Mastercard framed the debate around three questions
Stablecoins and agentic commerce have been discussed heavily over the past two years, but the firms that have to make real balance-sheet choices are not the stablecoin issuers or the agent builders. They are the card networks. For them, the issue is sharper: if money stops moving on their rail, what exactly is left?
Mastercard CEO Michael Miebach offered a clear answer across two appearances on July 30, the day the company reported second-quarter results. He first spoke with analysts on the earnings call, then sat down with Motley Fool co-founder Tom Gardner for a 46-minute podcast. The tone differed by audience, but the logic did not. Settlement may spread across multiple rails; trust, security and interoperability are the layer Mastercard still wants to control.
Three days later, on August 3, Mastercard completed its acquisition of BVNK for $1.8 billion, five months ahead of the original year-end timeline.
The article ties those remarks to three questions: why Mastercard is willing to loosen its grip at the settlement layer, why it is holding the line at the trust and security layer, and what BVNK adds to that strategy.
What Mastercard says it really sells
At the start of the podcast, Gardner asked Miebach to explain the relationship among banks, merchants and cardholders. Miebach first corrected a figure: not 4 billion cardholders, but 3.7 billion. He then made a point that shapes the rest of the argument. Cardholders are not Mastercard’s customers. Its customers are banks and, in some cases, merchants, whether that is Walmart or JPMorgan.
That distinction matters because it means Mastercard does not primarily sell payment itself. It sells the layer that makes payment trustworthy.
In practice, that trust looks like a guarantee. A consumer can place an order online and leave the site expecting delivery, or walk out of a store with goods in hand, because Mastercard stands behind the merchant with a payment promise: let the person go, and the merchant will get paid. The transfer of funds in the four-party model happens afterward, as the cardholder’s bank sends funds to the merchant’s bank. Mastercard has to make that work across 220 countries and territories and 3.7 billion cards, even though regulation and infrastructure vary by market.
The other side of that promise is risk control. If a consumer pays on a fake website, the user is still protected if the fault was not theirs. To stop fraud earlier, the system scans enormous volumes of data points at nanosecond speed, asking whether the customer could really be in that place at that moment and whether the spending amount exceeds any prior transaction.
Miebach described Mastercard as the operating system for the digital economy. The line sounds polished, but the structure behind it is more revealing. He broke that operating system into a security layer, a money movement layer and a data-and-insights layer. The money movement layer is plural: it stretches across cards, account-to-account systems and stablecoins. The trust and security layer is singular.
The company’s revenue mix is moving in that direction as well. Second-quarter net revenue grew 12% on a currency-neutral basis. Value-added services and solutions grew 18%, outpacing the company total by 6 percentage points, and roughly 60% of that business remains network-linked. Security, identity, fraud management and personalization are not payment processing in the narrow sense. They are judgment and protection sold around payment flows.
Miebach also put a scale on the threat side. By 2030, fraud and cyber risk losses are expected to reach $15.6 trillion, he said. If cyber risk were a country, it would rank as the world’s third-largest economy. Mastercard’s strategy, in his description, has shifted from defense to offense. Using threat intelligence capabilities built on Recorded Future, the company identified more than 7 million card-testing transactions across 192 countries in the first three quarters and estimated that it prevented $172 million in losses.
Stablecoins: useful in some flows, not yet a complete payment answer
Miebach’s clearest summary of stablecoins came in the opening remarks on the earnings call rather than the podcast. Stablecoins have enormous potential, he said, but they still need a few required principles to scale in a meaningful way: reliability, safety and interoperability. In his telling, that is what Mastercard provides.
He did not reject stablecoins outright. He rejected the idea that they already come with those conditions built in.
He then drew a line around use cases. On the call, he said stablecoins have clear utility in some B2B and P2P flows, but in P2M, consumer-to-merchant payments, there is no problem that needs solving.
On the podcast, he turned that into a simple contrast. Buying coffee at a neighborhood shop does not require a stablecoin. Paying an overseas supplier through correspondent banking can be a different matter. If a small business wires $100 and two intermediaries each take $5, the recipient ends up with $90. In that setting, stablecoins can solve a real problem.
His standard is practical rather than ideological. The issue is not whether a technology is fashionable, but whose problem it solves. Mastercard has been investing in account-to-account systems since 2016, he said, and is now one of the largest A2A solution providers. Cards remain a large part of the answer, but not the answer to every payment type. The network can handle U.S. dollars, other fiat currencies and stablecoins, but the goal is not simply to let stablecoins move through the system. It is to move the same protection people expect from card payments into those flows.
