无法撼动的支付双寡头：Visa与万事达卡的权力护城河与地缘博弈

Have you ever thought about why anywhere you go, it's always Visa and MasterCard? Amex works somewhere, but not everywhere. UnionPay is also a maybe, but Visa and MasterCard works with everything, everywhere, all at once. Stripe, Apple Pay, stablecoins, AI shopping agents, every single one of them posed a threat, and every one of them ended up running on Visa and MasterCard's network. And it's not just companies, countries have tried to build their way out of this dependency, and when Brazil actually succeeded, its payment system ended up being named in a US tariff action. A single warning, just a warning from the US Treasury, made Turkish banks drop Russia's card network within days. And this year, the US president personally asked Xi Jinping to let Visa into China. So, why can't anyone displace them? Can't or won't? As it turns out, there is one factor that decides whether any alternative survives, and it isn't technology. If anything, the next wave of AI and commerce, agents buying things on your behalf, might make them even stronger than ever. To understand why, we have to go back 70 years when Bank of America mailed 60,000 credit cards to people who never asked for a credit card. That experiment almost failed, and yet, it's the reason there is a Visa in your wallet today. Today is about Visa and MasterCard, ladies and gentlemen. Let's dive in.

Before 1958, credit cards were tied to specific stores, like Macy's credit card or Sears credit card. And banks needed a way to solve the chicken and egg dilemma. They had to convince people to use a neutral card, the bank's card, not associated with any store. But there was a risk that merchants wouldn't accept a neutral card because back then the logic was if you're a Sears, why would you accept a card that isn't yours? And on the other hand, they had to convince businesses to accept a card that nobody cared. So, to solve this problem, the Bank of America did two things. First, they bypassed credit card applications entirely and flooded an entire city of Fresno, California with 60,000 active pre-approved credit cards. And then they went to around 300 local businesses pitching them access to tens of thousands of people with active lines of credit, and it worked. But there was a big problem. 22% of Fresno accounts were delinquent. The fraud rate was completely out of control, and it almost destroyed the experiment, but they persevered. And it showed that if you solve the chicken and egg problem, credit cards work. So, instead of building one side of the equation and hoping for the other, Bank of America manufactured both sides of the market. Now, put yourself in the shoes of the Bank of America. You have a card that can be used with businesses that you have relationship with, but you can't scale that cuz you're not going to be knocking on every door of every shop. Plus, you only have access to people who have accounts in your bank. But sky's the limit, isn't that right? Why not give your cards to those who have accounts at other banks as well? And so, Bank of America realized to go national, you have to have your competitors carry the card. In 1970, Bank of America turns the credit card program into a cooperative owned by the banks. The cooperative later becomes Visa. Now, what's interesting here is that Visa never had to look for customers, unlike all other financial businesses, including banks, who had to spend money on advertising and marketing and all sorts of things to attract customers. Visa got a a massive sales engine in a form of banks. Banks were issuing Visa cards and giving them to both customers and businesses. But, banks weren't going to hand Bank of America the entire market. So, Wells Fargo, the Bank of California, and a few others formed their own cooperative, the eternal rival to Visa, the name of which is, you guessed it, Mastercard. At first, banks were divided into camps. They were either Visa or Mastercard. But, as of 1976, they could join both networks and offer either brand. And this became the next step in their journey. At first, they were competing for customers and businesses, then they were competing for banks, and then, once they got the banks, they had to retain those banks for them to stay loyal. And loyalty is a problem.

