Agents Can Spend Now. Everything Changes.
Visa announced today that AI agents can initiate payments across its global network. That sentence should stop you cold.
There are announcements that extend a trend, and there are announcements that close a chapter. What Visa said at its Payments Forum in San Francisco this morning falls into the second category. Not because the technology is entirely new, but because of who is saying it, the scale at which they are operating, and the specific words their chief product officer chose to frame it.
"AI is transforming the front end of commerce. Stablecoins are reshaping the back end."
That is not a startup pitch. That is the product officer of a company that processes over 200 billion transactions a year describing his operating thesis. The agentic economy now has a payment card. The infrastructure question that has hung over this space for three years has been answered.
What Visa Actually Built
At Visa Payments Forum 2026, Visa announced new AI, stablecoin, and token capabilities built around a single objective: ensuring trust, security, and control evolve alongside increasingly fast, automated, and intelligent commerce experiences.
The most consequential piece is not the technology. It is the governance architecture around it. Visa introduced an Agentic Directory, a registry of verified agents and merchants that gives merchants confidence in which agents can be trusted to transact on their platforms, and gives agents confidence they are interacting with legitimate counterparties. Alongside it, Agent Score allows merchants to evaluate whether AI agents can navigate and complete tasks on their websites.
This is the trust layer. Without it, agentic payments are a liability waiting to happen. With it, they are a scalable commercial infrastructure. Visa just built the identity and verification framework that every serious agentic commerce deployment needs, and made it available to every merchant and developer in its ecosystem simultaneously.
Visa also announced a strategic partnership with OpenAI to connect its payment network directly to AI agents, including ChatGPT tools, providing its global network, credentialing capabilities, and security infrastructure to support agentic commerce experiences. ChatGPT can now spend money on your behalf via Visa. That sentence was science fiction eighteen months ago.
On the settlement side, the numbers confirm that this is no longer experimental. Visa's stablecoin settlement has reached an annualized run rate of approximately $7 billion as of March 2026, with pilots expanding across multiple regions, blockchains, and currencies. The company is also introducing tokenized deposits, which allow banks to turn traditional deposits into programmable, always-on digital money.
The Developer Layer Was Ready First
The Visa announcements did not arrive in a vacuum. The open API layer that makes them programmable for every developer has been live for ten days.
On June 2, Crossmint launched the public release of its agentic card payments API, built in partnership with Visa Intelligent Commerce and Basis Theory. Agents in platforms including Claude Code, OpenClaw, and Hermes can now securely pay using cardholders' eligible Visa cards, with tokenized credentials and spending controls, without ever touching raw card data.
An audit of published agent skills found that card credentials were exposed in 7.1% of instances before this infrastructure existed, a level of risk that is unacceptable for corporate treasury and compliance teams. The API closes that gap. It gives every developer building agentic workflows a production-grade payment layer with PCI compliance, SOC 2 certification, auditable logs, and programmatic guardrails from day one.
Crossmint is also working with Mastercard and American Express to expand support for agentic card payments beyond Visa. The card networks are not competing for which one owns agentic payments. They are collectively building the infrastructure so the category exists at all.
The Companies That Saw This Coming Are Worth Billions
The market has been pricing the agentic finance thesis for months. Ramp recently raised $500 million in a round led by Iconiq Capital at a valuation of $22.5 billion. Iconiq's general partner said the firm believes Ramp is leading in agentic AI and setting the standard for AI-powered finance at a defining inflection point.
Ramp now powers over $10 billion in accounts payable spend annually, a ten-fold increase in just over two years. The company's stated goal is to become a one-stop shop for all financial operations, and it frames AI and automation as the mechanism that will upend back-end business processes entirely.
The valuation is not a bet on Ramp's current revenue. It is a bet on what happens when every finance workflow that can be automated is automated, and on Ramp being the infrastructure through which that automation runs.
The Companies That Did Not Are Paying in Headcount
The contrast with the companies that did not build this infrastructure fast enough is not subtle.
Bill.com cut 30% of its workforce in May, approximately 709 jobs, framing the decision as a pivot to becoming an AI-native company. Its CEO said AI is no longer one of three top priorities. It is the only one. The sequencing matters: Ramp saw this coming years ago and built toward it. Bill.com is restructuring toward it now, under competitive pressure, with a smaller team and more ground to cover.
Standard Chartered announced plans to cut approximately 7,800 roles through 2030, with the CEO explicitly stating the bank is replacing lower-value human capital with machines, targeting HR, risk, and compliance roles across multiple hubs. That sentence, from the CEO of a 160-year-old institution, is worth keeping. The banks are no longer hedging the language. They are saying it plainly.
The Open Question
The payment infrastructure for the agentic economy is now live. Visa's network. OpenAI's agents. Crossmint's API. Mastercard's Agent Pay framework. The stack is assembled.
What is not resolved is the governance layer above the payment layer. An agent that can spend money on your behalf needs rules: what it can buy, from whom, up to what amount, with what audit trail, under which regulatory framework. Visa's Trusted Agent Protocol provides part of the answer. Ramp's programmatic guardrails provide another part. But the complete governance infrastructure for autonomous financial agents operating inside regulated institutions does not yet exist at scale.
That is the remaining infrastructure problem. And it is the most valuable one left to solve.
Every other piece is in place. The agents are capable. The payment rails are live. The card networks are onboard. The developers have their API. The only thing standing between today and a world where autonomous agents manage material portions of corporate and consumer finance is the governance scaffolding that makes regulators, boards, and compliance teams comfortable enough to deploy at scale.
That gap will close. The question is who closes it.