One settlement rail standard, not two separate bets

Two decisions tend to get made separately once a company starts treating digital rails as real infrastructure instead of a curiosity. Treasury sets a policy for holding stablecoins as working capital — which issuer, how much, under what custody arrangement. Separately, whoever runs cross-border payments starts routing a growing share of them off traditional wires, onto whichever rail is fastest or cheapest for a given corridor that week.

Both decisions are reasonable on their own terms. The problem is that they're answering the same underlying question — which counterparties and rails does this company actually trust with its money — from two different desks, with two different vetting processes, and frequently with no shared list between them.

The same counterparty, vetted twice (or not at all)

In practice, this plays out in a predictable way. Treasury does real diligence before putting a stablecoin issuer on the balance sheet — reserve composition, redemption history, custody arrangement. Payments, meanwhile, picks a rail for a specific corridor based on speed and fee, and that rail may run through the exact same issuer treasury vetted — or a different one treasury has never looked at, because from the payments desk's seat, it's a transaction choice, not a balance sheet decision.

Neither team is being careless. But the company now has exposure to a counterparty through a channel nobody formally approved, sitting next to a counterparty that went through real diligence — and from the outside, both look identical: money moving through a digital rail.

Building the shared approval list

This doesn't mean centralizing every payment decision through treasury, and it doesn't mean slowing down a payments team that's found a genuinely faster corridor. It means making sure both sides are drawing from the same, already-vetted set:

• One list of approved rails and issuers. Whatever treasury has already done diligence on for balance sheet purposes is the same list payments should be routing through — and vice versa, since a rail payments trusts for volume is worth treasury's diligence too.
• A named owner for additions. New rail, new corridor, new issuer — someone signs off before it's used, in either direction, not after a quarter of volume has already gone through it.
• Separate limits, same list. Treasury's holding limit and payments' per-corridor concentration limit can be different numbers — they just need to reference the same approved counterparties.
• A review cadence tied to volume, not the calendar. A corridor that goes from pilot to meaningful volume should trigger a review before the next scheduled one, not wait for it.

A treasury policy that doesn't talk to the payments desk isn't cautious. It's vetting a counterparty once and trusting it twice.

Why this gets harder as more corridors come online

With one or two corridors, an informal conversation between treasury and payments usually catches a mismatch before it matters. The risk grows with the fifth and sixth corridor, each one added because it solved a specific, local problem — a slow partner bank, a currency with thin correspondent coverage — with nobody stepping back to check whether the new rail's issuer is one treasury has actually looked at. The exposure accumulates quietly, corridor by corridor, until a counterparty question in a board meeting reveals that finance can't produce one list of who it's actually trusting with company money.

What this doesn't require

• A crypto treasury function. This is ordinary counterparty-risk discipline, applied to a newer category of counterparty — the same instinct finance already applies to banking relationships.
• Giving up speed. A corridor that's genuinely faster stays fast — the standard just means its issuer was looked at once, by someone, before volume scaled.
• New software. A shared document listing approved rails, issuers, limits, and an owner is the entire starting point.

The balance sheet policy and the payments choice, under one roof

Holding stablecoins deliberately and routing payments through faster rails are both the right instinct — digital settlement infrastructure is genuinely better than what it's replacing in a growing number of corridors. What keeps that advantage from quietly becoming an unexamined concentration of counterparty risk is making sure the list of who's trusted is the same list, no matter which desk is asking.