Most teams that ship payments think about treasury too late. They model the fee, the FX margin, and the customer experience, then discover that the rail demands real money sitting in real accounts before anything moves. That money is not a fee. It is working capital you have committed and cannot use for anything else, and it scales with your volume.
This lesson is about the three forces that govern that committed capital: float, prefunding, and settlement-timing risk. They are not exotic. They are the plumbing that decides whether your unit economics survive contact with reality.
Float is the gap between two timestamps
Float exists because payment systems debit the sender at one moment and credit the recipient at another. The money is real, but during that window nobody can spend it. Whoever holds it earns on it, and whoever is waiting for it has a hole in their cash position.
That gap is not an accident or a bug. It is a design feature of clearing and settlement. Batch rails like ACH were built around windows and netting, not instant movement, so the delay is baked into how funds clear.
A worked example
Say you run a payout product moving $10 million a month cross-border, and funds spend on average one day in transit. At any moment, roughly $333,000 of customer money is mid-flight ($10 million divided by 30).
If you have to fund that float from your own balance sheet, value it at your cost of capital. At a 5 percent annual cost, $333,000 of permanently-in-transit money costs you somewhere in the range of $17,000 a year just to keep the lights on. Push to two days of float, or run higher volume, and the number climbs in lockstep. The lesson: float is a balance-sheet line, not a rounding error, and it grows with you.
Prefunding is capital you commit before the payment clears
Many rails will not move your money until you have already parked money with them. This is prefunding, and it is the single most underestimated cost for builders.
On instant rails the requirement is structural. The Clearing House's RTP network settles through a joint, prefunded account at the Federal Reserve Bank of New York, and every participating bank must keep that pooled balance topped up to cover its outgoing payments. FedNow takes a different route, settling through a participant's own Fed master account or a correspondent's, but the same truth holds: instant settlement means funds have to be sitting there in advance, because there is no overnight batch to net against.
The pattern repeats one layer down. Lenders and fintechs that originate ACH are frequently required by their sponsor bank or processor to prefund a settlement account before a batch clears. You are setting aside cash to back payments that have not yet left, on top of whatever float you already carry.
Cross-border makes it worse
To settle quickly in a foreign market, you generally hold local-currency balances in that market, often in nostro accounts. That capital is stranded by geography. It cannot fund your home-market operations, it is exposed to currency moves while it sits, and you need a separate pool for every corridor you open. This is why a new corridor is a treasury decision before it is a product decision, and it is the thread the cross-border and stablecoin-corridor lessons later in this module pick up.
Settlement-timing risk is the cost of missing a window
Settlement-timing risk is the danger that money arrives, or fails to arrive, at the wrong moment relative to your obligations. It comes from cutoffs, return windows, and the mismatch between when you owe and when you get paid.
Cutoffs are earlier than you think
Same Day ACH runs on fixed windows. The Federal Reserve delivers same-day funds at 1:00 p.m., 5:00 p.m., and 6:00 p.m. Eastern, fed by submission windows during the day, with the final file deadline at 4:45 p.m. Eastern. Critically, your bank sets its own customer-facing cutoffs, and they are usually earlier than the network's, sometimes early afternoon. Miss your bank's cutoff and the payment slides to the next business day, which can break a payroll run or a settlement promise even though nothing failed.
The per-payment ceiling matters too. The Same Day ACH limit is currently $1 million, with Nacha having approved an increase to $10 million that takes effect September 17, 2027. If you are moving large sums today and assume the higher limit, you will split files and recompute your timing.
Faster returns cut both ways
Quicker rails surface failures faster, which is good for cash management because you learn sooner that a debit bounced. The trap is that you also have less time to react before the reversal lands in your settlement account. If you have already paid out against an inbound debit that then returns, you are short, and you are short fast. On instant rails this compresses to near zero, so your reconciliation and exception handling, the subject of a later lesson, have to keep pace with settlement.
Putting it together
Picture an instant-payout product. A customer's funds clear to you on a one-day ACH cycle, but you promise them money in seconds on RTP. To honor that, you prefund the RTP joint balance, so you are fronting the payout before the customer's deposit has settled. Now layer in same-day return risk on the inbound leg: if that ACH debit returns, you have already pushed out an irreversible instant payment. Your prefunded buffer is what absorbs the gap, and the size of that buffer is set by your timing mismatch, not your revenue.
That is the core mental model. Float is money you cannot use while it moves. Prefunding is money you commit before it moves. Settlement-timing risk is the penalty when the money moves at the wrong time. Size all three before you launch a corridor or a faster-payout feature, because the rail will size them for you if you do not, and it will do it on the day you can least afford the surprise.