Dependency map
Four chains worth knowing
reUSD → sUSDe → USDe → CEX perp basis. About 86% of reUSD's on-chain reserve is staked Ethena dollars, so a reinsurance token quietly inherits centralized-venue basis risk two links down.
sUSDat → STRC → Strategy → BTC. The staked Saturn layer is a bitcoin-linked preferred-stock position wearing a stable ticker. It fell ~26% peak-to-trough in June 2026 while USDat, the par layer one hop up, held its peg. Same protocol, different tokens, different risk.
USD3 → Slope. One originator carries 90% of the credit book behind the senior token, a single-counterparty concentration you cannot see from the peg chart.
apxUSD → STRC → Strategy → BTC, plus a loop. ~74% of net reserves are the same Strategy preferred that sank sUSDat, held off-chain at a brokerage with no atomic redemption, while apxUSD vaults also loop PT-apyUSD through Pendle and Morpho. When STRC broke below par in June 2026 the token depegged and still trades deeply below a dollar (live on our peg radar).
The boards
How to read a board
An arrow points at the dependency: the tail is the thing at risk. Each step right is one hop further from you, and a chip with no arrow leaving it is where that chain ends.
Board A US T-bills / cash cluster
12 tokens · 11 dependencies11 of 33 dependencies land here, and the chains end at 4 different places: CEX perp basis, Reinsurance premiums, US T-bills / cash, USDC (Circle).
Board B Pendle (PT/YT) cluster
9 tokens · 8 dependencies8 of 33 dependencies land here, and the chains end at 4 different places: BTC (Strategy treasury), Discretionary op book, Morpho markets, Pendle (PT/YT).
Board C Aave v3 cluster
7 tokens · 6 dependencies6 of 33 dependencies land here, and the chains end at 3 different places: Aave v3, Crypto collateral, Slope receivables.
Board D Operator credit lines cluster
4 tokens · 3 dependencies3 of 33 dependencies land here, and the chains end at 2 different places: Operator credit lines, Restaked ETH collateral.
Board E GPU / AI-infra loans cluster
3 tokens · 2 dependenciesOne token, one backing
3 tokens whose entire dependency is a single link. They are separate components of the graph, so nothing else on this page touches them.
Every dependency, in full
33 rows, one per line on the boards above. Every fact names the dated report it was quoted from, which is not always one of the two tokens in the row - and this table is what a narrow screen or a reader without JavaScript gets instead of the drawings.
| At risk | Link | Leans on | Sourced fact | Source |
|---|---|---|---|---|
| USDe | backed by | CEX perp basis | institutionalized basis trade: spot/LST long vs centralized-venue perp shorts | ethena |
| reUSDe | junior to | reUSD | mezzanine tranche: the reinsurer's own capital absorbs first, then reUSDe, then reUSD | re-protocol |
| sUSDe | wraps | USDe | staked layer accrues the basis yield | ethena |
| reUSD | backed by | Reinsurance premiums | off-chain capital in Reg-114 trusts backing Cover Re treaties; attested daily | re-protocol |
| reUSD | backed by | sUSDe | ~86% of the on-chain reUSD reserve sits in sUSDe ($58.59M of $68.00M, six Ethereum protocol addresses, read on-chain 2026-07-19) - inherits Ethena tail risk | re-protocol |
| USDS / DAI | depends on | USDe | Spark Liquidity Layer CAN deploy Sky reserves into Ethena, but today it does not: the ALM Proxy holds 80.9 sUSDe, about $100, dust since Nov 2025, read on-chain 2026-08-05 (the second ALM proxy, which the DeFiLlama adapter misses, holds zero). This is live headroom, not a dead letter - the combined sUSDe+USDe position peaked at $953.65M on 2025-08-02, 87% of the $1.1B governance target, and the 20%-of-USDe-supply cap is about $777M against today's $3.8828B supply | spark |
| USDS / DAI | depends on | USDC (Circle) | PSM redeems 1:1 into USDC - Circle issuer risk at the base | sky-lending |
| USDS / DAI | backed by | US T-bills / cash | RWA/T-bill income plus stability fees | sky-lending |
| sUSDS | wraps | USDS / DAI | Sky Savings Rate accrual | sky-lending |
| USD1 | backed by | US T-bills / cash | claimed reserves with monthly attestations - no named attestation firm confirmed; owner can mint, freeze and drain via upgradeable proxy | wlfi |
| USDG | backed by | US T-bills / cash | cash + short-dated T-bills; custody DBS and Standard Chartered; KPMG monthly attestation | global-dollar |
| USDat | backed by | STRC (Strategy pref.) | M0-based dollar minted against off-chain STRC purchases; audits flag a settlement lag in which USDat can be briefly unbacked | saturn |
