Module 1

Why New Rails Exist

The structural limitations of card infrastructure and why faster money demands fundamentally different plumbing

---

The Card Rails Are Not Designed for Real-Time Settlement

The card infrastructure we use today was engineered in the 1960s for a world that did not expect real-time anything. When Bank of America built the foundation of the card network, computers filled rooms, settlement happened by mail, and a two-day clearing window was considered fast. That architecture survived through digitisation, but it was never rebuilt for an era where money moves at the speed of electrons.

Here is how card settlement actually works: you swipe a card (or tap, or click). The merchant terminal sends the transaction to the acquiring bank. The acquiring bank sends it to the card network (Visa, Mastercard, or American Express). The network sends it to the card issuer. The issuer authorises and responds. The network and acquiring bank batch the transactions throughout the day. Then, at the end of the day, they all settle simultaneously. The money moves the next morning. Sometimes two mornings. T+1 or T+2 settlement is not a feature. It is the vestigial remains of batch processing.

The card networks were engineered to settle transactions in batches because that is how computers worked in the 1960s. Real-time payment networks were engineered for a world where settlement should be instantaneous.

This matters because businesses and consumers increasingly expect money to move instantly. A freelancer should see payment arrive in their account before they finish their lunch break. A merchant ordering inventory should get confirmation of payment the moment they submit the order, not 24 hours later. An employer paying workers in emerging markets should know the money has arrived for certain, not hope it clears in a 72-hour window across multiple correspondent banks. The batch architecture is not just slow. It is economically incompatible with modern business expectations.

The Hidden Costs of Batch Settlement

Slow settlement is expensive. A merchant waiting for payment gets locked into the working capital problem: they have shipped goods but have not received payment. They float their own cash until the payment clears. For small merchants and gig workers, this is the difference between payroll and bankruptcy. Over a year, the cost of working capital is not trivial.

Batch settlement also creates reconciliation friction. A merchant sees an order come in at 3 pm Thursday, but the payment does not arrive until 9 am Saturday. Between order and payment, the customer might dispute the charge, the shipping might get lost, the customer might call customer service. The longer the gap between transaction and settlement, the longer the dispute window, and the higher the operational cost to resolve it.

For global payments, batch settlement is catastrophic. A US company paying a contractor in the Philippines sends money through a correspondent banking chain. The money leaves the US bank T+1, arrives at an intermediary bank T+2, and reaches the Philippines T+3 or later. Three to four days minimum, often longer if the intermediary bank has processing windows or if the payment routes through multiple hops. The contractor sees money in their account a week later, deducts from their own working capital the entire time. The sender never knows if the payment reached the right person or got stuck in a queue somewhere.

---

The Interchange Cost Model is Uneconomical at Scale

Every card transaction is taxed. The tax is called interchange. It is the fee the merchant's bank pays to the cardholder's bank each time you run a card. For debit cards, interchange is typically 0.05 to 0.5 percent of the transaction value. For credit cards, it is 1.5 to 3.5 percent. For premium credit cards, it can exceed 3.5 percent.

That tax is non-negotiable. The networks set the rate and no one can undercut it. The merchant cannot negotiate a lower rate. The acquiring bank cannot offer a discount because it would violate the network rules. The cardholder pays no visible fee (it is baked into the merchant's cost structure). The tax is a structural transfer: from merchants, to card networks, to issuing banks, which distribute it as interchange revenue and rewards subsidies.

Scale makes the problem worse. A large merchant processing $1 billion in card volume annually is paying $15 million to $35 million in interchange just to move money that should move for free. That is not a payment fee. That is an invisibly large tax on commerce. At that scale, even a basis point reduction in interchange is millions of dollars. For a merchant, the incentive to find alternative payment rails is enormous.

Card networks have no incentive to reduce interchange because interchange is the revenue model. For merchants facing billions in annual interchange costs, the search for alternatives is not about payments. It is about cost structure survival.

The economic reality drives innovation. Large merchants have already shifted volume off card rails to alternative methods. Amazon, Walmart, and other mega-merchants use bank transfers, private payment schemes, and real-time payment networks wherever possible. They reserve card payments for customers who literally have no other option. For everyone building infrastructure today, the question is not "why would we use cards?" but "why would we accept this tax when alternatives exist?"

Who Pays and Who Profits

The merchant pays the cost, but the benefit of the interchange tax is not equally distributed. Interchange revenue goes primarily to card-issuing banks. The acquiring bank takes a cut. The card networks take a cut. The merchant bears the cost. The consumer sees no cost (the cost is embedded in prices).

This creates perverse incentives. Card-issuing banks have no incentive to migrate customers to cheaper payment methods. Their entire revenue model depends on transaction volume and the associated interchange revenue. Real-time payment networks and open banking APIs represent existential threats to issuer profitability, which is why large banks have been slow to embrace them.