In other words, Miebach did not say stablecoins fail as a concept. He said they still lack a broad acceptance network.
Mastercard spent 60 years building what he called the largest acceptance network in the world. A payment method that reaches only a small set of terminals is not enough. Scale matters. Predictability matters. Protection matters. On the call, he laid out the broader view in more structured terms: the future will be plural, with many coins and many chains, and those systems will need a trusted interoperability layer because people will transact across different forms of value.
The company has been active in digital assets this year. It partnered with SoFi in March, obtained a BitLicense from the New York State Department of Financial Services in May, and in June expanded settlement support to six regulated stablecoins across eight blockchains. Earlier efforts include the Mastercard Multi-Token Network, Crypto Credential and One Credential, which places fiat and stablecoin balances under a single credential.
Even so, the scorecard remains thin. On the earnings call, Miebach gave only one growth figure linked to digital assets: transaction volume on crypto co-branded cards has more than tripled over the past two years, with Bitget and Kraken added in the quarter. That number is real, but it measures spending by crypto holders through cards. It does not show how much stablecoin settlement is already running directly on Mastercard’s network.
He did not provide that figure. The article notes that some analysis has pointed out Mastercard has never published stablecoin settlement volume comparable with Visa’s, and many initiatives may still be at pilot stage. Visa’s disclosed figures, by contrast, are public: by April this year, its annualized stablecoin settlement run rate had reached $7 billion, up 50% quarter on quarter, covering nine chains and more than 160 stablecoin card programs. In on-chain card settlement share, the split cited in the article was 97% for Visa and 3% for Mastercard, even though the two companies support nearly the same number of projects.
The result is a gap between framework and flow. Mastercard appears to have built a broader conceptual architecture, but less visible traffic has landed on it so far. The evidence Miebach can point to today is stronger on stablecoin holders spending through cards than on stablecoins settling at scale on Mastercard rails.
Agentic commerce: if humans are still buying, Mastercard thinks cards still fit
On the earnings call, analyst Ramsey El-Assal asked one of the strongest questions of the day: are there scenarios in agentic commerce that truly require stablecoins, or can conventional Mastercard credentials cover the full range?
Miebach did not answer with a clean yes or no. Instead, he split the market into two categories.
AI buying for people or companies: cards can carry the transaction
The first category is AI acting on behalf of a person. Miebach used a camping example. A user asks AI what they need for a trip, and it returns a list of 15 items, already aware that the person owns a tent and does not need another one. The friction comes next: the user still has to visit different sites and manually buy everything. If checkout could happen right there, with an agent completing the purchase on top of a Mastercard credential, the process becomes far smoother.
That also creates a new risk. A new entity sits in the middle of the transaction. How does the consumer know the agent is genuine? How do they know it bought what was intended? If it orders two grills instead of one, what evidence can the customer use after the charge posts?
Mastercard Agent Pay applies existing card tools to that problem: tokenization, zero-liability protection and dispute handling. The most important piece, Miebach said on the call, is Verifiable Intent. It gives users a way to challenge a transaction by saying they never intended to buy that item, which reopens the chargeback process. He added one detail on the call that was not in the podcast version: the capability was built with Google.
Chargebacks matter because dispute handling is expensive. Merchants still need reach and predictable user experience, and so do consumers. For that reason, Miebach said Mastercard believes cards will win in that world.
He extended the same logic to businesses. On the call, he said there will be B2B agentic transactions in which an agent procures goods or services for a company. Those transactions can also run inside the card ecosystem because the fit around ticket size, speed, intended use and protection requirements is similar. In his framing, AI buying for consumers and AI buying for enterprises do not amount to a direct challenge to the card rail.
The revenue angle is visible as well. Tokenization is itself a service Mastercard sells, and token penetration just moved above 40% of switched transactions in the second quarter. The remaining share is still a runway for expansion.
The article also notes a limit to this thesis. Agentic commerce of this kind looks more like a replacement of existing volume than the creation of entirely new demand. People are not going to buy five extra tents just because an agent made shopping easier.
Machines buying from machines: on-chain permissioning, off-chain settlement
The dividing line appears when the buyer is no longer a person but a machine.
If a company is purchasing APIs, compute power, data or digital content, why should it wait for an invoice? A more efficient model would be usage-based payment, scaling compute up by 10% when needed, scaling it down later, and paying only for what is consumed. That improves working capital efficiency, which matters both to chief procurement officers and to CFOs.
Miebach’s point was that this kind of market needs a payment ecosystem that is always on, high frequency and priced in tiny fractions of a dollar. In his view, that ecosystem does not yet exist today.
