Follow a single $50 card payment from the moment a shopper taps to the second a merchant can spend the money, and it passes through at least five hands in under two seconds, then waits a day or two to actually settle. That gap between speed and finality is the heart of how payments systems and infrastructure work in the US financial market. The United States runs several parallel networks that together clear more than $2 trillion a day, according to The Clearing House, and each one solves a different problem. This guide traces the money through the machine.
The four core networks
American payments run on four main rails. Card networks (Visa, Mastercard, and others) route retail purchases between a shopper’s bank and a merchant’s bank. The automated clearing house, or ACH, batches payroll, bills, and transfers and settles them on a schedule. Wire systems move large, time-critical sums with same-day finality. And instant rails, the RTP network and the Federal Reserve’s FedNow service, settle individual payments in seconds at any hour.
Each rail exists because the others fall short somewhere. ACH is cheap but slow. Wires are fast but expensive and manual. Cards are convenient but carry interchange costs. Instant rails are quick and final but irreversible. A modern payment stack uses software to send each transaction down the rail that fits its cost, speed, and risk profile, an approach explored in this guide to ERP-centric payments and treasury.
Authorization, clearing, settlement
Every payment runs through three phases. In authorization, the payer’s bank confirms the funds or credit line exist and places a hold. In clearing, the two banks exchange and reconcile the transaction details. In settlement, the actual money moves between the banks, often through accounts held at the Federal Reserve.
On a card payment these phases are separated in time, which is why a pending charge can vanish or change. On an instant-rail payment they collapse into one moment: authorization, clearing, and settlement happen together in seconds, which is what makes the payment final and unrecoverable.
The behind-the-scenes plumbing for instant payments is busy. The RTP network alone averaged 1.18 million payments a day in 2025 and processed 107 million payments in a single quarter, equal to roughly 98% of all instant bank-to-bank payments in the country, per The Clearing House’s Q2 2025 network report.
The players in the chain
A retail payment touches several specialized firms. The merchant’s payment processor captures the transaction. An acquiring bank holds the merchant’s account. The card network routes the message. The issuing bank approves or declines. And a settlement layer moves the funds. Fintech platforms increasingly bundle these roles behind a single interface, but the underlying handoffs remain.
The table below traces a typical card purchase through the chain.
| Step | Who acts | What happens |
|---|---|---|
| 1. Tap | Shopper and terminal | Card data captured and encrypted |
| 2. Route | Processor and network | Request sent to issuing bank |
| 3. Authorize | Issuing bank | Funds checked, hold placed, approval returned |
| 4. Clear | Acquirer and issuer | Records matched, usually end of day |
| 5. Settle | Banks via Fed accounts | Money moves, 1-2 days later |
Source: industry payment-flow standards.
Where data rides along
Older rails carried little more than an amount and an account number. The shift to the ISO 20022 messaging standard changes that by letting a payment carry structured data, such as an invoice number, a tax reference, or remittance details, in the same message as the money.
This matters because reconciliation, the work of matching incoming money to the right invoice, has long been manual. Rich data lets software do it automatically. For businesses processing thousands of payments, that is the difference between a finance team that chases payments and one that supervises a system that reconciles itself.
How instant rails change the flow
Instant rails compress the whole sequence. Because settlement is immediate and final, there is no clearing window and no settlement risk between banks. The trade-off is that the payment cannot be clawed back, so fraud controls have to run before the money moves rather than after.
That single change reshapes product design. Payroll can run on the day it is due. Marketplaces can pay sellers the moment a sale closes. And businesses gain precise control over when cash leaves and arrives, which is why more than 340,000 firms now use instant rails each month.
Why the US runs several rails at once
Other countries built a single dominant instant-payment system and pushed everyone onto it. The United States took a different path, layering new rails on top of old ones rather than replacing them. The result is more complexity but also more choice, since a business can keep cheap ACH for predictable bills while adopting instant rails only where speed earns its keep.
This coexistence is deliberate. ACH still moves the overwhelming majority of payroll because it is cheap and reversible, which protects against errors. Wires remain the rail of record for property closings and large corporate transfers. Cards dominate retail because of their fraud protection. Instant rails fill the gap for payments that need to be both fast and final. Each survives because it is genuinely better at its own job.
For anyone designing a payment product in America, the practical takeaway is that there is no single rail to learn. The skill is knowing the map, the cost, speed, and reversibility of each network, and routing money accordingly. That map, not any one rail, is the real infrastructure.
The raw rails move money, but the experience consumers and businesses actually use is built in software above them. Payment orchestration platforms decide which rail a transaction should take, retry failed payments, and present a single record regardless of the network underneath. This is why two apps can feel identical while routing money completely differently.
That software layer is also where cross-border and B2B complexity gets managed. Paying an overseas supplier, for example, may combine domestic rails, foreign-exchange conversion, and a partner network abroad, a process broken down in this overview of B2B cross-border payment solutions and in this guide to paying suppliers in China. The rails are standardized; the orchestration is the hard part.
For a business, the lesson is that choosing a payment provider is really choosing an orchestration layer. The rails are largely the same for everyone, so the difference between providers is how intelligently they route, reconcile, and protect each payment that crosses them.
Seen up close, the US payment system is not one machine but several, each tuned for a different job and increasingly stitched together by software. The direction of travel is clear: toward payments that settle in seconds and carry their own paperwork, leaving the slow, blind transfers of the past behind.
