Multi-Currency Treasury Operations for a CASP (2026)

Single-currency treasury is the default for early-stage CASPs. Customers fund in EUR, the firm holds in EUR, withdrawals settle in EUR — one bank, one ledger, one reconciliation. Multi-currency operations is where treasury becomes a discipline rather than a back-office function. Adding USD doubles the operational complexity. Adding GBP triples it. Adding stablecoin treasury alongside fiat creates a fourth ledger with its own settlement finality, its own counterparty risk model, and its own OFAC perimeter.

The CASPs that scale are the ones that build a four-currency architecture (EUR, USD, GBP, stablecoin) by month 12 of operation, with deliberate FX execution, an explicit OFAC overlay on every USD flow, and daily three-way reconciliation across bank ledger, internal ledger, and on-chain. The CASPs that fail at scale are the ones running multiple single-currency operations in parallel — each with its own gaps, each invisible to the others, none reconciled to a central source of truth.

This guide covers the four-currency architecture, the three FX-execution models, the OFAC overlay specifically on USD flows, the reconciliation engineering, and the five most common failure patterns we see in 2026.

A single-currency CASP can run a clean treasury with a ledger, a bank, and a reconciliation cadence. A multi-currency CASP is running four parallel ledgers that must reconcile to each other intra-day, with FX exposure between them, settlement-finality differences (T+0 for stablecoin, T+1 for SEPA, T+2 for SWIFT), and a sanctions perimeter that varies by currency.

The complexity is multiplicative, not additive. Two currencies is roughly 3x the operational load of one; four currencies is roughly 10x. Treasury teams that scale by hiring more analysts and adding spreadsheets compound the failure mode. The right answer is to invest in treasury infrastructure before adding the second currency.

EUR is the home leg for any EEA-domiciled CASP authorised under MiCA¹^[1]. Settlement finality via SEPA Instant (T+0 within 10 seconds for participating banks) or SEPA SCT (T+1). Bank relationship pattern: one operating bank, one safeguarding bank, both EU-domiciled. Capital, customer EUR balances, OPEX. Standard architecture; the lowest-friction leg.

USD is the hardest leg by a wide margin. Every USD flow that touches a US-domiciled correspondent bank inherits extraterritorial OFAC obligations under the Office of Foreign Assets Control²^[2]. The CASP must screen against the OFAC SDN list with the same depth as a US bank, geo-fence sanctioned jurisdictions at onboarding, and have a documented OFAC compliance officer (often distinct from the MLRO).

Settlement finality on USD via SWIFT is T+2 worst-case, same-day for FedWire / CHIPS where access exists. The CASP without direct correspondent access pays 30–80 bps per transaction through a third-party FX house and inherits the FX house's settlement risk.

GBP requires a UK banking relationship and FCA cryptoasset registration³^[3] for any CASP serving UK customers. Faster Payments gives T+0 settlement; CHAPS gives same-day large-value. Post-Brexit complications: EU CASPs cannot passport into the UK, so a separate UK entity or appointed-representative structure is needed for any volume.

Stablecoin treasury operates 24/7 with T+0 finality on settlement (subject to chain-finality assumptions). The stablecoin leg supplements rather than replaces fiat — because fiat off-ramps still go through banks. Whitelist policy: which stablecoins are accepted (USDC, USDT, EURC, others), which issuers are on the approved list, and the MiCA EMT vs ART classification logic for each. Counterparty risk on the issuer is real and concentrated.

FX executed at the operating bank's prevailing spot rate. Lowest operational complexity, highest cost. Spreads typically 50–150 bps over interbank for major pairs, materially worse for minors. Acceptable for low FX volume; punitive at scale.

Dedicated FX provider with multi-currency multi-bank account access, executing on tighter spreads via prime-broker relationships. Spreads typically 5–20 bps for major pairs at institutional volumes. Adds a counterparty between the CASP and the bank — evaluate the broker's own balance sheet, segregation, and regulatory authorisation as carefully as a sponsor bank.

Some pairs (notably EUR↔USD via EURC↔USDC) can be executed on-chain at near-zero spread, with T+0 finality. The trade-offs: counterparty exposure to the issuers, on-ramp/off-ramp friction back to fiat, and MiCA classification of the stablecoins used. Best suited as a treasury-management tool, not for customer-facing FX.

USD touches US correspondent banks. US correspondent banks are subject to primary US sanctions regardless of where the originator or beneficiary sits. By contracting with a USD correspondent the CASP imports those obligations into its own perimeter. Three operational consequences:

Customers from comprehensively sanctioned jurisdictions (Iran, Cuba, North Korea, Syria, the Crimea region of Ukraine) cannot be onboarded for USD-denominated services. IP-level geo-fencing is the floor; document-level KYC must align.