He summarized the structure in one line on the earnings call: on-chain permissioning, off-chain settlement. He then said Mastercard is the only network that supports machine-to-machine payments.
The split is deliberate. Permissioning happens on-chain because machines need immediate recognition of a valid transaction. Settlement happens off-chain because the route money takes is a different question. In the Q&A, he said settlement could happen across different types of rails, potentially with stablecoins or with something else, and Mastercard is open on that point.
The podcast version was even plainer. The underlying rail and infrastructure are likely to differ from the card rail. It could be stablecoins or something else. Enterprises will choose. Mastercard wants to stay neutral on the base rail while keeping hold of the trust and interoperability protocol above it.
The first ecosystem includes more than 30 partners, among them Adyen, Ant International, BVNK, Checkout.com, Coinbase, OKX and Cloudflare. The presence of a CDN company like Cloudflare signals where Mastercard thinks this market is headed: not traditional purchase-order workflows, but native digital spend such as compute and API billing.
That makes the company’s concession unusually precise. As long as the thing being bought is still being bought by a person, whether for personal or corporate use, Mastercard thinks cards remain the winning rail. Only when the buyer becomes a machine does it explicitly acknowledge that the settlement rail may no longer be a card rail. Even then, it wants to own the upper layer of trust and interoperability.
AI: the companies with proprietary data have the edge
Gardner also pushed Miebach on a larger macro question. If AI reaches a point where, as Elon Musk has suggested, it exceeds the aggregate of human intelligence within five years and makes most work cheaper, while large cash-rich technology companies continue cutting staff, what does that mean for consumer spending?
Miebach began with the technology itself. AI needs to be explored, he said. Deployed well, it can drive prosperity and growth. It also has a downside. Generative AI is helping fraudsters, scammers and hackers at the same time it helps defenders. The same technology can attack and protect, which makes it an arms race.
On the earnings call, he put that in more operational terms. Cybersecurity is now one of the central concerns occupying CEOs and boards. A major question around frontier models is whether they are a threat or a tool for finding weaknesses inside a company’s own systems. His answer was both. Mastercard is already using frontier models to speed up identification of internal vulnerabilities and then sharing those practices with customers.
At the human level, Miebach argued that this industry, and in fact most industries, still depends on having the best talent. That means helping staff upgrade skills. He described the direction as human-centric AI applications: use the tools to improve work and remove repetitive tasks. He used his own workflow as an example, saying he built an AI assistant to handle email processing. He still reviews the output, but the more tedious work is stripped away.
He also said many customers now want to talk with Mastercard about agentic commerce and stablecoins. Internal teams use AI to organize public information before those conversations, freeing time for the parts that matter most.
As for why Mastercard does not see itself as an industry that needs to be fundamentally reinvented because of AI, Miebach said the company has always been about technology rather than headcount. It is a network company operating across 220 countries and territories with a labor base that is relatively light compared with its market value. In his words, this is not the kind of industry that requires a complete rethink.
He also moved Mastercard out of the AI infrastructure trade. The company is not an AI trade and not AI infrastructure, he said. It is focused on applied AI.
That view showed up in capital allocation too. Mastercard repurchased $4.9 billion of stock in the second quarter. During a period when AI-linked stocks dominated capital flows, Mastercard had become a source of funds for investors rotating elsewhere, and the share price at one point fell from $570 to $470. Miebach said the buyback was opportunistic. The company does not treat repurchases as a business in themselves; it acts when the timing makes sense.
His simplest line on the AI era may have been the most revealing. The companies that will stand out are the ones that can use many different models and also possess proprietary data to feed them. Mastercard, he said, has one of the most unique datasets available: transaction data. That, in his view, is what gives the company staying power and a right to participate in what comes next.
Why BVNK matters: Mastercard is filling the settlement role on a new rail
The August 3 close of the BVNK acquisition fits directly into the four-party model Miebach was describing.
In the traditional fiat system, final movement of funds is handled by banks, while Mastercard supplies the instruction set and clearing logic. In the familiar shopping example, the cardholder’s bank and the merchant’s bank are the institutions that actually move money. Mastercard provides the guarantee and the routing. Because member banks make up the fiat rail, Mastercard does not need to hold the settlement license itself and does not directly control the money.
That changes once settlement may happen over blockchain-based stablecoins. On that rail, there is no equivalent member-bank structure automatically doing the work that banks do in fiat cards.
BVNK fills that gap. On the earnings call, Miebach described the role this way: with BVNK, Mastercard can act as the trusted interoperability layer that lets customers send, receive, store and convert assets.
