Visa moved $4 trillion in a single quarter, cut 7% of its staff, and is building the stablecoin rails that could route around it

Every figure in this article is drawn from Visa's fiscal third-quarter 2026 results and earnings call, court filings and reporting on the merchant interchange settlement, or from CNBC and Payments Dive as attributed. Visa's fiscal year ends in September. Sections marked as analysis are identified as such.

In the three months to the end of June, $4 trillion was spent on Visa cards. Not in a year — in a quarter, for the first time. There were 72 billion processed transactions, up 10 per cent, and net revenue of $11.6 billion, up 14 per cent. Adjusted earnings of $3.32 a share rose 11 per cent.

On the same day, Visa announced it was cutting about 2,600 jobs — 7 per cent of its workforce — and taking a $563 million severance charge, with the savings redirected into artificial intelligence, cross-border and business payments, affluent customers, and stablecoins. It is an unusual combination: record throughput, a smaller company, and a large investment in the technology most often described as the threat to card networks.

The quarter

The headline volume number is the least interesting part, because Visa's volume grows more or less with global consumption and always has. The revealing line is value-added services, which reached $3.8 billion, up 34 per cent in constant dollars — roughly a third of total revenue. This is the fraud scoring, tokenisation, dispute handling, consulting and issuer-processing business Visa has spent a decade building on top of the network, and it is growing three times faster than the payments it sits on.

Cross-border volume excluding intra-Europe rose 12 per cent, the highest-margin traffic Visa carries. Visa Direct, its push-payments rail for payouts and remittances, processed $4 billion of transactions, up 21 per cent. Client incentives — the rebates Visa pays banks to keep their card portfolios — rose 18 per cent, a reminder that the toll booth negotiates with the road builders.

Chief financial officer Chris Suh addressed the question every consumer-exposed company has been asked this year, and gave the most emphatic answer of the season: "We do not see signs of the lower-spend consumer weakening in our volumes." Operating expenses rose 17 per cent, but nearly all of that was the severance charge.

Visa guided the full fiscal year to revenue growth at the low end of the low teens and earnings growth at the low end of the mid teens — a deceleration from the quarter just delivered, and the kind of guidance a company issues when it expects the comparisons to get harder rather than the business to get worse.

The stablecoin position

Visa launched a Stablecoin Platform this year for minting, moving and managing stablecoins on behalf of banks and issuers, and joined the consortium behind the OpenUSD standard. Ryan McInerney, its chief executive, has been careful about what that means: "Visa will remain multi-coin, multi-chain. Our role is not to pick winners. Our role is to help clients connect to the stablecoin ecosystem securely and at scale."

Read that sentence twice. It is not a claim that stablecoins will fail, and it is not a bet on any of them succeeding. It is a statement that whatever settles the transaction, Visa intends to be the layer that connects a bank to it — the same position it holds today between an issuer and a merchant acquirer, translated to a different rail.

McInerney was blunter about the other technology in the room: "Agentic commerce is a when, not an if." Visa has deployed more than 150 internal AI applications and is building credential standards for software agents that transact on a cardholder's behalf.

The agentic point is the more immediate of the two. If a software agent books the flight, orders the groceries and renews the insurance, somebody has to guarantee that the merchant is real, the buyer authorised the purchase and the money can be reversed if it was not. That is precisely the service Visa has sold for sixty years, and an autonomous buyer needs it more than a human one does, not less.

Analysis: the toll booth's actual risk

Stablecoins get the headlines, but they are the slower of the two threats to Visa's economics and the easier one to co-opt. A merchant that accepts a dollar-pegged token avoids interchange, which is real money — but it also gives up chargeback rights, fraud liability transfer, credit, rewards and universal acceptance, all of which the merchant's customers are attached to even when the merchant is not. Visa's answer, sensibly, is to sell the plumbing to whoever wants to try.

The economics also flatter Visa more than the rhetoric suggests. Interchange — the fee that gets attacked in Congress — is not Visa's revenue at all; it is set by Visa and paid by the merchant's bank to the cardholder's bank. Visa earns network and processing fees on top. A world of lower interchange damages the banks that issue cards long before it damages the network that connects them, which is one reason Visa can afford to sound relaxed.

The nearer risk is political and legal, and it is already in front of a judge. In June, a federal court granted preliminary approval to a settlement with merchants — reported at roughly $200 billion in value over its term — that would cut interchange by about 0.1 percentage points for five years and let merchants surcharge more freely and refuse specific card categories, including premium consumer and commercial cards.

The retailers who would receive that money do not want it. The National Retail Federation, the Retail Industry Leaders Association and Walmart have told the court they oppose the deal and would rather go to trial. When the plaintiffs' largest members reject a $200 billion settlement, they are saying they think the asset being settled is worth considerably more.

Behind that sits the Credit Card Competition Act, which would require banks with more than $100 billion of assets to enable at least one unaffiliated network on their credit cards — attached, at one point, as an amendment to stablecoin legislation. It has not passed. It has also not gone away in five years of trying, and its logic is the one thing that would genuinely change Visa's business: not a new rail, but a mandated second one on the existing cards.

It is worth being clear about the scale of what is being contested. The settlement's 0.1-percentage-point reduction sounds trivial and is not: applied to American card volume it is billions of dollars a year, permanently, and it establishes the principle that interchange is a negotiable number rather than a posted one. Both sides are fighting about the precedent more than the basis points.

Analysis: what the layoffs actually say

Cutting 7 per cent of staff in the same quarter you announce record volume is not a distress signal; Visa's operating margins remain among the best in the S&P 500. It is a reallocation, and the destination list tells you what management is worried about. Every item on it — stablecoins, agentic commerce, cross-border, business payments — is a place where the transaction might one day happen without a card being present.

That is the correct anxiety. Visa's franchise was never really about plastic; it was about being the default set of rules by which a stranger's money can be trusted. Cards were the expression of that, not the substance. The company appears to understand this better than its critics do, which is why it is spending severance savings on rails it does not own instead of defending the ones it does.

The number to watch is not $4 trillion, which will keep rising with global consumption whatever happens. It is the 34 per cent — the growth rate of the services Visa sells that are not interchange. If interchange is legislated or litigated downward, that line is the company. On current evidence, it is being built fast enough.

There is one more reading of the quarter worth holding onto. A company that processes a tenth of the world's consumer spending, grows its highest-margin division at 34 per cent, and still decides it must remove 2,600 people and rebuild around agents and tokens is not a complacent monopoly. Whether that turns out to be foresight or over-correction, the pattern is the same one that separated the technology companies that survived the last platform shift from the ones that explained why it would not affect them.