish autonomous agent activity from scripted human-initiated flows at scale. Watch for repeat usage — real machine commerce means the same agents transacting continuously, not one-off integrations. And watch whether any of the claimed volume shows up in settlement demand for specific stablecoins, which is harder to fake than a payments counter.