Single-Custodian Concentration in Stablecoin Reserves
On March 11, 2023, USDC traded as low as 0.86 dollars — roughly 1,400 basis points below par — because 3.3 billion of its roughly 40 billion in reserves, about 8 percent, sat as cash at Silicon Valley Bank when it failed. The remaining 77 percent in short-dated Treasuries was never at risk, and USDC re-pegged by Monday's open once the FDIC backstop landed; no holder who waited out the weekend lost principal. That episode is the load-bearing data point in stablecoin reserve analysis: a small, concentrated cash sleeve at a single failed custodian moved a 40-billion-dollar token 14 percent off peg. Three years later USDC circulation is near 73 billion, the cash roster has been deliberately spread across banks — and the concentration has migrated. The majority of USDC reserves now sit in the Circle Reserve Fund, a single SEC-registered 2a-7 government money market fund managed by BlackRock and custodied at BNY Mellon. For an allocator, the question is not whether a stablecoin is transparent. It is where its reserve breaks first.
The Trigger and the Mechanics
Concentration risk in a stablecoin reserve has two distinct forms, and they fail through different doors. The first is the cash sleeve: the ~20 percent held as bank deposits for minute-to-minute redemption liquidity. A deposit above the FDIC limit at a failing bank is an unsecured claim, and when that sleeve concentrates at one bank, a single failure freezes the redemption buffer — the SVB mechanism exactly. The second is the reserve vehicle: the ~80 percent in Treasuries, now routed through one BlackRock-managed fund at one custodian. This sleeve is structurally safer — 2a-7 constraints, sub-60-day weighted-average maturity, Treasury collateral — but it concentrates operational and counterparty dependency on a single asset manager and a single custodian bank. The failure here is not credit; it is operational or legal: a custody freeze or a redemption bottleneck at the fund gate propagates directly into the token's ability to honor par. A stablecoin can be fully backed and still depeg if the reserve it is backed by cannot be accessed on the day redemptions spike.
Where the Losses Land
The blast radius runs through everyone who treats the token as cash. A treasury holding USDC as an operating balance faces a redemption gate exactly when it needs liquidity most. A DeFi protocol using the stablecoin as base collateral inherits the depeg into every position built on it — the 2023 episode cascaded through lending markets that had marked USDC at par. And the holder who cannot tell a temporary access freeze from a genuine backing shortfall sells into the depeg and realizes the loss the patient holder avoided — the 0.86 print was paid by sellers, not by the reserve. The concentration determines the correlation: when the reserve funnels through one manager and one custodian, every holder shares the same single point of failure regardless of how diversified their own book is.
What Breaks First and How to Detect It
The cash sleeve breaks first, because it is the thinnest, most bank-exposed, and most redemption-critical layer. Detection is unusually tractable because USDC publishes the evidence: weekly reserve disclosures, monthly Deloitte attestations, and daily CUSIP-level portfolio reporting including which banks hold the cash. The early-warning signals are concrete — rising concentration of the cash sleeve at any single bank, a deposit roster that stops being disclosed, an attestation that lengthens its gap or narrows its scope, and the structural reminder that an attestation is a point-in-time snapshot, not a full audit, so it says nothing about the days between reports. The deeper signal is the reserve-vehicle layer: if the fund manager, the fund, and the custodian are all singular, that is the concentration to price even when every attestation is clean, because it is invisible to a backing check that only asks "is it fully reserved" rather than "through how many independent parties."
Defenses, Residual Risk, and the Scale Question
Real defenses are specific: diversify across stablecoin issuers rather than assuming the most transparent is the safest, because transparency and concentration are different axes; size stablecoin operating balances to what can survive a weekend redemption freeze; hold native rather than bridged tokens so redemption runs directly to the issuer; and read the reserve report for custodian and manager concentration, not just the headline backing ratio. Fake defenses are the trap: treating a clean attestation as proof of resilience, treating "fully backed" as "always redeemable," and treating a Treasury-heavy structure as immune when the operational access path funnels through one custodian. The residual risk is irreducible: some concentration is the price of the safety and transparency that make these reserves evaluable in the first place, and a 2a-7 fund at a G-SIB custodian is a genuinely strong structure — the risk is that its very strength hides how many independent failure points it has. For stablecoin reserves to support 10x institutional adoption, three things must generalize: multi-custodian and multi-manager reserve structures as the disclosed norm, competition among dedicated reserve funds (State Street's SSRXX launched in June alongside the incumbents) diversifying the manager layer, and resolution of the Federal Reserve master-account question so issuers are not forced to route liquidity through commercial-bank chokepoints at all. The constructive signal is that the industry priced the 2023 lesson correctly on the cash sleeve and spread it — the unfinished work is applying the same discipline to the reserve vehicle before it becomes the layer that teaches the next lesson.
For informational purposes only. Not an offer to buy or sell any security. Available only to accredited investors who meet regulatory requirements.
