A payment that arrives on a chain your settlement partner does not support is not a payment yet. Normalising is the act of moving it onto one that works — issuer-native, never through a wrapped token, and priced at what it cost us.
Circle’s CCTP does this for USDC across 17+ chains; Tether’s USDT0 does it for USDT. The issuer destroys the token on the source chain and issues a genuine one on the destination. What arrives is real USDC or real USDT, backed by the issuer, not a claim on anybody’s contract.
A wrapped token is a receipt for value held in a contract, worth face value only while that contract holds. Bridges have lost more money than any other category in crypto. Where no issuer-native path exists for a pair, we do not support the pair — we do not substitute a riskier one and hope.
And we will never write our own. A contract we authored, holding your funds while they are in transit, is custody in substance whatever we call it in the marketing. The whole regulatory position of this company rests on not doing that.
Cross-chain has more cost surface than an ordinary transfer. Every part of it is passed through at what it cost us, on its own line.
Paid on the chain the money arrived on. Ours to fund and ours to manage; yours to see.
What the issuer’s own path charges. Not marked up, because marking it up would make the route choice quietly ours to profit from.
Where any exists. Recorded as measured, not as estimated.
Paid on the chain it lands on. Same rule.
Transport is mostly a fixed cost, so it matters at $2,000 and barely registers at $100,000. That is a real limit on who this works for, and we would rather put it on the page than discover it with you.
If your client paid on a rail the settlement partner already accepts, there is nothing to move. We say so, skip the step, and charge nothing for it.
That costs us the fee every time it happens, and it is still the right answer. A router that invents work for itself is not a router, and the first customer to notice would be right to leave.
Illustrative. The first one costs nothing to route because there was nothing to route.
A transfer submitted with no arrival and no failure is the single most dangerous moment in this product.
Re-bridging $40,000 that quietly landed sends $80,000, and one of them is unrecoverable. There is no version of that we can apologise our way out of.
Poll the source chain, the destination chain and the transport’s own status endpoint until the state is actually known. Slower, and correct.
Escalation to a person, with a plain explanation of exactly what we do and do not know. Nothing resolves itself on a timer.
Every value-moving call carries an idempotency key, so a retried request returns the original result rather than making a second transfer.
Tell us the chain, the asset and roughly the size. We will tell you what path we would have used and what the transport would have cost.
