The onchain leg of a stablecoin payment costs cents. The payment does not. The distance between those two statements is where most stablecoin cost claims go to die, and the best evidence for that now comes from a central bank rather than from sceptics.
The end-to-end bill
A cross-border stablecoin payment has 5 cost components, and the marketing quotes only the middle one.
- On-ramp. Converting sending-side fiat into the stablecoin, through an exchange, broker, or payment institution, at a fee plus an FX spread when the fiat is not the peg currency.
- Chain fee. The transfer itself. Cents on most networks, and genuinely negligible.
- Off-ramp. Converting back to receiving-side fiat, again at a fee plus spread, plus local banking charges for the last hop into an account.
- Compliance. Travel-rule data exchange between virtual-asset service providers, sanctions screening, and the operational cost of exception handling. This scales with regulatory maturity, not with ticket size, which is why it lands hardest on small transfers.
- Float and treasury. Whoever holds the stablecoin between legs carries peg, counterparty, and yield-forgone costs. For a payment company running corridors at scale, managing that float is a real treasury function with a real cost of capital.
What the Bank of Italy found
The cleanest published audit to date is a Bank of Italy mystery-shopping study reported on 1 August 2026: researchers executed 200 USDC remittances across 10 international corridors and found total end-to-end costs ranging from 0.3% to almost 9% of the amount sent, with blockchain fees a tiny fraction of the total and exchange fees, FX spreads, and local banking charges accounting for the bulk (CoinDesk). For calibration, the World Bank's Remittance Prices Worldwide series puts the global average cost of sending 200 dollars at 6.36% in Q3 2025 (Issue 54, published April 2026). A 0.3% corridor embarrasses the incumbent average; an almost-9% corridor is worse than it. Both are stablecoin corridors. The study's own conclusion is the balanced one, and worth carrying: stablecoins have solved the speed of moving value and have not yet solved the costly last mile between crypto and local fiat, while 24/7 settlement and programmability remain real advantages.
Where the economics actually clear
Ramp costs are not uniform, and the dispersion is the story. Industry estimates, and they are vendor estimates rather than independent research, put retail off-ramps at roughly 1.5% to 2.5%, institutional API-based ramps at roughly 0.15% to 0.35%, and OTC (over-the-counter) execution for operators moving 100 million dollars a month or more at 5 to 10 basis points. Treat those numbers as reported-only, but the shape they describe is well evidenced by behaviour: stablecoin payment volume concentrates in flows large enough, or repeated enough, to amortise fixed compliance cost and negotiate institutional ramp pricing.
That is exactly what the volume data shows. Artemis Analytics' bottom-up survey work, the methodology most other estimates trace back to, put B2B stablecoin payments above 6 billion dollars a month by mid-2025, and McKinsey's 2025 estimate built on the same panel put annual stablecoin payment volume around 390 billion dollars, roughly 226 billion of it B2B, with about 60% of volume originating in Asia. Artemis itself flags the panel as a lower bound. The composition matters more than the totals: the market that has actually adopted stablecoin payments is corporate, cross-border, and Asian, not consumer point of sale.
The honest comparison, per job
- Remittances. Stablecoins win the corridors where the incumbent is worst and the ramps are competitive, with the Philippines the standing example, and lose corridors where a scheme link or a competitive incumbent already clears at low single digits. Corridor-level, not categorical.
- Cross-border B2B. At 6 to 7 figure tickets, even a retail-grade 2% ramp cost is dominated by the working-capital value of same-day finality, and institutional pricing takes the cost case from arguable to strong. This is the lane the displacement map scores for new rails.
- Point of sale. The end-to-end bill has to beat a marginal cost the consumer perceives as zero, while giving up the card bundle from Part 2. It does not, and the pitch that it does is the least defensible claim in the category.
The durable stablecoin advantages, on this reading, are the ones that survive the audit: finality on a clock that never closes, programmability, and reach into corridors the incumbent system serves badly. Cost is an outcome of corridor structure, not a property of the rail.