The thesis behind the money
In a Fortune essay published July 20, Circle chief executive Jeremy Allaire made the argument driving a wave of infrastructure spending. An economy run by software agents, he wrote, must run on software money, software contracts, and software governance, or it cannot run at all. The logic is that an autonomous agent completing a purchase cannot pull out a plastic card or wait three days for a bank settlement window. It needs value that moves at software speed with programmable rules, and that requirement is why stablecoin issuers now see agentic commerce as the demand event they have waited years for, rather than a speculative sideshow tied to crypto trading.
The framing matters because it reframes stablecoins away from trading and toward payments. For most of the past decade, dollar-pegged tokens were plumbing for crypto exchanges. The pitch in mid-2026 is that they become the default settlement layer when a Gemini, Copilot, or ChatGPT agent buys something for a shopper, because programmable money lets the agent hold funds, authorize a spend inside preset limits, and settle instantly across borders. Whoever controls that rail collects fees on a rising share of commerce, and the size of that prize explains why the incumbents of traditional payments are moving into stablecoins rather than defending against them from the sidelines.
What actually shipped
The activity is concrete, not theoretical. Following passage of the GENIUS Act, which set federal rules for dollar-backed stablecoins, Circle secured approval to establish a national trust bank, giving its USDC token, redeemable one to one for dollars, a regulated home inside the banking perimeter. A separate coalition is forming around an open standard: a group that includes Stripe, Visa, BlackRock, and roughly 140 firms is preparing to launch a stablecoin called Open USD this year, with reserves held in a distributed structure rather than by a single issuer. That governance choice is aimed squarely at the objection that a stablecoin concentrates too much risk in one company's balance sheet.
The card networks and banks are not sitting still. Visa has launched a stablecoin services platform that reaches more than 200 million merchants, positioning the network to settle in tokens where it already handles cards. SWIFT has deployed a blockchain-based ledger, with 17 banks piloting tokenized cross-border payments through it. Visa chief executive Ryan McInerney told Fortune that agentic commerce will expand the network's addressable market and that stablecoins show significant opportunities. When the company most exposed to disruption by tokenized money describes it as market expansion, the industry has moved past the question of whether this rail gets built.
Why retailers are paying attention
The number that gets a CFO's attention is interchange. Every card transaction carries a fee, often around two to three percent, that flows to issuers and networks, and for a high-volume retailer that tax runs into the hundreds of millions of dollars a year. A proprietary stablecoin lets a brand settle a purchase for a fraction of a cent and keep the spread. Fortune's reporting notes that brands including Walmart, Amazon, and Starbucks could bypass credit card fees through their own tokens, converting a cost center into an owned rail. Starbucks already proved consumers will preload money into a branded wallet, and a stablecoin generalizes that model to any purchase.
The strategic upside runs deeper than fee savings. A retailer that issues its own settlement token controls the data trail of every transaction, holds customer balances as float, and can program loyalty and refunds directly into the money rather than bolting them on afterward. That is a meaningful shift in leverage away from the card networks that have sat between merchants and shoppers for decades. The catch is that issuing regulated money is a serious undertaking, and most retailers will consume stablecoin rails through a provider like Circle or Visa rather than stand up their own, which is precisely the dependency those providers are building the infrastructure to create.
The retailer calculus
For a commerce leader, the decision is not whether stablecoins arrive but how much of the stack to own. Building a branded token demands licensing, reserve management, and compliance capacity that only the largest retailers can justify, so the practical choice for most is which settlement partner to route agent-driven transactions through. That choice is consequential because the partner sees the transaction data, holds the float, and sets the terms for disputes and refunds on autonomous purchases. Picking a rail in 2026 is closer to picking a core banking relationship than choosing a payment gateway, and it carries the same switching costs once volume concentrates on one provider.
The competitive risk is being late. Retailers that wait for a single dominant standard may find that agent traffic has already routed around them to merchants who wired up stablecoin checkout early, because an agent optimizing for price and speed will prefer the seller whose settlement clears instantly and cheaply. The counter-risk is committing to a rail that regulators later constrain or that fails to reach consumer adoption. The prudent path is to run controlled pilots on more than one settlement provider now, measure real cost and dispute outcomes, and keep the option to switch before autonomous volume makes any single rail load-bearing for the business.
What still has to be proven
The open questions are governance and trust. A stablecoin is only as good as the reserves behind it and the clarity of who makes users whole when something breaks, which is why the Open USD group is emphasizing distributed reserves and why Circle pursued a national trust bank charter. Agentic commerce adds a second layer of hard problems on top: how an agent proves it is authorized to spend, how limits are enforced, and how a disputed autonomous purchase gets reversed. None of that is fully settled, and the first high-profile failure of an agent-initiated stablecoin transaction will shape regulation more than any white paper.
For enterprise buyers, the signal in this week's essay is that the settlement layer for AI-driven commerce is being poured now, while the standards are still wet. The firms setting terms today, from Circle and Visa to the Open USD coalition, are trying to become the default rail before agent traffic scales. Commerce leaders can hold off betting the business on any single token while still forming a clear point of view on tokenized settlement and getting a small live experiment underway. Treating stablecoins as a distant crypto story rather than a near-term payments decision is the posture most likely to leave a retailer reacting to a rail it did not help design.
