Salman Syed took Marqeta through its IPO as the company's North America business lead. His next company has Visa and Mastercard writing strategic checks into the same seed round. That alone is unusual. Card networks rarely line up like this without the underlying market being strategic.

Astrada closed a $3.8 million seed round on May 5, 2026, led by Bain Capital Ventures, QED Investors, and Nyca Partners, with participation from Visa, Mastercard, and several smaller institutional investors. The company has been operating since 2024. Astrada has processed more than $750 million in card spend and over three million transactions across all three major card networks through a single API. Early customers include Workday, Zoho, Payhawk, and Miter.

The product description is dry. The strategic implication is not.

The networks are not buying a payments product. They are buying a stake in the data layer agents will use to decide what to spend, on whose behalf, against which mandate. Whoever owns that layer owns the decisions.

What Astrada actually does

Astrada describes itself as "the data layer for autonomous finance." In practice, it offers a single API that surfaces real-time card and banking data across Visa, Mastercard, and American Express. The integration work that has historically required separate connections to each network and each issuer collapses into one endpoint. Spend categorization, transaction detail, and balance visibility flow through the same pipe.

Salman Syed framed the company's thesis directly: "Finance is moving from manual to autonomous, and the infrastructure hasn't kept up. We built Astrada to be the real-time data layer that finance platforms need." The company itself has stated that "as businesses increasingly adopt AI agents to manage financial workflows, the need for a unified, real-time data layer has become critical."

The customer list points at the same thesis. Workday is the largest enterprise HR and finance platform. Zoho operates business productivity software at mid-market scale. Payhawk runs corporate spend management. Miter is a vertical finance platform for construction. Each one is a context where an agent might be authorized to read spend, categorize it, route approval, or initiate a follow-on action. None of those workflows can run cleanly on batch data delivered overnight.

Why the data layer is the new battleground

For two decades the card networks have competed on rail volume. Issuers, acquirers, merchants, and processors have built businesses on that competition. The agent layer breaks the model. Agents do not need a card to be issued in their name. They need to know what cards are spending, in real time, with enough categorization to make a decision against a mandate.

That is the data layer. Whoever controls it controls the surface where agent decisions get made.

State of the Stack: Agentic Commerce 2026 named six pieces of infrastructure agentic commerce needs before it works at scale: commitment governance, identity, dispute resolution, settlement, observability, and the audit layer. The data layer underneath all of those was treated as a given. It is not. The historical aggregators, Plaid, Codat, TrueLayer, and the bank-direct integrations a handful of fintechs have built, all share the same architectural problem. They were built for human-initiated workflows where data freshness in minutes was acceptable. Agent workflows need it in milliseconds, with the same fidelity across networks and the same field structure across issuers.

Astrada is a new entrant in that space with a clean technical premise (single API across all three networks), a credible operator (Marqeta IPO experience), and as of this week, strategic alignment with both card networks at once. That combination is rare.

The networks' converging agent strategy

Visa and Mastercard have been making their own moves into the agent stack. Mastercard's Agentic Commerce Developer Kit on the Amex side of partner work, Visa's stablecoin pilots through Tempo, the joint Fiserv work on agent-addressable rails, and Stripe's machine payments protocol are all moving toward the same destination: a stack where agents can transact, with cards still in the loop somewhere, but the rails alone do not deliver the value.

The Astrada investment makes the networks' bet on the data layer concrete. Both networks already see the spend that crosses their own rails. What they do not see, and what Astrada offers, is a clean cross-network view at the platform level, surfaced to the buyer applications agents will operate inside. A finance platform integrating Astrada once gets visibility across all three networks. That is a different kind of footprint than either network can build alone.

It is also a hedging move. If agents start routing decisions through a data layer the networks do not control, the rails get commoditised faster. Owning a stake in the dominant data layer is a way of ensuring the rails remain visible inside the agent's decision context, even if the agent never asks which network it is using.

What gets squeezed

The existing data-aggregation layer just gained two well-funded competitors with native rail integration.

Plaid has been the dominant US bank data aggregator for a decade. Its recent integration with Perplexity is one signal of where it expects to compete. Codat and TrueLayer have built strong positions in the SMB and European markets respectively. None of them have the simultaneous structural support of both major card networks. Astrada, with $3.8 million on a seed round, does.

That does not mean the existing aggregators lose. The market is large enough for several layers, and bank account data sits adjacent to card data, not in the same field. What changes is the pricing power and the integration default. If Workday, Zoho, Payhawk, and Miter establish that "single API for autonomous finance" means Astrada, the gravitational pull on the next ten enterprise finance platforms is meaningful.

For acquirers, processors, and issuer-side fintechs, the implication is that the value-added layer above the rails is moving. Owning settlement and authorization is not enough. The platforms agents will run in are going to read from a unified layer. Either you contribute to it, or you are read about by it.

What to watch

Three things will tell us whether Astrada is the autonomous-finance data layer or one of three viable contenders.

The first is integration depth. Single API across three networks is a clean pitch. The execution is the question. Issuer-level coverage, latency at scale, and field-level fidelity across networks are where these claims succeed or fail.

The second is the network response posture. Both Visa and Mastercard participating in the same round is unusual. If either network later builds a competing first-party data product, the relationship gets complicated. If they continue investing through Astrada, the data layer becomes a shared resource the rails reach through.

The third is the regulatory layer. Two duopoly card networks investing in the same financial infrastructure provider will draw attention from competition authorities, particularly in the EU, where the interchange revolt and the broader scrutiny of network coordination are already live. The structure of the investment matters. So does what it enables.

Sources

When the duopoly invests together, is it cooperation against a common threat, or is it the data layer becoming an oligopoly before agentic finance ships?

Charlie Major is a Product Development Manager at Mastercard. The views and opinions expressed in Major Matters are his own and do not represent those of Mastercard.