The Real Cost of a SWIFT Transfer in 2026: Fees, Correspondents and FX

How much a SWIFT transfer really costs is a question almost no CFO in LATAM can answer precisely. The fee charged by the sending bank is visible. Everything else is not: the deductions taken by correspondents in transit, the margin embedded in the exchange rate, the receiving bank's charge and the cost of tracing a delayed payment. The practical result is familiar to any treasury team: the supplier receives less than what was sent, and nobody knows in advance how much less.

At Soulbit Academy we analyze that cost with verifiable data, not marketing figures. We use the World Bank price monitor, the BIS correspondent banking statistics and the targets of the G20 program coordinated by the FSB. And at the end we compare, with the same honesty, the cost structure of the stablecoin rail, which is not free but is transparent before you confirm.

Why nobody knows what an international payment costs

The opacity is not an accident: it is a consequence of the architecture. A SWIFT transfer is not a single movement of money. It is a chain of messages between banks that hold accounts with each other, and every link in that chain sets its own price independently.

The sending bank charges its fee when the payment is ordered. Intermediary banks can subtract charges in transit that the sender never sees. The receiving bank can charge for crediting the funds. And the currency conversion carries a margin that is not invoiced as a fee, but built into the rate applied.

None of these actors has an effective obligation to show the total cost before executing the payment. In fact, the G20 program includes a specific transparency target: that providers disclose the total transaction cost, the delivery time and payment tracking. That regulators had to set that goal for 2027 says a lot about the starting point.

For a company paying suppliers or international payroll every month, that opacity has a direct financial cost. You cannot budget accurately for something you only learn when the payment arrives.

The cost stack, layer by layer

It helps to separate the price of an international transfer into its real components. There are at least six, and only the first is fully visible when the payment is ordered.

First, the issuing fee of the sending bank. Second, the correspondent deductions, known as lifting fees: each intermediary in the chain can subtract its charge from the amount in transit. Third, the FX margin whenever there is a currency conversion. Fourth, the receiving bank's fee upon crediting. Fifth, investigation or tracing charges if the payment is delayed and has to be located. Sixth, the financial cost of time: cash locked up while the payment travels.

Why does the beneficiary receive less than what was sent?

Because several of these layers are charged by deduction, not by invoice. Correspondent deductions and the receiving fee come out of the transferred amount itself. The company orders one figure, the supplier receives a smaller one, and the difference is split among actors the sender never chose and does not know.

Cost component	Who applies it	How it shows up
Issuing fee	Bank sending the payment	Visible when the transfer is ordered
Correspondent deductions (lifting fees)	Each intermediary bank in the chain	Subtracted in transit; the beneficiary receives less
FX margin	Whoever executes the currency conversion	Embedded in the exchange rate, no separate line
Receiving fee	Beneficiary's bank	Deducted when the funds are credited
Tracing and investigation	Sending bank, at the company's request	Extra charge if the payment is delayed and must be located
Cost of time	The company itself	Cash locked up during business days in transit

The FX margin, the largest and least visible cost

Of all the layers, the exchange rate margin tends to be the most expensive and the least understood. It never appears as a fee on any statement. It is the difference between the interbank rate at that moment and the rate actually applied to the transaction.

The World Bank considers this so central that its methodology for measuring sending costs always includes two components: the explicit fee and the exchange rate margin over the reference rate. Without the second, the figure would be incomplete.

The monitor's own data confirms where it hurts most. In the third quarter of 2025, banks were the most expensive type of provider for sending money abroad, with an average cost of 14.99% according to Remittance Prices Worldwide. That figure measures personal remittances of 200 dollars, not corporate payments, but it illustrates the pattern: the traditional banking channel is the most expensive in the measured market.

For a B2B payment the margin percentage is usually lower, because the amounts are larger. The logic, however, is identical: if the company does not know the interbank rate at the moment of the transaction, it cannot know how much it paid to convert. And if the rate is set after the payment is ordered, it cannot even negotiate it.

Correspondents: fewer banks, longer chains

International payments depend on the correspondent banking network, and that network has been shrinking for more than a decade. According to data from the payments committee of the BIS, the number of active correspondent banks fell by about 22% between 2011 and 2019, and active corridors declined by around 12% over the same period.

The regional figure is the one that matters most to a CFO in LATAM: Latin America recorded the largest regional decline in correspondent relationships, with a 34% drop over that period, compared with 13% in North America. Fewer correspondents means longer, more concentrated payment chains to reach the same destination.