What BVNK brings in practical terms
The deal closed on August 3 and was structured as $1.5 billion in base consideration plus a $300 million earnout, bringing the total to $1.8 billion. That was five months ahead of the original year-end target.
According to the article, BVNK brings about $30 billion in annualized stablecoin transaction volume, more than 25 licenses across 130 markets, MiCA authorization obtained in February, and direct access to the SEPA euro rail through Lithuania. Its customer list includes Worldpay, Deel, Rapyd, Flywire and Visa Direct.
Inside Mastercard, BVNK is expected to handle three concrete tasks:
- provide 24/7 stablecoin settlement for processors and acquirers;
- add stablecoin checkout capability to Mastercard’s payment gateway;
- support conversion between fiat and stablecoins.
Those are functions member banks perform on the fiat side. BVNK performs the parallel set on the stablecoin side.
Miebach’s own explanation for the acquisition contained one line the article highlights: BVNK’s payment orchestration, license base and connectivity are highly differentiated, and they are already running in the market today. What Mastercard bought for $1.8 billion, then, was not just technology. It bought time.
Chief Product Officer Jorn Lambert put the official framing in the closing announcement: in a multi-currency world where fiat, stablecoins and tokenized deposits coexist, the next payment paradigm will be defined by how efficiently different rails connect to each other. The article’s reading of that line is straightforward. Mastercard does not want to be every rail. It wants to be the connector. But to connect effectively, it first needs to hold at least one real rail in its own hands.
The bidding process says more about strategy than the press release does
The article also cites comments from BVNK early investor Kjartan Rist, founding partner of Concentric, who said he entered the company in 2019 at a $4 million valuation and had not sold a single share in eight years. His account adds detail the official deal language did not.
First, the pressure point was Stripe. At the end of 2024, Stripe bought Bridge for $1.1 billion. Rist said Mastercard has deep respect for Stripe, then translated that into plainer language: Mastercard is looking back over its shoulder. He broke the threat into three parts: execution, product simplicity and the absence of legacy baggage.
Second, Coinbase reportedly bid higher but lost on fit. The article says reports put Coinbase’s offer at one point as high as $2.5 billion, but BVNK’s founders cared more about chemistry with the eventual buyer. Coinbase is an exchange; Mastercard is a financial services company.
Third, Visa was in the boardroom and still chose not to pursue the asset. Visa is both a shareholder and a board observer, but did not make a move. Rist’s interpretation was that Visa prefers a different strategy: work with multiple operators rather than own one.
The contrast is clear. Faced with the same target, one network is renting pipes and the other is buying a pipe.
On price, the article argues $1.8 billion was not excessive. Mastercard spent $4.9 billion on share repurchases in the same quarter, so the acquisition cost was only a bit more than one-third of that. But the two uses of capital are not equivalent. Shares bought back can later be reissued. An acquired company cannot simply be handed back.
The BitLicense Mastercard obtained in May is presented as another signal. Card networks traditionally do not need to hold these licenses because member banks hold the regulated position. Mastercard instead applied directly to the New York State Department of Financial Services for authority linked to clearing tokenized deposits and payment stablecoins. Combined with the more than 25 licenses and SEPA connectivity BVNK brings, the picture is of a company building a new rail itself, at a pace that the article says may exceed some fintech peers.
On the stablecoin rail, there is no member-bank system waiting for Mastercard. So it is taking on part of that role itself.
The underlying question has not changed: who is responsible when something breaks?
The article closes on one of Miebach’s simpler lines. What consumers want from payments has not changed for decades, he said: simplicity, safety, and clarity about what happens if something goes wrong.
That is what card networks have been selling for 60 years. Not clearing in the abstract. Not switching in the abstract. The real product is the answer to a harder question: who bears responsibility when there is a problem?
In the article’s framing, stablecoins do not yet offer that answer because there is no native chargeback on-chain. Agents do not offer it either, which is why Mastercard and Google are building Verifiable Intent. AI models can recommend and automate, but they do not by themselves provide accountability.
That is why Miebach’s three answers on stablecoins, agentic commerce and AI reduce to the same position. Pick whichever rail you want underneath. But when something goes wrong, someone still has to be responsible. Mastercard intends to charge the party that needs that assurance. It is betting on the trust, security and interoperability layer, and to make sure that layer has something real to connect to, it has bought a new rail it once might not have needed to own.
The article leaves one premise unresolved. It asks whether responsibility itself will remain as valuable in the next payment system as it was in the last one.