Stablecoin Concentration Risk in DAO Treasuries

When USDC depegged in March 2023, MakerDAO had roughly half of DAI's collateral sitting in the Peg Stability Module as USDC. Three years later, you would expect treasury concentration to be a settled question. It is not — and not for the reason most write-ups assume. Across the largest DAOs by treasury size, the dominant exposure is not which stablecoin they hold. It is the protocol's own token, often above 95% of the balance sheet. Uniswap's $848.5M treasury is 99% UNI. Curve's "treasury" is $16M, all CRV. We pulled the rest of the numbers and the picture is consistent: stablecoin concentration in DAO treasuries is a real risk, but for most DAOs it is a sub-question to a much bigger one they have not yet asked.

Our thesis: the DAOs that have done meaningful work on treasury concentration share three traits — a written policy with a hard cap, a rotating manager with monthly public reporting, and a stable book that includes RWAs and non-USD stables, not just USDC. Everyone else has a token bag and some leftover stables. The regulatory clock — GENIUS Act yield prohibition, MiCA volume caps on USDC — is now forcing the question whether they are ready or not. We think the next cycle's biggest treasury blowups will not look like 2023's USDC depeg. They will look like a 2022-style native-token drawdown that catches a DAO without a stable runway, because the stable runway was always optional.

What the data says

Method: we read March and April 2026 treasury monthly reports for every DAO that publishes one (ENS, CoW, Compound), pulled the multichain Collector balances directly from Etherscan for those that don't, and cross-referenced supply and pricing snapshots against Stablecoin Insider, the Madres Travels stablecoin tracker, and DefiLlama summaries dated April 11–30, 2026.

Two numbers are worth pausing on. The first is the ENS Endowment's explicit 30%-per-protocol cap — the only public hard concentration limit we found at any major DAO. The second is that ENS's largest single stable position is not USDC. It is $23.5M of USDS deployed into the Sky Savings Rate, 25.6% of the Endowment as of March 2026. After the 2023 SVB scare, the DAO that publishes the most legible treasury reports decided its yield-bearing dollar exposure should not be Circle's bank-deposit liability.

CoW's report is the other useful one. 11.4% of the treasury is EUR-denominated stables. Most DAOs do not hold any non-USD stables at all. CoW does it because Karpatkey treats currency as a concentration vector the same way it treats issuer or protocol.

Why USDC ends up in the stable slice

USDC dominates the stable slice when it shows up at all, and the canonical case is still the Peg Stability Module. The PSM's tin and tout fees were cut to 0% in November 2021. By mid-2022, USDC in the PSM was over 50% of DAI's backing collateral. The LitePSM wrapper deployed in September 2024 routes USDS-USDC swaps through the same module today, with both fees still at zero. The mechanism that produced the original concentration is still active; it just has a different label.

The compliance posture compounds the gravity. Circle received conditional OCC approval for a national trust bank in December 2025, achieved MiCA EMI authorization in France in 2024, and IPO'd on NYSE under CRCL after filing the S-1 on April 1, 2025. Tether, by contrast, is not on ESMA's MiCA-authorized E-Money Token list; Coinbase Europe delisted USDT in December 2024 and Binance followed for EEA spot pairs in March 2025. For a DAO whose contributors are global, USDC is the path of least friction. Concentration follows friction.

What the depeg cost in March 2023 is the case against this default. Circle had $3.3B of $40B reserves stranded at SVB (~8%); USDC traded as low as $0.87; DAI mechanically followed it down because the PSM exchanges over-collateralized assets for distressed USDC at par. Aave saw 3,400 liquidations on $24M of collateral, 86% of it USDC, in the 60 hours before the FDIC systemic-risk announcement. MakerDAO responded by cutting USDC's share of DAI collateral from ~50% to ~8% by mid-2023 and pivoting hard into RWAs. The lesson scaled inside Maker; it did not scale across DAO treasuries broadly.

What the diversifiers actually do

Sky is the most aggressive case. As of April 2026, Spark holds over $1.5B of tokenized US Treasuries — $800M in BlackRock's BUIDL, $400M in Anemoy via Janus Henderson, $300M across Superstate's products — backing USDS issuance against off-chain duration rather than another issuer's bank deposits. The combined USDS + DAI float is ~$13.4B, which makes the diversification material at sector scale, not just self-protective.