OFAC SDN screening on customer onboarding, on every transaction, at refresh cadence (daily list updates), with secondary-sanctions screening for connected counterparties. FinCEN⁴^[4] expectations on cross-border correspondent banking apply.

OFAC publishes SDN-listed crypto addresses. Any USD-denominated stablecoin movement to or from those addresses creates a sanctions violation. The CASP must screen on-chain in real time and block the deposit or withdrawal before it settles. This is non-negotiable.

Treasury reconciliation in a four-currency CASP runs three ledgers in parallel: the bank ledger (what the bank says you have), the internal ledger (what your accounting system says you have), and the on-chain ledger (what the blockchain says you have). All three must reconcile every business day with documented break investigation and signed-off attestation.

Every break has an owner, an investigation deadline (typically T+1), a documented root cause, and a remediation. Recurring break categories — e.g. mempool-stuck withdrawals, FX rate slippage between booking and settlement, fee mis-postings — get systematic engineering attention rather than manual workarounds.

Daily reconciliation signed off by the treasury operator. Weekly attestation by the CFO. Monthly attestation to the board. Quarterly external proof-of-reserves by a Tier-1 audit firm. This cadence is what the bank's risk team will inspect at supervisory review.

Stablecoins are issuer credit, not cash. Concentration risk on a single issuer is real — USDC's brief depeg in March 2023 and the periodic concerns around USDT reserves are reminders. Treat each stablecoin as a credit exposure with documented limits, not a cash position.

Customer deposits in EUR, withdraws in USD an hour later. The CASP carries the FX exposure between booking and execution. At scale, intra-day FX positions can be materially larger than overnight ones. Document the intra-day FX limit and the hedging mechanism.

A liquidity buffer denominated only in EUR cannot meet a sudden USD withdrawal request. Buffer must be currency-matched to liabilities plus a margin for FX-execution time. Stress-test against a 24-hour USD withdrawal spike of 20% of liabilities.

Spreadsheets do not scale past two currencies. The treasury team that adds analysts every quarter to keep up with reconciliation is a team accumulating hidden breaks. Invest in automated reconciliation infrastructure before adding the third currency.

Customers onboarded under EUR rules, then later allowed to transact USD without re-screening. The result is OFAC-list customers transacting USD through the correspondent bank, and the bank's compliance team finds it before yours does. Re-screen at the moment a customer activates USD, with full OFAC SDN coverage.

When the inability to settle in a second currency is costing more than the operational investment to add it. As a rule of thumb: when more than 15% of customer demand is in a non-home currency, the second-currency build is overdue. Add the infrastructure (reconciliation, FX execution, sanctions overlay) before opening the floodgates on customer-facing flows.

Not strictly. Many EU CASPs offer USD via a USD correspondent banking relationship without a US entity — but the OFAC compliance burden is the same as if you had one. A US entity becomes necessary when serving US customers (which triggers FinCEN MSB or state MTL licensing) or when the volume justifies direct FedWire access.

MiCA Class 3 minimum is the floor (€150,000 own funds), but practical liquidity buffer is currency-matched to 24-hour redemption pressure. A reasonable starting point: 25% of customer balances in each currency held in immediately-available form, with a 24-hour stress test of an additional 20% spike.

No. Customer fiat off-ramps still require fiat banking. Stablecoin-only treasury creates single-issuer concentration risk and supervisory questions about safeguarding. Stablecoins are a treasury tool, not a treasury substitute.

Quarterly external attestation by a Tier-1 audit firm is the institutional standard in 2026. Internal Merkle-tree attestation can run monthly or even continuously. Banks now require quarterly external attestation as a condition of the relationship for any custodial CASP — expect this to harden into a regulatory expectation by AMLR application in 2027.

• How to Get a Bank Account for a VASP or CASP — the banking foundation underneath multi-currency treasury.

• Bank Diligence File for a Regulated Crypto Firm — what banks ask about treasury architecture during diligence.

• MiCA Compliance Guide for CASPs — the authorisation framework underpinning the treasury obligations.

• Bank Account for an EMI: 2026 Buyer's Playbook — the parallel three-bank treasury for non-crypto regulated fintechs.

• Banking Access for Regulated Fintechs — our service: multi-bank introductions, treasury architecture, FX broker selection.

Multi-currency treasury is not a back-office function; it is a first-class engineering and risk discipline. Build the four-currency architecture, the FX-execution stack, the OFAC overlay, and the daily three-way reconciliation deliberately — ahead of the customer demand that will otherwise force you into them under operational pressure. The CASPs that scale do this in year one. The ones that fail at scale wait until year three.