When a platform holds funds with a bank, those funds rarely sit idle. Banks deploy customer deposits into lending, securities and interbank markets. This is the fractional reserve model that underpins modern banking. It is also why the question of where your operational float sits has become a strategic decision rather than an administrative one.
For platforms managing multi-currency payments, payroll disbursements, or marketplace settlements, the choice of custody architecture directly affects counterparty risk, liquidity access, and settlement speed. A growing number are shifting away from traditional bank-held float toward 100% reserve models. The reason is not ideology. It is operational.
How bank-held float actually works
When a platform deposits funds with a universal bank, those funds enter the bank's general balance sheet. The bank owes the platform a liability, but the underlying assets are deployed elsewhere. This is the foundation of banking economics.
The model works until it doesn't. When Silicon Valley Bank collapsed in 2023, platforms with operational accounts faced immediate liquidity freezes. The funds were nominally theirs, but access depended on the bank's solvency. The FDIC's resolution process ultimately protected most depositors, but the operational disruption was severe. Payroll runs stalled. Settlement cycles broke. Platforms learned that "funds on deposit" and "funds available" are not the same thing.
This is the counterparty risk embedded in fractional reserve custody. The platform's liquidity is entangled with the bank's balance sheet health, lending decisions, and risk appetite.
What 100% reserve architecture changes
A 100% reserve model operates differently. Client funds are held in segregated accounts, fully backed by cash or cash-equivalent assets. The custodian does not lend against those balances. There is no maturity transformation, no deployment into higher-yielding instruments, no balance sheet entanglement.
The operational implications are significant:
- Counterparty risk is structural, not reputational. The platform's funds are not exposed to the custodian's lending book or investment decisions. Segregation is architectural rather than contractual.
- Liquidity is immediate. Funds are not locked in instruments that require unwinding. When the platform needs to move capital, it moves.
- Insolvency treatment is cleaner. Segregated client funds typically sit outside the custodian's estate in bankruptcy. Recovery does not depend on resolution proceedings.
This model sacrifices yield. A bank holding float can offer interest because it earns a spread on deployment. A 100% reserve custodian cannot offer the same return because it does not take the same risk. For platforms optimizing treasury yield, this is a cost. For platforms optimizing operational resilience, it is the point.
The settlement speed connection
Reserve architecture also affects payment velocity. When a bank holds float, settlement timing is influenced by the bank's own liquidity management. Funds may be batched, queued, or delayed to align with internal treasury operations. The Bank for International Settlements has documented how correspondent banks' economic incentives create friction in cross-border settlement.
A 100% reserve operator has no yield incentive to delay. Funds in segregated accounts do not generate return while sitting still. The operator's economics favor throughput, not duration. Settlement speed becomes a feature rather than a tradeoff.
This distinction matters most for platforms with high transaction volumes or time-sensitive disbursements. Payroll platforms cannot explain to workers that their salary is delayed because the bank batched settlement. Marketplace operators cannot tell merchants their funds are queued behind the bank's intraday liquidity targets.
Who is adopting this model
The shift toward 100% reserve custody is not theoretical. It is visible in how platforms are restructuring their treasury operations.
Payment service providers increasingly separate operational float from yield-generating reserves. The float that funds daily settlement sits in segregated, instantly accessible accounts. Longer-duration capital may remain with traditional banks for yield, but the operational layer is isolated from balance sheet risk.
Payroll and EOR platforms face particularly acute pressure. Their funds are not discretionary capital. They are wages owed to workers, often across multiple jurisdictions. The reputational and regulatory cost of a liquidity freeze is asymmetric. These platforms are leading adopters of reserve-backed custody.
Fintech infrastructure providers have responded to this demand. Lorum operates a 100% reserve model with direct access to local clearing rails across 30+ markets. Client funds are held in segregated accounts, not deployed for yield or entangled with lending operations. The architecture is designed for platforms that cannot afford to discover their liquidity is illiquid.
The tradeoff is narrowing
100% reserve custody is not free. Platforms forgo the interest income that bank-held float can generate. For some treasuries, that yield matters. The decision depends on what the platform optimizes for.
But the yield gap is narrowing. Platforms increasingly have access to tokenized money market instruments that allow treasurers to deploy idle balances without surrendering custody or introducing balance sheet entanglement. In practice, this means a platform can hold operational float in a segregated reserve account and separately allocate surplus capital to short-duration instruments that remain under the platform's control. The decision to pursue yield remains with the platform, not the custodian. This preserves the segregation architecture while giving treasury teams optionality they previously lacked.
The relevant questions are:
- What is the cost of a 24-hour liquidity freeze to your operations?
- How much counterparty risk are you comfortable embedding in your payment infrastructure?
- Does your treasury function prioritize yield or availability?
There is no universal answer. But the platforms that experienced SVB, or watched peers scramble during the 2023 banking stress, increasingly answer those questions the same way. The yield on float is measurable. The cost of losing access to it is harder to price until it happens.
100% reserve models do not eliminate all risk. They eliminate a specific category of risk: the entanglement of operational liquidity with a custodian's balance sheet. For platforms where payment continuity is non-negotiable, that tradeoff is becoming easier to make.