I'm not a big fan of podcasts. I mostly stopped to listening to them about a year ago, because at some point, when you listen to a bunch of them, they all start sounding the same. A lot of them are too futuristic and vague, and therefore very far from what's happening on the ground, or the arguments are not strong enough for me to buy in. And then, ironically, I got approached by the AI forecast, which is a podcast by Cloudera, offering to sponsor this video. And at first, I was skeptical and didn't think much of it until I checked out their content. Let me start by saying that this podcast is not for beginners. The vast majority of guests are executive leaders, founders, entrepreneurs, and the format of the podcast is the opposite of what you'd expect. The host doesn't interview people about AI in general. Their episodes are hyper-specific, and you can tell simply by looking at the titles of the episodes. The host picks one person to talk about a very specific business process in one specific domain, and that person was on the ground when it was happening. Let me give you two examples. One of the guests, Bob van Luijt, co-founder of Weaviate, breaks down the unit economics of AI search, which is something very few people talk about. Marlon Davis talks about AI wrappers that don't solve any real problems, something very close to me because we've made several episodes on this topic. And an EVP from Mercy Corps who walks through an agentic system they shipped for humanitarian crisis response. This podcast is for those of you looking for a very deep expertise in a specific domain that you are operating in. And if you need to know how other businesses solved problems that you were dealing with, head over to the AI forecast. It is available for free on YouTube, Spotify, and Apple Podcasts. And now, back to the video.

Say you buy a sweater at Zara for $100 and pay for it with a credit card. From the $100 that you paid, Zara's bank pays a cut, roughly $2, and they pay it to the bank that issued your card. That's the transaction fee. And Visa and MasterCard only take a tiny cut of that, a few cents for moving that transaction across their network. So, how did Visa keep banks from leaving? They raised the fee. Remember, the transaction fee goes to the bank that issued your card. The business accepting your payment on the other side doesn't have a choice. They'll take what's left. So, a higher fee makes the network more attractive to banks and more expensive to businesses. So, the incentives reinforce each other. Networks attract banks. Banks attract customers with rewards, and customer demand puts pressure on businesses to keep accepting the cards. Now, put yourself in the shoes of a business who looks at this entire situation and goes, "Okay, you guys are having a party over there, but I'm the one paying for this party, and I didn't sign up for this. So, I'm going to refuse accepting these cards that rip me off." And to that, Visa and MasterCard came up with something called honor all cards. If you're a business and you sign up to accept a Visa card, you have to accept any Visa card a customer comes to you with, including expensive premium rewards cards. So, the business, no matter how you look at it, is the party that absorbs all of these costs.

Now, for decades, the banks also owned these networks because a cooperative serves its members. But then, MasterCard went IPO in 2006, and so did Visa in 2008, raising $17.9 billion in what was the largest US IPO up to that point in the middle of financial crisis brewing. Which means they were no longer owned by banks. They became public companies. And you know what else happens around the same time? The age of online commerce. Just as the networks became public companies, commerce began to move online, which became a blessing and a curse for networks. Here's why. Early online shoppers needed two things: a card they could pay with and protection if the merchant was a scam. Credit cards had both, and chargebacks became the consumer protection layer. The online commerce brought in the third layer into the structure, the payment processing software. PayPal, Stripe, Square, Apple Pay. So, now when you buy a sweater on zara.com, the chain goes like this. Your bank, the network, Zara's bank, and Zara. You pay $100, your bank keeps the transaction fee, Visa or MasterCard keeps the fee for carrying this transaction through, Zara's bank takes the cut, and then they pay Zara whatever is left. On one hand, you look at this and that there are more hands in this chain. On the other hand, every online checkout makes Visa and MasterCard more valuable because every payment provider becomes their new distribution channel.

This was the blessing and now on to the curse. The same payment platforms became potentially dangerous to the two because Stripe, for example, has relationship on both sides. They can move money between accounts directly without needing Visa and MasterCard at all. And yet they don't. Why? Because the network is where the users are. When Apple launched Apple Card, arguably the most powerful consumer tech company on the planet, they didn't build an Apple payment network. They partnered with Goldman Sachs to issue the card, and they put it on Mastercard's network. So, if even Apple couldn't bypass them, and if every business and merchant complains about the outright dependency of them, why can't other countries simply build their own networks?

