HHooww  iitt  wwoorrkkss
PPrroodduuccttss
AAbboouutt
DDeevveellooppeerrss
FFAAQQ
BBlloogg
RReeqquueesstt  aacccceessss
HHooww  iitt  wwoorrkkss
PPrroodduuccttss
AAbboouutt
DDeevveellooppeerrss
FFAAQQ
BBlloogg
RReeqquueesstt  aacccceessss

Stablecoin treasury management for fintechs

Lukasz Dec

Co-founder and Chief Executive Officer

The moment a fintech starts settling on stablecoin rails, its treasury function inherits a set of questions it never had to answer for bank balances: how much token float to hold and on which chains, who custodies it and how, what happens if an issuer wobbles, why the balance earns nothing, and how to reconcile positions that live in three places at once. None of these is exotic. All of them are new to most finance teams. This is the practical guide, written for the CFO or head of treasury at a payments company or platform, not for a crypto desk.

Float: how much, and where

The first instinct is to hold as little as possible, and it's the right one: regulated payment tokens are movement instruments, not savings instruments. But "as little as possible" has a floor set by operations: enough to fund payouts between top-ups, enough to absorb timing gaps between fiat inflows and token outflows, and enough on each chain and in each currency token you actually settle in.

The float question is really three sub-questions:

Dimension

The question

The trade-off

Size

Days-of-payouts to hold as tokens

Lower float = less issuer/custody exposure; higher float = fewer top-up cycles and less risk of a missed cut-off

Currency

Which tokens — USD, AED, EUR, others

Native-currency float removes the dollar detour but concentrates you in thinner pairs

Chain

Where the balances sit

Multi-chain float serves more counterparties; more chains = more custody surface and more rebalancing — the N×N problem landing on the balance sheet

A useful discipline: treat token float like nostro balances, because that's what it is — working capital parked at the point of settlement. Nostro management has a century of practice behind it (target balances, sweep rules, intraday limits); apply it.

2. Custody: who holds the keys

Bank money is custodied by the bank. Token float is custodied by whoever controls the private keys, and that's a design choice with real consequences:

  • Self-custody (your keys, your infrastructure) maximizes control and maximizes operational risk; a fintech running payroll-scale flows on self-managed keys is running a security operation on the side.

  • Qualified/regulated custodian moves key management to a supervised specialist - segregation, insurance, controls - at the cost of a counterparty and an integration.

  • Settlement-provider custody keeps float within the licensed perimeter of the party executing your settlement - the model where the same entity converts, holds, and transfers under one supervisory umbrella (the custody & transfer activity the PTSR licenses).

Whichever you choose, the treasury questions are the same as for any custodian: segregation of client assets, insolvency remoteness, audit and attestation, and - for tokens specifically - the key-management architecture (MPC vs. multisig vs. HSM) and who can move funds under what approvals. If a provider can't answer those crisply, they aren't a custodian; they're a wallet.

Issuer and depeg risk

A stablecoin is a claim on its issuer. Treasury has to underwrite that claim the way it would underwrite a bank: reserve composition and where reserves are held, redemption terms, regulatory status, and concentration. Two facts frame the risk. First, regulated regimes now mandate full backing, segregation, and redemption at par - the licensed dirham and euro tokens carry supervision unregulated tokens never had. Second, even a fully backed token can trade below par under stress: USDC briefly lost its peg during the March 2023 US banking turmoil when part of its reserves sat at a failed bank, and recovered as reserves were confirmed accessible. Fully reserved is not the same as unconditionally liquid at par.

Practical rules: diversify across issuers where liquidity allows; prefer tokens with transparent, frequent attestations and regulated status in your jurisdiction; know your redemption path (who, how fast, at what cost) before you need it; and set exposure limits per issuer just as you set them per bank.

The no-yield reality

Finance teams used to sweeping idle balances into overnight instruments meet an unfamiliar constraint: regulated payment stablecoins can't pay you for holding them. The UAE's PTSR and the EU's MiCA both prohibit issuers from paying interest or time-based rewards - the rule that keeps a payment token from becoming a shadow deposit.

The consequence is a treasury design principle: stablecoin float is a cost center, and its cost is the yield forgone. That makes the float-minimization discipline in section 1 an economic imperative, not just a risk preference - and it means the returns from stablecoin settlement come from speed, cost, and reach, not from the balance itself. Products marketed as "yield-bearing stablecoins" are typically tokenized funds or unregulated structures; treat them as investments with a different rulebook, not as float.

Multi-currency and reconciliation

Once you hold AED tokens, USD tokens, and euro tokens across two or three chains, you have a multi-currency, multi-venue book. Two disciplines keep it sane:

FX policy for tokens. Decide which currency exposures you hold deliberately and which you convert immediately; treat token conversions as FX events with rates, spreads, and timestamps; and prefer direct pairs where depth allows, dollar legs where it doesn't - with routing that minimizes the detour rather than defaulting to it.

Reconciliation across three ledgers. Your bank statements, your on-chain balances, and your settlement provider's ledger must agree, continuously. On-chain data is unusually good for this - every movement is timestamped and referenceable - but only if references are attached at origination and your provider surfaces them per transaction. Ask for per-payment identifiers, real-time balance visibility, and exportable ledgers before you sign; reconciliation retrofitted after go-live is the most expensive kind.

Putting it together

The teams that run stablecoin treasury well don't treat it as crypto. They treat token float as nostro balances, custody as a counterparty decision, issuers as banks to be underwritten, no-yield as a design constraint, and on-chain data as the reconciliation gift it is. Most of that discipline already exists in a good treasury function - the settlement layer's job is to make the token-specific parts invisible: custody within a licensed perimeter, conversion with visible rates, per-transaction references, and float that goes to work the moment it's needed and not a minute before.

Frequently asked questions

How much stablecoin float should a fintech hold?
As little as operations allow - typically a target of a few days of payouts, sized per currency token and per chain, managed like nostro balances with top-up rules and limits.

Can a company earn interest on stablecoin balances?
Not from regulated payment stablecoins - the UAE's PTSR and the EU's MiCA both prohibit issuers paying interest or time-based rewards. Yield-bearing products are typically tokenized funds or unregulated structures with different risk and rules.

Who should custody a fintech's stablecoins?
Either a regulated custodian or the licensed settlement provider executing your flows - under clear segregation, attestation, and key-management controls. Self-custody at scale is a security operation most fintechs shouldn't run.

What is depeg risk?
The risk a stablecoin trades below its reference value - even a fully reserved token can, briefly, under stress. Manage it with issuer diversification, exposure limits, and a tested redemption path.

How do you reconcile stablecoin balances?
Continuously, across bank statements, on-chain balances, and your provider's ledger - using per-transaction references attached at origination and real-time balance visibility.

Sources