TRM Labs has delivered the most sobering dataset yet on the agentic commerce narrative, estimating that AI agents account for somewhere between 0.6 percent and 7.5 percent of the value flowing through Coinbase’s x402 payment protocol after anomalous flows are stripped out.
The blockchain intelligence firm examined roughly 52.7 million USD across 198.9 million x402 settlements processed by known facilitators on Base, Solana, and Polygon since May 2025. After screening out self-payments, bulk flows from one or two payers, and sellers with fewer than ten buyers, researchers were left with 25.62 million USD in what they classified as likely commerce. Within that filtered pool, the share attributable to AI agents remained remarkably small under every model TRM tested.
Why measuring agent payments is genuinely hard
The core problem, according to the report, is that x402’s payment flow does not require intelligence to execute. Launched by Coinbase in 2025, the protocol integrates payments directly into web requests: a buyer receives a price quote, signs a payment authorization, and a facilitator verifies it, submits the blockchain transaction, and covers the network fee. An ordinary script can follow that exact sequence without any AI involved. Scheduled jobs, load tests, and self-dealing wallets can all produce blockchain records indistinguishable from genuine machine commerce at the transaction level.
TRM’s approach was behavioral rather than technical. “The assumption is that a true agent explores across multiple services and products, while an address repeating the same price behaves more like a script hitting one service over and over,” the report explained. Researchers screened for facilitator-broadcast payments with varying amounts averaging under 1 USD, then applied a stricter test requiring that pattern to persist across months, combined with public agent registration or payments to multiple distinct sellers.
The firm was candid about the limits of its own methodology. Many deployed agents today are single-purpose tools that buy the same service repeatedly, which the exploration test would misclassify as a script. The modeling choice, TRM noted, “may understate the space.”
What the activity actually looked like
The on-chain history behind the numbers tells a story of experimentation rather than an autonomous economy taking root. In late 2025, activity included apparent meme-token minting and recurring payments to a single AI-analysis service. Volume then concentrated in one payment contract in early 2026, before AI-service payments returned through an agent-payment router around the middle of the year.
Across the entire period, USDC dominated settlement almost completely, accounting for 99.6 percent of settled value, or 52.47 million USD. The preference reflects the stablecoin’s role as the default denomination for machine-to-machine micropayments, where price stability matters more than speculative upside.
Industry keeps building ahead of demand
The measured reality has not slowed investment in agent payment infrastructure. Binance’s August Agent OS launch included an x402 payment layer, while Coinbase’s Base has targeted agent and payments startups through a 1 million USD accelerator program. In May, Amazon announced AgentCore Payments with Coinbase and Stripe, enabling AI agents to pay for online services using stablecoins.
That gap between infrastructure investment and observed usage is precisely why TRM framed its research as a calibration exercise. If agents are to transact at scale, the industry needs baselines that separate genuine autonomous commerce from scripted noise, otherwise adoption metrics become self-congratulatory fiction.
Compliance is the missing rail
Beyond measurement, TRM identified a more structural gap: assigning responsibility for what agents do. On-chain agent registries exist and let individuals declare ownership of an agent address, but the declaration is voluntary “and currently not utilized by the majority of participants.”
The firm called for three building blocks: accurate registration of agent ownership, counterparty reputation information that agents can check autonomously before transacting, and monitoring systems designed for high volumes of tiny payments rather than low volumes of large ones.
“The rail already works,” TRM concluded. “What is needed is accurate registration, counterparty reputation an agent can check on its own, and monitoring built for volume rather than value. Agentic commerce will need agentic compliance.”
A reality check, not a eulogy
The report lands at a moment when agentic commerce has become one of the most heavily marketed narratives in crypto, with x402 positioned as the native payment layer for an anticipated explosion of machine-to-machine transactions. TRM’s data suggests the explosion, if it comes, has not started yet.
For developers and investors, the takeaway is twofold. First, protocol transaction counts are a poor proxy for agent adoption, since scripts and self-dealing can inflate totals dramatically. Second, the compliance layer may prove as decisive as the payment layer, because enterprises will not delegate spending authority to autonomous wallets that cannot be attributed, audited, or held accountable.
The 0.6 to 7.5 percent range is wide, and TRM admits even the upper bound could be conservative. But until registration becomes standard and behavioral analytics mature, the honest answer to how much AI agents are actually spending online remains: far less than the hype suggests, and nobody can yet say precisely how much.
Disclaimer: This article is for informational purposes only and does not constitute financial advice.
52.7 million filtered down to 25.6 million and agents are still under 8 percent of that. the agentic commerce thesis is carrying a lot of weight on vibes right now
wide band because most x402 flows were a couple payers washing volume. strip that out and barely anything is left, thats the real story
The part about an ordinary script following the exact payment sequence is the real story. Nothing in x402 verifies intelligence, it just verifies a signed request.
0.6 to 7.5 percent of value after filtering. the agentic commerce thesis is running on vibes and self-payments lol
to be fair the protocol launched may 2025, still early. but yeah 8 percent is not the revolution the pitch decks promised
fair on timing but TRM filtered for actual commerce, not just protocol age. if agents were driving real demand youd see it even in a young protocol
a year and 198.9 million settlements is not early anymore. if agents mattered the share would trend up, not sit under 8
a year and 198.9 million settlements and the band still tops out at 7.5 percent. things that find product market fit do not stay flat for 12 months
counterpoint, api billing was never going to hockey stick in year one. but agreed, under 8 percent torches the q3 pitch decks either way
ruel has it, pm fit doesnt hide for a year. a 7.5 ceiling after filtering 52.7M down to 25.6M is a rounding error dressed as a thesis
flat for 12 months is the kill shot. even the optimistic model tops out at 7.5, and optimistic models never stay that polite
the strict test needed months of varying sub 1 dollar payments across multiple sellers lol. a patient bot could pass that eventually and the agents barely did
TRM actually putting numbers on this instead of quoting vendor decks is rare. 0.6 to 7.5 is a wide band but even the ceiling is tiny
funny part is the simplest x402 use case, paying for api calls, barely needs an agent at all. a cron job with a wallet does the same job
thats the part nobody wants to say out loud, half these agents are cron jobs with a system prompt
cron job with a system prompt is generous, half of them are a bash script and a prepaid card. trm counting wallets instead of press releases is the real innovation here
exactly, and the trm data basically proves it. real agent demand would show long tailed distinct wallets, instead its a couple payers recycling the same volume
Agreed, payer concentration was the tell. Two wallets recycling volume through a protocol is a demo dressed as demand.
25.6M of likely commerce out of 52.7M settled is the real story here. half the volume evaporates once you screen for wash flows
x402 signs a payment inside a normal web request. clever design, but nothing about it requires intelligence, TRM got that right
exactly, x402 is a metering layer at heart. the agentic part was always marketing riding on top of an http 402 in-joke
52.7M filtered down to 25.6M is the number vcs will skip. half the headline was two payers recycling volume. brutal but necessary work by trm
TRM filtered 52.7M down to 25.6M and the pitch decks will still quote the 52.7 number tomorrow. watch
calling it now, some keynote this month says 50 million in agent payments and cites the unfiltered number. screenshot this
already happened, saw a thread this morning quoting 52.7M like it was all agent volume. the filtering appendix might as well be invisible
0.6 to 7.5 percent after 198.9 million settlements is a funeral for the agentic commerce thesis. base and solana volume and the agents still didnt show up
Half the volume filtered out as self-payments and bulk flows is the real headline. 52.7 million down to 25.6 million says more about x402 than the agent share does.