Of course they can, and many of them are doing it, but why can nobody come close to the big two? Can't or won't? There is one factor that predicts whether a domestic network survives, the owner of the network. If the bank owns it, it loses. If the state owns it, it wins. But that win comes at a price. Canada has its own network called Interac. Follows the exact same structure that Visa and Mastercard had back in the day. It's a cooperative of five biggest banks in Canada. I live in Canada, as you guys know, and yes, Interac is what we use every single day when we tap our debit cards. Businesses love it. Interac charges a flat amount, 1.4 cents per transaction. So, why are we still on Visa, even domestically? Because Canadian banks earn more from Visa. See, the thing is, Interac is just debit. There is no Interac credit card, and most of the spending is done on credit cards. The fact that Interac is a cooperative means that it is owned by banks, and it charges a small fee, but that leaves no money for rewards, marketing, fraud tools, or global acceptance network. Visa and Mastercard charge percentage, and those fees fund all of those things. Interac is often co-badged with Visa or Mastercard, and the Interac portion of it works when you tap your card in store. But Visa handles everything online and abroad. So, in a way, Interac is another distribution channel for Visa. Dutch banks also had a domestic card and phased it out. In fact, 13 out of 21 Euro countries have no domestic card scheme, and 2/3 of the European payments run on American networks. There's one factor that predicts whether a domestic network survives. Asking banks to build a Visa competitor is basically asking them to cut their own revenue. So, they don't.

Now, look at the countries where a domestic network did work. India and its RuPay. Been around since 2012, but for years it was a debit card for people who didn't qualify for a regular one. Then, in 2022, India's central bank, which basically means the state, made a decision to link RuPay credit cards to domestic payment system UPI. Then it gave RuPay credit cards zero transaction fees on transactions under 2,000 rupees. And RuPay boomed. It now has 10% of India's credit card market and rapidly growing. Brazil had its own network called Elo, which was owned by banks and therefore, predictably, struggled against Visa and Mastercard. But in 2020, Brazil launched a payment system called Pix, which is not a card, but an instant payment system owned and run by Brazil's central bank. In other words, the state. Pix market share in Brazil is 51% on domestic payments. One of the largest market shares in domestic payments globally. Now, put a pin in this. I'll come back to it because Washington noticed. And lastly, China. UnionPay is not the best example here because Chinese residents did not have a choice. For years, foreign networks could not clear domestic card payments in China. But, put a pin in this one, too. I'll come back to it because Washington noticed this one, too. Europe is catching up pretty late. Only in July 2026, the European Parliament began discussing the introduction of the digital euro to reduce reliance on non-EU providers, but let's be honest, non-EU means American. So, you can see the pattern. When banks own domestic networks, it stays small and local because banks earn more on Visa. But, when the state enforces the network, it can become quite powerful. But, when it becomes powerful, Washington starts to pay attention.

To understand the extent of power that Visa and MasterCard have, you can look at how they act as regulators. 2010, Visa and MasterCard stopped processing donations to WikiLeaks. 2020, Visa and MasterCard cut off payments on Canada's very own Pornhub after a New York Times investigation. Steam delisted hundreds of games under their pressure. Another example of the extent of their influence on sovereign states: when Russia invaded Ukraine, the US imposed sanctions. Russia has had its own card network called Mir since 2014, and in 2022, Turkish banks were accepting Mir to facilitate transactions from Russian tourists. In response, the US Treasury Department issued a single warning letter to Turkish banks. Within days, five Turkish banks suspended the Mir network completely. You see, building a network is one thing, but making sure it actually works internationally requires permission. And that permission is not given by other countries or by banks, but by the US government. If you try to build a network, you either hit a wall with the US Treasury, or in case of Brazil, you get hit with tariffs.