Arbitrum is the smaller but more replicable model. STEP allocated 35M ARB (~$11.6M) into tokenized T-bills split across Franklin Templeton BENJI, Spiko USTBL, and WisdomTree WTGXX. The program is explicitly capped at ~1% of treasury — small enough to be politically possible, large enough to be a published precedent for the next ARB allocation. STEP 2.0 expanded the issuer count further in Q1 2026.

Frax v3 took the opposite route — collapse the stable's own backing into a tightly-defined HQLA list: short-dated US Treasuries, Fed reverse-repo, Fed master account deposits, and select MMF shares. The protocol's IORB-driven AMO rebalances on-chain when off-chain rates shift. This is treasury concentration handled at the issuance layer, not the holdings layer — a different solution to the same problem.

The pattern across the diversifiers: a single managing organization with a public mandate, a written cap, and monthly reporting. ENS uses Karpatkey under EP4.5. CoW uses Karpatkey under CIP-26. Compound stood up its own Treasury Management Committee in Q1 2026 and announced a $30M deployment envelope copying the same playbook. Sky and Frax are the special cases because they manage protocol balance sheets, not discretionary treasuries; the policy lives in the issuance contract, not in a Snapshot vote.

What we might be wrong about

The honest counterargument is manager risk. In February 2026, GnosisDAO terminated Karpatkey by an 88% vote and replaced them with Noca; no public composition snapshot under the new manager has been published yet. The Aave-Chan Initiative is leaving Aave governance in July 2026. Treasury management run by a single org with a public mandate is better than no management — but the model has key-person risk, and the "single-protocol cap" rule at ENS is enforced by manager attestation, not on-chain. We have not seen a Karpatkey transition trigger a treasury rebalance under stress, and we should be modest about claims of durability until that test happens.

The other counterargument is structural. Mantle holds 78% of its $6.2B treasury in MNT and frames the concentration as a deliberate alignment incentive — the treasury and the token rise and fall together, which keeps the DAO's interests aligned with the protocol. This works in a strong cycle. It worked badly for Uniswap, whose treasury fell from $19B in 2021 to $1.5B in 2022, a 92% drawdown that triggered no automatic policy response. The "alignment" frame is a real argument; it is also the frame that makes 92% drawdowns survivable as long as no operating expense came due during the trough. Most DAOs we looked at do not have a written policy for the trough.

We could not pull current itemized treasury composition for Uniswap, Lido, or Optimism — the gap is real and meaningful. Optimism's 50%-of-sequencer-revenue OP buyback program from February 2026 is a re-concentration into native token, not a diversification, but we cannot quantify its effect on the overall mix without an updated treasury report.

Where this goes

The GENIUS Act yield prohibition, signed July 18, 2025 and now moving through implementing rules after Treasury's April 3, 2026 NPRM on state-regime equivalence, separates the payment-stablecoin layer from the yield wrapper. USDC, USDT, and PYUSD cannot pay holders interest "solely in connection with" holding the token. sUSDS, sUSDe, and sFRAX can, because they are wrappers, not the payment instrument. The mechanical effect is that DAO treasuries holding raw USDC for yield will progressively rotate into wrappers — and concentration migrates from the issuer layer to the wrapper-issuer layer, where the credit risk is now Sky governance, Ethena's delta-neutral hedge book, or Frax's collateral list rather than Circle's bank account. The ECB flagged exactly this in its November 2025 Financial Stability Review: two issuers account for around 90% of the stablecoin market, and concentration risk is a sector-level concern, not a DAO-level one.

The watch-points for the next 6–12 months: whether GnosisDAO's Noca transition produces a published composition by Q3 2026; whether Uniswap's Treasury Working Group, first temp-checked at 74% approval in March 2024, produces an actual diversification proposal before the next major UNI drawdown; and whether the GENIUS Act final rules close or formalize the affiliate-rewards loophole that PayPal currently uses to pay up to 4% APY on PYUSD balances inside its app. Each one is a stress test of whether the diversification narrative is real or rhetorical.

If you run a DAO treasury and want this view across your holdings, we are building it.