sUSD and SIP-420: the depeg a governance vote chose

sUSD is minted against staked SNX, which makes it a collateralized debt position in the same family as Dai or crvUSD — value comes from over-collateralized backing, not from a bank reserve or a delta-neutral hedge. Through 2025 nothing went wrong with that backing in the conventional sense. There was no exploit, no oracle failure, and no insolvency event in which collateral fell short of the debt it secured. The coin simply stopped trading at a dollar and stayed there.

The proximate cause was a vote. SIP-420 restructured how Synthetix carries debt: instead of each SNX staker owning a personal slice of the system's outstanding sUSD, the protocol pooled that debt into a single contract it owns and manages, and lowered the issuance ratio for the pool to 200%. sUSD broke toward $0.664 as the change took effect in April, oversupplied through the summer, and reached an all-time low near $0.21 in August. The discount narrowed at times but never closed, and after more than a year below par Synthetix stopped trying to repair it: in June 2026 SIP-423 froze the token and retired it, repaying holders in vested SNX. What began as a wound became, by governance choice, a decommissioning.

Every CDP stablecoin needs a force that pushes the price back up when it trades below peg. In sUSD's original design that force was the staker. Because each staker personally owed a share of the outstanding sUSD, a price below a dollar was an opportunity: buy sUSD cheap on the market, burn it against your own debt, and close your position for less than a dollar of value per unit retired. Thousands of self-interested stakers running that trade was the repeg engine. It was reflexive, decentralized, and required no treasury.

SIP-420 turned that debt over to the protocol. Once a staker migrated into the 420 pool, their obligation was scheduled to be forgiven over twelve months and was no longer theirs to defend — "it's not your debt anymore." The arbitrage that depended on individual ownership of debt simply had no one to run it. The lower 200% issuance ratio compounded the problem from the other side by making fresh sUSD cheaper to mint. So the change weakened the demand for sUSD below peg and strengthened the supply of it at the same time.

It is worth being precise about what kind of risk this is. The collateral was sound; the disclosures were honest; the protocol was solvent. What failed was a behavioral mechanism that the governance process chose to remove on purpose, in exchange for capital efficiency. That is self-inflicted mechanism risk on a live CDP — a failure mode that does not show up in reserve attestations or collateral-quality scores because it lives in the incentive design, not the balance sheet.

With the repeg buyer gone and minting cheaper, sUSD accumulated faster than the market wanted to hold it. The clearest on-chain tell was pool composition: sUSD came to make up the large majority — at points well over 75%, and on some pairs far higher — of the Curve liquidity meant to keep it near a dollar. A stablecoin dominating its own pools is the visible shape of one-way selling: holders rotating out, with the automated market maker absorbing the imbalance and the price grinding lower as it does.

Two reinforcing pressures deepened the slide. A falling SNX through the period reduced confidence in the collateral asset and the staking yield that was supposed to anchor demand, and the absence of a reflexive buyer meant nothing leaned against the drift. Crucially, though, this never became a Terra-style death spiral. There was no reflexive mint-and-dump link between sUSD and SNX of the kind that destroyed UST; the supply stayed roughly flat in the low-30-millions while the price fell. The system found a depressed equilibrium and sat in it, which is exactly why the outcome here is a durable discount rather than a collapse.

Synthetix chose not to reverse SIP-420. Rather than re-raising the collateral ratio to resurrect the old per-staker arbitrage, it set out to manufacture demand and throttle supply directly. The first lever was the 420 Pool itself: locking sUSD for a year against SNX rewards (with rewards vesting over a further three months) pulled float out of circulation and out of the liquidity pools that had been pricing the discount.

The second lever was a ratcheting staking requirement. By early 2026 the protocol required pool participants to hold a rising share of their forgiven debt as staked sUSD — escalated from an initial 10%, through 20%, to 50% of jubileed debt with the "Rebuilding sUSD" update, plus automatic 10% increases every two weeks until the requirement reaches 100% or sUSD trades above $0.98. In effect the protocol re-imposed a synthetic version of the obligation SIP-420 had forgiven, but routed through pool rules rather than individual incentives.

Underneath both sat demand from protocol revenue: fee-funded buybacks, including a portion of perps trading revenue and capped on-market purchases, plus integrations meant to create structural sinks for sUSD. The team's stated arithmetic was that on the order of $5 million of coordinated support would be enough to re-establish stability, with a repeg targeted around early Q2 2026 and sustained stability by mid-2026. The shape of the attempted recovery is the inverse of the break: where one vote could sever the incentive overnight, rebuilding demand by hand is a months-long grind — and in this case one that never reached par.

The demand engineering bought time but not a peg. Lockups, the staking ratchet, and fee-funded buybacks narrowed the discount in stretches, yet sUSD never reclaimed a durable dollar; it kept changing hands well below par on a float that refused either to recover or to disappear. Synthetix's own framing shifted from repair to triage — founder Kain Warwick described the remaining tail as functionally insolvent without exchange revenue to backstop it, and the treasury had already absorbed roughly a third of the supply over the prior year.

In June 2026 governance drew the line. SIP-423 — the first Synthetix proposal to wind sUSD down rather than fix it — froze the contract, took a holder snapshot, and made sUSD permanently non-transferable. Rather than a cash buyback the protocol judged value-destructive, holders are repaid four SNX per sUSD, valuing SNX at $0.25 against sUSD's $1.00 face, under a one-year lock and one-year vest; sUSD held in LP and deposit contracts is handled through a separate claims process. The same governance machinery that severed the peg in SIP-420 used the same lever to bury the token.

Governance is part of the peg. A CDP stablecoin's stability rests not only on its collateral but on the incentive that pulls it back to par, and that incentive can be legislated away by a vote even when the backing is untouched. When assessing a crypto-collateralized dollar, the question is not only "what backs it" but "who is structurally motivated to buy it below a dollar, and could that motivation be changed by governance?" SIP-420 answered that question the hard way.

Capital efficiency and peg defense can be in direct tension. Lowering a collateral ratio and pooling debt genuinely improves capital efficiency and reduces the death-spiral risk a founder rightly pointed to — but the same change can remove the reflexive arbitrage that defended the peg in the first place. There is no free lunch: a more efficient CDP may be a less self-correcting one. The honest read on sUSD is that it never recovered from its own redesign: more than a year of demand-engineering failed to manufacture the buyer the vote had removed, and in the end the same governance process retired the token rather than restore it. Solvency was never the question — an incentive was legislated away, and a stablecoin died of it.