Stablecoins are not stable: The Stream crash reveals the structural eyewash of Decentralized Finance.

Author: YQ Source: X, @yq_acc Translator: Shan Ouba, Golden Finance

In the first two weeks of November 2025, decentralized finance (DeFi) exposed fundamental flaws that the academic community has been warning about for years. After the collapse of Stream Finance's xUSD, Elixir's deUSD and many other synthetic stablecoins also fell, which is by no means an isolated incident caused solely by mismanagement. These events reveal structural issues in the DeFi ecosystem regarding risk management, transparency, and the construction of trust mechanisms.

What I observed in the Stream Finance collapse was not the traditional complex smart contract exploits or oracle manipulation attacks, but rather a more concerning situation: a lack of basic financial transparency wrapped in decentralized rhetoric. When an external fund manager lost $93 million with zero effective oversight, triggering a $285 million cross-protocol contagion; when the total value locked in the entire “stablecoin” ecosystem evaporated by 40%-50% within a week while maintaining the pegged exchange rate, we must acknowledge a core fact about the current state of decentralized finance — this industry has made no progress.

To be more precise, the current incentive mechanism rewards those who ignore the lessons and punishes those who act conservatively, while making the entire industry bear the losses when an inevitable collapse occurs. An old saying in finance is painfully validated here: if you don’t know where the returns come from, you are the source of the returns yourself. When some protocols promise an 18% return rate through undisclosed strategies, while the yield in mature lending markets is only 3%-5%, the true source of this high yield is actually the principal of the depositors.

The Operating Mechanism and Risk Contagion of Stream Finance

Stream Finance positions itself as a yield optimization protocol, offering a 18% annualized return for USDC depositors through its interest-bearing stablecoin xUSD. Its claimed strategies include “delta-neutral trading” and “hedged market making,” terms that sound complex and profound but do not reveal the actual operational situation. In comparison, during the same period, the annualized yield for USDC deposits in mature protocols like Aave was 4.8%, while Compound was just slightly above 3%. Faced with a return rate three times higher than the market level, basic financial common sense should raise suspicions, yet users still deposited hundreds of millions of dollars. Before the collapse, the trading price of 1 xUSD was 1.23 USDC, reflecting the so-called compound returns. xUSD claimed a peak management asset size of 382 million dollars, but DeFiLlama data shows its TVL peak was only 200 million dollars, indicating that over 60% of the claimed assets are in unverifiable off-chain positions.

The actual mechanism revealed by Yearn Finance developer Schlagonia after the crash shows that this is a systematic fraud disguised as financial engineering. Stream creates unsecured synthetic assets through recursive lending, with the specific process as follows: after the user deposits USDC, Stream exchanges it for USDT via CowSwap; then uses these USDT to mint deUSD from Elixir (the reason for choosing Elixir is its high yield incentives); transfers deUSD cross-chain to networks like Avalanche, deposits it in the lending market to borrow USDC, completing a cycle. At this point, although the strategy has concerning complexity and cross-chain dependencies, it is still similar to standard collateralized lending. However, Stream does not stop there — it does not use the borrowed USDC only for additional collateral cycles, but instead mints xUSD through its StreamVault contract, resulting in an xUSD supply far exceeding actual collateral support. With only 1.9 million dollars in verifiable USDC collateral, Stream minted 14.5 million dollars in xUSD, leading to a synthetic asset scale that is 7.6 times the base reserves. This is equivalent to a fractional reserve banking system without reserves, lacking regulatory oversight, and having no lender of last resort support.

The circular dependency with Elixir further exacerbates structural instability. In the process of driving up the supply of xUSD, Stream deposits 10 million USDT into Elixir, expanding the supply of deUSD. Elixir exchanges these USDT for USDC and deposits them into Morpho's lending market. As of early November, there were over 70 million USDC in supply and over 65 million borrowed on Morpho, with Elixir and Stream being the two dominant participants. Stream holds about 90% of the total supply of deUSD (about 75 million USD), while Elixir's reserves are mainly composed of Morpho loans to Stream. These stablecoins are collateralized against each other, destined to collapse together. This “financial inbreeding” creates systemic fragility.

