Stablecoins were supposed to route around the banking system. Instead, the companies scaling them are building deeper into it than anyone predicted.

Stripe paid $1.1 billion for Bridge, whose core product is orchestrating banks. Citi is launching crypto custody. Standard Chartered is testing stablecoin settlement in Singapore. One by one, the operators moving institutional volume keep landing on the same architecture.

An enterprise cross-border payment has three legs. The payer’s money moves in local currency over local rails—a Brazilian importer paying in BRL via Pix. The payee receives local currency on their end—the supplier collecting dollars in their account.

Between them sits the middle leg: getting value across the border from one institution to the other. That leg used to run through correspondent banking, SWIFT messages hopping between intermediary banks, each holding accounts with the next, each adding a day and a fee. When both institutions accept a stablecoin, that leg settles on-chain in seconds. Banks still own the other two.

That is the division of labor. Stablecoins settle the middle leg. Every flow still begins and ends in fiat, and that is where banks are non-negotiable: the entry point, the compliance anchor, and the local rails in every market a payment touches. The companies scaling on stablecoin rails built into the banking system, corridor by corridor.

The money starts in fiat

Every enterprise payment flow begins in a bank account. Payroll, vendor invoices, customer revenue, and capital distributions live in fiat, they move through regulated financial infrastructure, and the companies sending and receiving them keep that infrastructure regardless of what a payment provider prefers.

The scale of the gap makes the point concrete. The cross-border payments market reached $208 trillion in 2025, according to FXC Intelligence. Genuine stablecoin payments ran at roughly $390 billion annualized as of late 2025, per McKinsey and Artemis, which works out to about 0.02% of global payment volume across both cross-border and domestic flows. The headline figures of $30 trillion or more in annual stablecoin “volume” mostly reflect bots, exchange flows, and automated trading rather than payments. A hedge fund treasury desk, an enterprise running payroll across thirty countries, and an exchange settling institutional withdrawals all begin in fiat.

The question these operators ask is which banking infrastructure connects reliably to which settlement rails, and who built it with enough depth to hold at institutional volume. Most stablecoin companies struggle to answer.

The gap between $50M and $10B

At $50 million in annual payment volume, one banking relationship, one stablecoin issuer, and one compliance layer carry the load. At $500 million, the flows outgrow them. At $10 billion, the question is no longer how good your technology is—it’s how many corridors your banking, FX, and licensing stack can actually carry. Volume follows infrastructure.

Brazil illustrates the pattern. Pix, the country’s instant payment system, moved more than R$35 trillion in 2025, roughly $6.3 trillion, and B2B transactions made up 47% of that value by the central bank’s own breakdown. BRL settlement, local rail access, and FX infrastructure all become mandatory at institutional volume.

Each layer takes years of relationship-building with banks, regulators, and local counterparties. The stablecoin mechanism itself works cleanly. The regulated fiat-to-crypto bridge, the multi-corridor banking stack, and the FX infrastructure that handles multi-currency conversion at scale are where growth stalls. Companies that hit a ceiling at mid-scale are almost never stopped by the crypto layer. They’re stopped by the banking layer they never built.

Why one bank is never enough

Single-bank dependency is the most underrated operational risk in crypto payments today. Most companies on stablecoin rails lean on one primary banking partner.

Banks exit fintech and crypto programs with little warning. They leave corridors after a regulatory shift. They revise their risk appetite when management changes or a compliance review lands badly. Recent history supplies the evidence. The Silvergate wind-down, the Signature Bank receivership, and the FDIC “pause letters” that Coinbase later obtained through public records requests all show the same pattern. In March 2026, the FTC sent formal warning letters to PayPal, Stripe, Visa, and Mastercard over debanking practices, part of a broader federal effort that traces back to an August 2025 executive order.

For a company with a single banking counterparty, losing that relationship means an immediate operational shutdown. The answer is banking depth that makes the loss survivable. That means multiple regulated connections, redundant rail access, and compliance architecture that satisfies every jurisdiction in the operating corridors. That foundation costs time and money to build, and it holds when demo-grade setups collapse.

Compliance is the moat

The crypto-native instinct treats compliance as friction and banking as legacy overhead. That framing survives at small scale, where the counterparties are retail users. It falls apart when the counterparties are CFOs at multinationals, treasury teams at global funds, and compliance leads at tier-one exchanges. These buyers hold their infrastructure partners to the same regulatory standard they answer to themselves.

The regulation now reinforces the point. The GENIUS Act, signed in July 2025, ties compliant stablecoin issuance to bank-grade reserve, disclosure, and licensing requirements. Even where the rules permit nonbank issuers, they push serious volume toward bank partnerships and bank-custodied reserves. An EY-Parthenon survey found that 13% of financial institutions and corporates currently use stablecoins, while 80% of non-users are actively exploring adoption.

Demand is building. The bottleneck is the supply of regulated, institutional-grade infrastructure that enterprise buyers trust. A stablecoin company that cannot show licensing depth, banking connectivity, and defensible compliance loses institutional deals to one that can.

What durable infrastructure looks like

The companies building payment infrastructure that survives at enterprise scale are integrating with banks rather than pulling away.

They combine regulated banking connectivity across several counterparties, local rail access in the corridors that matter, FX infrastructure that handles multi-currency conversion, and stablecoin settlement as the programmable layer on top. B2B stablecoin payments reached a roughly $226 billion annualized run-rate by late 2025, up 733% year over year, and that growth concentrated in companies that solved the banking layer first.

Stablecoins add value through speed, programmability, around-the-clock settlement, and reduced correspondent friction. Those advantages become accessible once the banking foundation exists beneath them. Infrastructure that performs in a demo and fails at production volume is a product that never finished being built.

Bernardo Brites is co-founder and CEO of Trace Finance, which builds regulated banking and stablecoin settlement infrastructure for Brazil, the U.S., and emerging markets. The views here reflect his own commercial vantage point in that market.