The operational consequences are direct. More links mean more points where a charge can be deducted in transit. They also mean more cut-off windows, more sequential compliance checks and more places where a payment can sit for days without any visible explanation. A cross-border transfer can take between one and five business days depending on the corridor and the number of intermediaries, and a payment launched on Friday may not settle until the following week.

When that happens, tracing the payment carries its own price: the investigation is requested through the sending bank and usually generates an additional charge. It is the only industry where finding out where your money is costs money.

What World Bank data and the G20 targets say

The best public data on the cost of the traditional rail is published by the World Bank. Its Remittance Prices Worldwide monitor measured a global average cost of 6.36% for sending 200 dollars in the third quarter of 2025, a slight improvement on the 6.49% recorded in the first quarter of the year.

Are remittance figures useful for a B2B payment?

Useful as a thermometer, not as a budget. They measure small personal transfers, where fixed costs weigh more than in a corporate payment. But they run over the same rail, the same correspondents and the same FX margin structure. If the rail were cheap and transparent, those figures would not have spent two decades refusing to come down.

The opposite reference point is the official targets. The G20 program, coordinated by the FSB, sets a global average cost of no more than 1% for retail cross-border payments by the end of 2027, with no corridor above 3%. For 200-dollar remittances, the target is 3% on average in 2030, with no corridor above 5%. On speed, the goal is for 75% of payments to reach the beneficiary in less than one hour.

The gap between the measured 6.36% and the 3% target speaks for itself. And the FSB's consolidated progress report of October 2025 acknowledges it plainly: the efforts have not yet translated into tangible improvements for end users, the average cost remains rigid and satisfactory improvements at the global level are unlikely within the 2027 timeline.

The cost structure of the stablecoin rail, without the marketing

The alternative gaining ground in B2B treasury is settling in digital dollars on blockchain. USDC is a digital dollar issued by Circle, backed by cash and short-term US Treasury bonds, with independent monthly attestations. We develop the full comparison between the two rails in SWIFT vs stablecoin for international payments.

Is paying in stablecoin free?

No, and it is worth saying so clearly. The stablecoin rail has two cost components. The first is the blockchain network fee, which is paid per transaction and does not depend on the amount sent. The second is the conversion between fiat and stablecoin on the way in or out, which has its own price in the form of a quote.

The structural difference is not that the cost is zero. It is that the total cost can be known before confirming. There are no correspondents deducting charges in transit, no margin embedded in a rate set after the fact, and settlement happens in minutes, any day and at any hour. There is also an honest trade-off: a confirmed transfer is irreversible, without the bank recall of the traditional rail.

Dimension	Traditional SWIFT transfer	Stablecoin rail (USDC)
Fee structure	Issuing, correspondents, receiving and tracing, in layers	Network fee per transaction plus a conversion quote
FX margin	Embedded in the applied rate, no breakdown	Quote visible before confirming the transaction
Total cost known upfront	No; deductions are known when the payment arrives	Yes; network plus conversion are visible before sending
Settlement time	1 to 5 business days depending on the corridor	Minutes, any day and hour
Traceability	Limited; tracing a payment can cost money	On-chain record verifiable immediately
Reversibility	Bank recall and mediation possible	Irreversible once confirmed

Where Soulbit fits and what its V1 delivers

Soulbit is a stablecoin payment and treasury rail for companies, and its relationship with this problem is direct: it turns the cost of an international payment into a price known before confirming.

The real V1 offers a business account with balances in USDC and USDT, and fiat in USD, EUR and GBP. Conversion between crypto and fiat works on a quote-on-request basis: the company asks for a price, sees it on screen and decides whether to confirm. There is no hidden FX margin and no deductions in transit. It includes KYB verification, batch payments for payroll and suppliers, payment links and QR collection. The local banking rail exists today only in Colombia; in other countries, funding from local currency and cash-out are handled by the company on its own.

Just as important is what V1 does not do: it offers no cards, no yield on balances, no proprietary token and no native mobile app. To understand the full platform, you can read what Soulbit is and how it works. And every payment is recorded on-chain with a unique identifier, which simplifies the accounting close, as we explain in reconciling stablecoin payments in your accounting.

The honest balance: the traditional rail charges in opaque layers and settles in days; the stablecoin rail charges in two visible components and settles in minutes. For a CFO, the question is no longer just what each rail costs, but which of the two lets you know before you pay.