| sUSDat | wraps | USDat | staked layer, and where the STRC price actually lands: the reserve allocates dynamically between T-bills and STRC by credit LTV, the current weight is unpublished, and the ~26% June 2026 drawdown implies it was high | saturn |
| STRC (Strategy pref.) | depends on | BTC (Strategy treasury) | issuer Strategy's balance sheet is bitcoin | saturn |
| apxUSD | backed by | STRC (Strategy pref.) | Strategy STRC preferred, held OFF-chain (Alpaca brokerage, Preference Foundation), is 84.4% of asset reserves or 70.0% counting protocol-owned liquidity (2026-08-05). The two bases differ by 14 points, so a bare percentage here is meaningless without saying which. It did NOT rise: the ~74% this note used to carry was the wider base and is now 70.0%. STRC broke below $100 par June 2026 and apxUSD depegged - no atomic redemption to force it back | apyx |
| apxUSD | depends on | Morpho markets | loop leverage sits on Morpho markets; apxUSD $0.9177 (2026-08-05), up from $0.868 a month earlier because STRC rallied 24%, not because anything was repaired - attested collateral is still 92.24% and there is still no atomic redemption | apyx |
| apxUSD | depends on | Pendle (PT/YT) | self-referential: apxUSD vaults loop PT-apyUSD through Pendle | apyx |
| dUSD (Makina) | depends on | Pendle (PT/YT) | Pendle + Spectra PT positions in the operator book; deposits AND redemptions allowlist-gated on-chain | makina |
| dUSD (Makina) | backed by | Discretionary op book | discretionary multi-strategy USD book; $4.13M AUM/share-price manipulation 2026-01-20 | makina |
| crvUSD | backed by | Crypto collateral | only about 17% of circulating crvUSD is user CDP debt across the nine LLAMMA mint markets (collateralized ~191%, soft-liquidation bands). Another ~15% is PegKeeper inventory the protocol minted straight into the crvUSD/USDT pool, and the remaining two thirds traces mostly to DAO-authorized minting - including a credit line for Yield Basis, run by Curve's own founder (2026-08-05) | curve-dex |
| sUSD3 | junior to | USD3 | junior tranche, $7.94M = 10.58% of the pool (2026-08-04), absorbing the first 14.4% of a Slope-book loss before USD3 is impaired. The $1M insurance fund that used to sit above it covered the crypto credit lines only, never the fintech conduits that are now 90% of the book - and the current suppliers page no longer mentions it at all | 3jane |
| GHO | backed by | Crypto collateral | minted against Aave collateral | aave-v3 |
| GHO | depends on | Aave v3 | native to the Aave protocol | aave-v3 |
| USD3 | depends on | Aave v3 | idle-cash buffer, and it is thin: 4.74% of the pool on 2026-08-04 ($3.56M), down from 20.28% on 2026-07-19. maxWithdraw() returns that same figure for every large holder, so it is the global exit, not a per-holder allowance | 3jane |
| USD3 | backed by | Slope receivables | 90.93% of the facility book from one originator ($55.25M of $60.77M, 2026-08-04), carried at LTV 1.0 with no overcollateralization - and Slope services the loans it sold, so origination, servicing and collections sit with the same counterparty | 3jane |
| stcUSD | wraps | cUSD | staked layer; unwraps to cUSD | cap |
| cUSD | backed by | Operator credit lines | yield = whitelisted operator credit spreads; redeems as a proportional reserve basket (socializes a single-asset depeg). Only about 4% of supply is instantly redeemable, the rest waits on loans that do not repay on demand, and secondary DEX liquidity is about $6k (2026-08-05) | cap |
| cUSD | depends on | Restaked ETH collateral | operator defaults insured by restaked EigenLayer/Symbiotic collateral. The coverage question is now answered rather than open (2026-08-05): Cap sets NO protocol-wide minimum coverage ratio, so per-borrower LTV is the only floor - and the 3-of-5 developer Safe can whitelist a borrower and set that LTV in one transaction with no delay, while contract upgrades are timelocked 24h | cap |
| USDai | backed by | GPU / AI-infra loans | GPU-collateralized infrastructure loans; NAV re-marks only on default | usd-ai |
| sUSDai | wraps | USDai | staked layer; FIFO exit on 30-day windows | usd-ai |
| syrupUSDC | backed by | Institutional loans | over-collateralized institutional credit | maple |
| NUSD | backed by | OTC token deals (hedged) | discounted locked-token deals with short hedges; 2026 exploit + DNS hijack | neutrl |
| feUSD | backed by | HYPE collateral | licensed Liquity-V2 CDP fork; collateral concentrated in HYPE on one young chain, Redstone feeds | felix |
No two lines cross anywhere on this page, and that is checked rather than hoped for: the graph is a forest, so a crossing-free drawing exists, and the solver counts its own crossings on every build. It currently reports 0.