Now, remember the pins we put in Brazil and China? Let's come back to them. What did the US do when they saw a real player on the market? They imposed a 25% tariff on Brazilian products, and Pix was listed as one of the factors to justify those tariffs. The US conducted a section 301 investigation and concluded that Brazil unfairly favored Pix over American networks. Washington wants what it specifically describes as competitively neutral treatment for private payment companies rather than preferential treatment for a government-backed system. The irony is that at home, the Department of Justice is suing Visa for monopolizing debit, while outside of the United States, they impose tariffs on networks that displaces Visa Mastercard monopoly. And finally, China. In 2012, the US won a World Trade Organization case against China, which ended up forcing China to open card clearing. The US had a right to ask for retaliatory measures, but for years held off on enforcement because they wanted to use it as leverage. In 2020, American Express got a license to clear domestic payments, and in 2023, MasterCard got a license as well. But Visa, the largest one, is still waiting. And so this year, in Lima, Peru, the US president personally raised the issue with Xi Jinping. The president of the United States used bilateral discussions between the two superpowers to lobby for a private card network. If you want to know why no alternative can displace Visa and MasterCard, this is why. At home, you can't displace them because banks won't let you, and abroad, they hit a political wall.

And finally, AI. Why might the next wave of commerce make Visa and MasterCard even stronger? What is the alternative? How are companies trying to bypass the Visa Mastercard dependency? Stablecoins and AI. Stripe and Paradigm launched its own blockchain for stablecoin payments called Tempo. That launched in March on an open protocol that Stripe co-authored called MPP. Right on the launch day of Tempo, Visa, and a little bit later Mastercard, starts supporting the protocol so that agent payments could run on Visa's. Later this year, in August, Mastercard buys BVNK, one of the biggest stablecoin infrastructure companies. Do you see what's happening here? Mastercard acquired a stablecoin infrastructure to have full control over stablecoins and its network. The networks did to stablecoins the same thing they did to Apple Pay back in the day. It made them yet another distribution channel.

AI agents are a different story, however. When AI agent buys things for you, there are two things that work very differently from when you make a purchase online yourself. One, you don't make an active choice at checkout because the agent will pick whatever card is saved. To that you may say, "But yeah, but the saved card is most likely Visa. How's that different?" Yes and no. Today, when you make a purchase online yourself, a merchant has one moment when they can influence the way you pay, the checkout page. They can offer you a discount for paying in a certain way or offer some bonus for an alternative method of payment. But when an agent buys for you, nobody sees a checkout page. The agent will pay with whatever credential it was given. And two, the agent doesn't know your card number. If the agent is hacked or makes a mistake, your card will be exposed. So, how do they solve this problem? With tokens. And who issues the tokens? Visa and Mastercard. Visa alone already has more than 17 billion of them, and Mastercard built a version just for agents called Agented Tokens.

And this is what brings me to the final point, the protocol war that is only starting to unfold. Google has its protocol, OpenAI and Stripe have their protocol, Stripe and Tempo have a third. Everybody wants to own how agents shop. But Visa and Mastercard plug into all of them. They don't need to win the protocol war because every side ends up paying through them anyway. To be fair, this is still early. Open AI shut down its checkouts this year after weak usage. Anthropic released an open source shopping agent in September, but it stops at building the cart and the whole payment handling is done through the stores existing checkout.

So, are Visa and MasterCard unbeatable? Well, look at everyone who tried. Apple Pay became a wallet for Visa cards. Stripe became a distribution channel. Stablecoins are running on Visa and MasterCard. And AI agents have Visa and MasterCard tokens waiting for them before the agents are even shipped. Banks earn more from Visa than from anything they could build themselves. Businesses cannot refuse their cards because the customers carry those cards. So, is it that the countries can't build an alternative? Or do they simply choose not to? It is a choice. And when a country makes that choice, it ends up on Washington's radar. So, the next time a new payment app promises to change the way you pay, check what's behind it. It is probably a Visa. We hope this was helpful and we'll see you in the next episode. Bye.