Industry analyst CBB publicly pointed out these issues on October 28: “The on-chain backing of xUSD is about $170 million, yet approximately $530 million was borrowed from lending protocols, resulting in a leverage ratio of 4.1 times, with most directed towards illiquid positions. This is not yield farming; it is reckless gambling.” Schlagonia had warned the Stream team 172 days before the collapse, stating that it would only take five minutes to analyze its positions to see that a collapse was inevitable. These warnings were public, specific, and accurate, yet they were ignored by users chasing yield, platform managers pursuing fee income, and protocols supporting the entire structure. On November 4, when Stream announced external fund managers had lost about $93 million in fund assets, the platform immediately suspended all withdrawals. Due to the lack of a redemption mechanism, panic spread rapidly, with holders rushing to sell xUSD in the illiquid secondary market, causing its price to plummet 77% to about $0.23 within hours. This stablecoin, which promised stability and high yields, evaporated three-quarters of its value in a single trading session.

Specific Impacts of Risk Contagion

According to data from the DeFi research institution Yields and More (YAM), the total debt exposure related to Stream in the entire ecosystem reaches $285 million, including: TelosC has $123.64 million in loans collateralized by Stream assets (the largest single platform exposure); Elixir Network has borrowed $68 million through a private Morpho vault (accounting for 65% of deUSD reserves); MEV Capital faces a $25.42 million exposure, approximately $650,000 of which has formed bad debt due to oracles freezing the xUSD price at $1.26 (while the actual market price dropped to $0.23); Varlamore has an exposure of $19.17 million; Re7 Labs has exposures of $14.65 million and $12.75 million in two vaults, respectively; Enclabs, Mithras, TiD, and Invariant Group also have smaller exposures. Euler faces about $137 million in bad debt, with over $160 million in funds frozen across multiple protocols. Researchers note that this list is not exhaustive and warn that “there may be more stablecoins/vaults affected,” as the full picture of interconnected exposures remains unclear weeks after the initial collapse.

The deUSD of Elixir plummeted 98% from $1 to $0.015 within 48 hours due to 65% of its reserves being concentrated in loans issued to Stream through the private Morpho vault, becoming the fastest mainstream stablecoin to crash since Terra UST in 2022. Elixir provided redemption services for about 80% of deUSD holders not associated with Stream, allowing them to exchange at $1 for USDC, protecting most community users, but this significant cost of protection was borne by Euler, Morpho, and Compound. Elixir then announced it would completely terminate all stablecoin products, acknowledging that trust had been irreparably damaged. The broader market reaction showed systemic loss of confidence: according to Stablewatch data, although most interest-bearing stablecoins maintained their peg to the dollar, their TVL declined by 40%-50% within a week after the Stream crash, equivalent to $1 billion in funds flowing out of protocols that had not crashed and had no technical issues. Users could not distinguish between quality projects and fraudulent ones, leading them to choose a complete withdrawal. By early November, the total DeFi TVL had decreased by $20 billion, with the market pricing for widespread risk contagion rather than responding to the collapse of specific protocols.

October 2025: $60 million triggered chain liquidation

Less than a month before the collapse of Stream Finance, on-chain forensic analysis revealed that the cryptocurrency market experienced not an ordinary collapse, but a precision attack at the institutional level exploiting known vulnerabilities. From October 10 to 11, 2025, a well-timed market sell-off of $60 million triggered oracle failures, leading to large-scale cascading liquidations in the DeFi ecosystem. This is not an issue of excessive leveraging of legitimately damaged positions, but a design flaw in institutional-level oracles, replaying attack patterns that have been recorded and publicly reported since February 2020.

The attack began at 5:43 AM UTC on October 10, when a concentrated sell-off of 60 million USDe occurred in the spot market of a certain exchange. In a well-designed oracle system, the impact of this behavior should have been negligible — multiple independent price sources combined with a time-weighted mechanism can effectively prevent manipulation. However, the reality was that the oracle system adjusted the valuations of collateral assets (wBETH, BNSOL, and USDe) in real-time based on the manipulated spot prices of the trading platform, triggering large-scale liquidations immediately. Millions of simultaneous liquidation requests exceeded the system's processing capacity, leading to infrastructure paralysis; market makers were unable to place orders in a timely manner due to API interface interruptions and withdrawal queues, resulting in an instant depletion of liquidity, and the liquidation chain reaction self-reinforced.

Attack Methods and Historical Precedents

The oracle faithfully reported the manipulated price on a single platform, while prices in all other markets remained stable. The main exchange showed the USDe price at $0.6567 and the wBETH price at $430, while prices on other platforms deviated by less than 30 basis points from normal levels, and the on-chain liquidity pool was minimally affected. As Ethena founder Guy Young pointed out, “During the entire event, over $9 billion in on-demand stablecoin collateral was immediately redeemable,” which proves that the underlying assets were not impaired. However, the oracle still reported the manipulated prices, and the system executed liquidations based on these prices, forcing positions to be liquidated due to valuations that did not exist in any other market.

