In July 2026, Visa processed nearly $7 billion in stablecoin transactions. That number, buried in an earnings footnote, tells a story far larger than the sum it represents. Not because of the volume—though impressive—but because of what it reveals about the shift beneath the surface. For years, the narrative around stablecoins was about speed: cheaper, faster, borderless settlement. We compared TPS metrics, watched Lightning Network struggle, and celebrated when a single transfer took seconds instead of days. But the metrics were always a distraction. The real battlefield has moved.
I spent the 2020 DeFi Summer modeling yield farming mechanics for a Singapore-based fund, tracing over ten thousand on-chain transactions to understand where liquidity actually flowed. What I found then was a paradox: the more decentralized the protocol claimed to be, the more centralised the incentive structure became. Today, watching the stablecoin ecosystem mature, I see a similar illusion at work. The industry is obsessed with settlement rails—those pipes that move value from A to B. Yet the true prize, the one that Visa, Mastercard, Stripe, and Wirex are all quietly chasing, is not the rail itself. It is the layer that sits on top of it: the customer relationship, the data, the trust.
Context: The Forgotten Middle
The history of stablecoins can be divided into three acts. Act One: Tether and USDC as speculative tools—on-ramps to exchanges, nothing more. Act Two: the realisation that stablecoins could settle payments faster than SWIFT, leading to Visa and Mastercard launching pilot programmes. Act Three: the current moment, where stablecoins are becoming the core of an entirely new banking stack—one that combines payments, lending, yield-bearing deposits, and programmable automation.
As of 2026, the total stablecoin supply sits at $315.6 billion, with daily transfer volumes reaching $1,956 billion. These numbers are no longer niche. They represent a parallel financial system that is growing faster than any ten-year-old fintech startup. Visa and Mastercard have integrated stablecoin settlement for select partners; Stripe now allows merchants to accept USDC directly. But the most revealing data point comes from a company you may not have heard of: Wirex.
Wirex is a London-based crypto bank that recently announced its Banking-as-a-Service (BaaS) platform processed an annualised $1 billion in settlement volume within just 131 days of launch. That figure, while small compared to Visa’s $7 billion annualised stablecoin volume, is growing at a rate that suggests a new category is forming. Wirex does not just settle payments. It offers yield-bearing deposits through DeFi protocols like Morpho and Aave, a non-custodial wallet (EVEDEX), a loyalty credit line, and an upcoming “Agent Card” that lets users program rules for automated payments. In short, Wirex is building a complete banking operating system on top of stablecoins.
Core: The Three Layers and the Hidden War
To understand the real battle, we must separate the stack into three distinct layers: the settlement layer, the customer layer, and the automation layer. Each layer is being contested by different players with very different incentives.
Layer One: The Settlement Rail Visa and Mastercard entered the stablecoin space not because they wanted to become crypto banks, but because they needed to protect their core business: moving money. If stablecoins could settle transactions directly between parties without using a card network, the card networks would become irrelevant. So they adapted. Visa now processes stablecoin transactions through its existing infrastructure, effectively turning USDC into just another settlement currency. Mastercard launched its own “Mastercard Crypto Credential” to verify transactions. Stripe went further, allowing merchants to accept stablecoin payments and automatically convert them to fiat.
What these giants have is scale, trust, and regulatory compliance. But they lack something critical: the ability to offer yield or build a direct, sticky relationship with the end user. When you pay with a Visa card, the bank that issued the card holds the customer relationship, not Visa. Visa is just the pipe. That is the fundamental limitation.
Layer Two: The Customer Layer This is where Wirex and similar BaaS providers enter. By offering a full-stack financial product—debit cards, deposits with DeFi yields, lending, and multi-currency accounts—they are trying to become the bank itself. Wirex’s CEO Pavel Matveev told me that the company’s goal is “to own the client relationship end-to-end.” That means not just processing payments, but holding the deposits, offering the credit, and managing the risks.
The key differentiator is the yield. Wirex Earn offers up to 9.75% APY on stablecoin deposits, claiming the returns come from real lending demand on Morpho and Aave, not from token printing. In a bull market, that sounds sustainable. But I have seen this script before. During the 2021 liquidity mining craze, many protocols promised high yields from “real” activity—only to collapse when the activity turned out to be self-referential. In the code, I found the ghost of the architect. The difference this time is that the lending demand is from institutional partners like BingX and Crossmint, which use the liquidity for actual trading and merchant settlement. That adds a layer of authenticity, but it does not eliminate the risk of market downturn.
Layer Three: The Automation Layer The most intriguing development is the Agent Card, which Wirex plans to launch later this year. This is a digital card that can be programmed with rules—e.g., “pay my rent automatically if my balance is above X, but never exceed Y amount.” The card is tied to a software agent that executes payments based on user-defined logic. This is the logical endpoint of the stablecoin banking thesis: money that can be programmed to move itself.
But here lies the hidden danger. When money moves autonomously, who is responsible if it moves wrong? The user? The card issuer? The agent’s developer? There is no legal precedent for “algorithmic payment errors” on a large scale. In my years auditing smart contracts, I learned that every layer of automation introduces a new failure surface. The audit is not a check; it is a confession. A confession that the code may have bugs, that the incentives may misalign, and that the human who wrote it may have missed something. The Agent Card will require a new kind of risk management—one that the industry is not prepared for.
Contrarian: The Responsibility Void
The prevailing narrative is that stablecoin banking will democratise finance, giving everyone access to high-yield savings, low-cost payments, and programmable money. But the contrarian truth is that this new stack is built on a responsibility void. Each layer—settlement, customer, automation—assumes the other layer will handle the risks. Visa assumes Wirex will comply with sanctions; Wirex assumes Morpho will be secure; the Agent Card assumes the user will write correct rules. When something breaks—and it will—the blame will be passed around like a hot potato.
Identity is a protocol; soul is the private key. In traditional banking, the bank holds the key and the liability. In stablecoin banking, liability is fragmented across smart contracts, chain providers, and third-party custodians. The user is left holding the risk. The industry celebrates “composability” but forget that composability also compounds failure.
Takeaway: The Next Fight
When the pool empties, only the intent remains. The real battle in stablecoins is not about settlement speed or payment volume. It is about who will be trusted to hold the customer’s intent—and to bear the responsibility when that intent is mis-executed. Visa and Mastercard have trust but lack the ability to offer complete financial services. Wirex has the product but still needs to prove it can handle the risk. The winner of this war will not be the one with the highest TPS. It will be the one that builds a system where responsibility is as clear as the transaction itself.
I keep coming back to a line from a 2017 audit I did in Zurich, when I flagged a reentrancy vulnerability that would have cost $2.1 million. The team rejected my report for being “too academic.” They said the code worked in practice. It did—until it didn’t. The stablecoin stack works in practice today. But the ghosts are already in the code, waiting for the next market stress to surface. The question is: who will be holding the keys when the music stops?