USDC Settles 98.8% of AI Agent Payments

The machine-payments market has already picked its settlement asset, and the answer is boring. Data reported this week shows USDC clearing 98.8% of AI agent payment volume, a figure that quietly settles the running argument over whether autonomous software would transact in XRP, SOL or a stablecoin. Agents are not holding volatile assets between transactions; they are quoting, paying and settling in a dollar-pegged token. That is rational for a machine — a budget denominated in dollars cannot absorb a 10% drawdown — but it also means the emerging agent economy has a single issuer at its center.

The Numbers Behind the 98.8%

The figure, reported by 24/7 Wall St., covers observed AI agent payment activity across the protocols that make machine-to-machine settlement possible — the x402-style HTTP-native payment space, agent wallet platforms, and the emerging agentic commerce integrations built by exchanges and payment processors. USDC’s dominance is not a marketing outcome; it is a liquidity and tooling outcome. Circle’s token has the deepest integrations into the SDKs agents actually call, the most mature compliance story for merchants accepting machine customers, and a predictable dollar value that makes automated budgeting tractable. For any developer writing an agent that must pay for API calls, compute or data, USDC is the path of least resistance.

Where That Leaves XRP

Ripple has spent the week arguing the opposite case. Its XRPL starter kit now supports machine payments via the Stripe–Tempo protocol, and a separate integration wires XRP into Stripe’s agent payment sessions. The technical work is real. The commercial evidence is not: the XRPL integration is explicitly beta, and no volume data has been disclosed for agent-settled XRP transactions. When 98.8% of measured agent payments run through one stablecoin, a beta integration without numbers is a press release, not a market position. Notably, even Ripple’s own sessions reportedly run on XRP rather than RLUSD, which suggests the stablecoin angle of its stack is even further behind.

The Concentration Problem

A 98.8% share for a single stablecoin is not a healthy market structure, it is a dependency. Every layer of the agent-commerce stack — agent wallets, payment protocols, merchant acceptance — is effectively underwritten by one issuer’s redemption promise and blacklist policy. If that issuer freezes an address, throttles redemptions, or draws regulatory fire, the machine economy it clears does not fail over to a competitor; it stops. Circle’s USDC having won this market is not the same as the market being resilient. We noted earlier this month that x402’s claimed volumes deserve scrutiny; the denominator matters just as much as the 98.8% numerator here, and independent on-chain verification of what counts as an ‘agent payment’ remains thin.

What to Watch

Three things will test whether USDC’s share is durable. First, whether Solana-based agent activity — where SOL and USDC circulate on the same rails — shifts the mix as agent frameworks ship there. Second, whether Ripple publishes actual XRP agent-payment volume once its integrations leave beta; without it, the XRP narrative is unfalsifiable. Third, whether any large merchant or protocol operator deliberately multi-issuer their settlement to hedge concentration risk, as a regulated payments operator would with any single counterparty holding near-monopoly share. Until one of those moves, the working assumption for anyone building agent commerce is simple: the machines pay in USDC, and everyone else is fighting for the remainder.

USDC Settles 98.8% of AI Agent Payments

The Numbers Behind the 98.8%

Where That Leaves XRP

The Concentration Problem

What to Watch

Sources

Related reading