---

Cross-Border Friction is Economically Absurd

Try to send money across international borders on the card networks and you will understand why real-time payment networks exist. A US business paying a supplier in Europe sends money through the SWIFT system. The money leaves the US bank with an intermediary bank fee ($15 to $50). The intermediary bank sends it through a correspondent banking chain, possibly routing through multiple countries. Each hop deducts a fee. The receiving bank deducts another fee. The money arrives three to five business days later, sometimes longer.

The foreign exchange margin is brutal. The merchant's bank quotes an exchange rate 2 to 3 percent worse than the real market rate. The receiving bank quotes another 1 to 2 percent margin. The merchant is losing 3 to 5 percent of the transaction amount to currency conversion margins, and that happens in addition to the direct fees.

For small cross-border transactions, the fees often exceed the transaction value. A startup paying a freelancer $500 loses $30 to $75 in fees and currency slippage. The freelancer receives $425 to $470. Both parties lose. The system is economically designed for large, bulk transfers between banks, not for individual cross-border commerce.

ISA fees (International Settlement Account fees) add another layer of cost. Correspondent banking relationships are maintained through account fees that can be hundreds of dollars per month, even for banks not using the correspondent. The cost of maintaining the correspondent banking infrastructure is so high that only the largest banks can afford it. Smaller banks are forced to use intermediaries, adding more hops and more fees to the chain.

Cross-border payments through correspondent banking cost 3 to 7 percent in fees and currency margins, take three to five days, and are structurally incompatible with the speed and cost expectations of modern commerce. This is not a feature. This is the foundation of the demand for alternatives.

The pain is global. Companies in Asia sending money to Africa face even higher costs and longer delays. The correspondent banking chain is longer. The intermediaries take larger cuts. The money might sit in settlement accounts for days waiting for batch processing windows. Emerging market businesses are essentially frozen out of affordable international commerce. This creates enormous demand for non-bank payment networks and real-time payment alternatives.

---

Pull vs. Push: The Fundamental Architecture Problem

Card payments are pull payments. The merchant pulls money from the customer's account. The customer provides card details, and the merchant (or acquirer) initiates the payment by pulling from that card. The customer's bank checks if the account has sufficient funds, authorises the transaction, and completes it.

The problem with pull architecture is customer protection. If a merchant charges your card wrongfully, the burden falls on you to dispute it, prove you did not authorise the charge, and get your money back. The fraud protection exists, but reversing a fraudulent charge requires intervention: calling the bank, filing a dispute, waiting for the investigation. The merchant has already received and kept the money.

Real-time payment networks use push payments. The customer initiates the payment by pushing money to the merchant from their account. No merchant credentials are required. No account number is shared. The merchant receives the money directly and the transaction is complete. From the customer's perspective, it is structurally safer: you control whether the money leaves your account. You initiate it explicitly. You see confirmation immediately.

But push payments create a different customer protection problem: irrevocability. Once the money is pushed, it cannot be recalled (easily). There is no chargeback process, no dispute window, no automatic reversal. If you push money to the wrong account by accident, the receiving bank must be contacted to reverse it, and many real-time networks have no straightforward reversal process. The regulatory frameworks are still being written to address this.

The Customer Protection Tradeoff

Cards excel at pull-payment customer protection. Disputes are handled through a formal chargeback process. Fraud is reversed automatically. The customer is protected because the card network backs the protection. This is why consumers prefer cards for unfamiliar merchants and high-value purchases.

Real-time payments excel at fraud prevention and customer control. You decide exactly when the money leaves. You approve the specific amount and recipient. You see immediate confirmation. Fraud is less likely because the customer explicitly controls each transaction. This is why real-time payments are safer for intended payments between known parties.

The architecture tradeoff is real. Different use cases require different architecture. This is why the future is not "cards replaced by real-time payments," but rather a portfolio of rails where each rail optimises for its use case. Cards for consumer protection and universal acceptance. Real-time payments for speed, cost, and customer control. ACH and bank transfers for batch and scheduled payments.

---

Why Card Rails Still Win (And Where They Do Not)

Despite all these limitations, card networks have not disappeared. They still process trillions of dollars annually. Understanding where card rails win is essential to understanding why new rails exist alongside them, not replacing them.

Where Card Rails Dominate

Consumer acceptance: Consumers have spent 50 years habituated to card payments. They trust them. Merchants accept cards universally. Every ecommerce site takes cards. This network effect is enormous and creates a structural moat.

Rewards and incentives: Card networks fund massive rewards programs. Every cardholder gets cash back, points, or travel rewards funded by interchange revenue. Push payments offer no built-in rewards mechanism. A customer might earn 2 percent cash back using a card on every transaction. They earn zero on a bank transfer. The incentive structure favours cards.

Merchant chargeback protection: While chargebacks are the bane of merchants, they also protect customers from fraud. A customer sees an unauthorized charge and disputes it. The money comes back. The customer does not lose. This protection is valuable for unfamiliar merchants and high-risk purchases. Real-time payments offer no equivalent protection.

Universal sign-up: You can use a card at any merchant globally without pre-registration. You do not need an account with the merchant. You do not need to set up payment permissions. You hand over your card and it works. That frictionless experience is unmatched by real-time payment networks, which often require bank account registration and merchant-specific setup.

Where Card Rails Lose Decisively

Cost: The interchange tax makes cards uneconomical for high-volume, low-margin businesses. At sufficient scale, real-time payments and bank transfers are orders of magnitude cheaper.

Speed: Real-time payment networks settle instantly. Card networks settle next day. For businesses where cash flow is critical, that 24-hour difference is operationally significant.

Cross-border payments: Correspondent banking is broken. Real-time payment networks are cheaper and faster for cross-border commerce, assuming both countries have access to networks.

B2B payments: Businesses do not want rewards or consumer protections. They want cost efficiency and speed. Business-to-business payments have been migrating off card rails for decades. Real-time payments accelerate that migration.

Programmability: Real-time payment networks expose APIs that allow merchants to build custom payment flows, routing, and settlement logic. Card networks do not permit this level of control. For developers and technologists, real-time payments offer capabilities cards simply cannot provide.

---

The Global Picture: 70+ Countries Have Live Real-Time Payment Networks

Real-time payment networks are not a future hypothetical. They exist today in more than 70 countries. They are moving significant volume. In some countries, they have already eclipsed the card networks.

The United States launched FedNow in July 2023. It processes over 100 million transactions monthly and is approaching $2 trillion in annual volume. The United Kingdom has operated the Faster Payments Service since 2008, with 18 years of operational data showing that real-time payments can work at massive scale. India's UPI network processes over 10 billion transactions per month, more than all of India's card volume combined. Brazil's PIX network reached 150 million users in three years and is now processing more volume than Brazil's card networks.

The European Union is mandating real-time payment capabilities across all member states. SEPA Instant Credit Transfer became mandatory for all banks in October 2025. Every bank in the EU must now offer instant payments 24/7/365. The regulation is not an option. It is law.

The adoption rate tells the story. When governments and regulators mandate real-time payment infrastructure, adoption accelerates rapidly. When adoption reaches critical mass, the economics become obvious. Merchants shift volume. Consumers shift expectations. Card networks become marginal for certain use cases, not the default.

---

The Economics That Drive the Architecture Shift

At the core, the shift from card rails to real-time payment networks is driven by economics. Each layer in the card stack extracts a tax: the network, the acquiring bank, the issuing bank, the correspondent banks, the currency converters. The total cost is 3 to 5 percent of transaction value for domestic payments and 5 to 7 percent for cross-border payments. This makes the card rails uneconomical for all but the lowest-volume, highest-margin transactions.

Real-time payment networks eliminate most of these layers. There is no network interchange fee. There is no acquiring bank intermediary (the merchant connects directly to their bank). The settlement is direct and immediate. The cost is often a flat fee per transaction (a few cents) or a subscription fee, not a percentage of transaction value. At scale, the savings are enormous.

For merchants, the choice is binary once real-time payments are available and their customer base accepts them. A merchant offering both payment rails will naturally shift volume to the cheaper option. As volume shifts, the economics of card processing worsen. The card networks respond by... maintaining interchange rates (since they cannot lower them), adding fraud protection fees, or introducing surcharges. None of these responses reverse the fundamental economics.

The card networks are locked into their economics because their revenue model is based on interchange. Real-time payment networks are built on a cost-plus or fee-for-service model. The economics favour real-time payments at scale, and the gap widens as volume increases.

This is not speculation. It is happening now. Global card volume is growing, but it is growing slower than real-time payment volume. In countries with mature real-time networks, card volume is flattening or declining. The network effect is reversing. As more merchants and consumers adopt real-time payments, the incentive to stay on card rails decreases. The tipping point in most developed markets is 3 to 5 years away.

---

What This Means for the Future

The shift from card rails to real-time payments is not about technology. It is about economics, regulatory mandate, and the fundamental unsuitability of batch architecture for real-time commerce. New rails exist because the old rails were not built for a world where settlement should be instantaneous.

In five years, the payment landscape will look very different. Card networks will remain relevant for specific use cases: consumer protection, rewards programs, universal merchant acceptance. But the default payment method for commerce between parties with bank accounts will be a real-time payment network or an open banking payment initiation service. The infrastructure and regulations are already in place. The volume is already shifting. The only question is how quickly the transition completes.