This is strikingly similar to what happened with Compound in November 2020 - at that time, DAI skyrocketed to $1.30 on Coinbase Pro within an hour, while the trading price on all other platforms was $1.00, resulting in a $89 million liquidation. The trading platform has changed, but the vulnerabilities remain. This method of attack is entirely consistent with the bZx incident in February 2020 (which manipulated Uniswap oracles to steal $980,000), the Harvest Finance incident in October 2020 (which manipulated Curve to steal $24 million and triggered a $570 million run), and the Mango Markets incident in October 2022 (which manipulated across multiple platforms to extract $117 million). From 2020 to 2022, 41 oracle manipulation attacks collectively stole $403.2 million. The industry's response has been slow and fragmented, with most platforms continuing to use spot markets with excessive reliance and inadequate redundancy on oracles. The amplification effect indicates that as market scale expands, these lessons become increasingly important: in the 2022 Mango Markets incident, a $5 million manipulation led to a $117 million loss, amplified by 23 times; in October 2025, a $60 million manipulation triggered a chain reaction that resulted in an even larger amplification effect. The attack patterns have not become more complex, but rather the foundational systems continue to scale while retaining the same fundamental vulnerabilities.

Historical Patterns: Crash Events from 2020 to 2025

The collapse of Stream Finance is neither a new occurrence nor unprecedented. The DeFi ecosystem has experienced stablecoin collapses multiple times, each revealing similar structural vulnerabilities, yet the industry continues to repeat the same mistakes, with increasing scale. Recorded collapse events over the past five years show a consistent pattern: algorithmic stablecoins or partially collateralized stablecoins attract deposits by offering unsustainable high yields, which do not stem from actual income but rely on token issuance or new capital inflows; protocols operate with excessive leverage, the actual collateralization rate is opaque, and there exists a “protocol A endorses protocol B, and protocol B in turn endorses protocol A” circular dependency; when any shock exposes its potential insolvency or subsidies become unsustainable, a bank run occurs; users rush to exit, the value of collateral assets plummets, liquidation triggers in a chain reaction, and the entire structure collapses within days or hours; risk contagion spreads to all protocols that accept the stablecoin as collateral or hold related positions within the ecosystem.

May 2022: Terra (UST/LUNA)

Loss scale: $45 billion in market value evaporated in three days. UST is an algorithmic stablecoin backed by LUNA through a mint-and-burn mechanism. The Anchor protocol provides an unsustainable 19.5% annual yield on UST deposits, with about 75% of UST deposited in the protocol to earn rewards. The system relies on continuous capital inflow to maintain the peg. Triggering event: On May 7th, the Anchor protocol experienced $375 million in withdrawals, followed by a massive sell-off of UST that led to its de-pegging. Users exchanged UST for LUNA to exit, causing the supply of LUNA to surge from 346 million to 6.5 trillion in three days, resulting in a death spiral, with both tokens plummeting to nearly zero. This collapse caused significant losses for individual investors and led to the bankruptcy of several major cryptocurrency lending platforms, such as Celsius, Three Arrows Capital, and Voyager Digital. Terra founder Do Kwon was arrested in March 2023 and faces multiple fraud charges.

June 2021: Iron Finance (IRON/TITAN)

Loss scale: The TVL dropped from 2 billion USD to nearly zero within 24 hours. IRON is partially collateralized stablecoin, 75% backed by USDC and 25% backed by TITAN. It attracted deposits by offering an annualized yield of up to 1700%, creating unsustainable yield farming incentives. When large holders began to redeem IRON for USDC, the selling pressure on TITAN self-reinforced, causing the price to plummet from 64 USD to 0.00000006 USD, leading to the complete failure of IRON's collateral. Lesson: In a pressured environment, partial collateral mechanisms are insufficient to maintain stability; when the backing token itself faces a death spiral, the arbitrage mechanisms fail under extreme pressure.

March 2023: USDC

De-pegging situation: Due to $3.3 billion in reserves being trapped in the near-bankrupt Silicon Valley Bank, the price of USDC dropped to $0.87 (a 13% decline). This situation should have been “impossible” for a “fully-backed” stablecoin that is regularly audited. It was only when the Federal Deposit Insurance Corporation (FDIC) invoked the systemic risk exception clause to provide full protection for Silicon Valley Bank deposits that USDC regained its pegged exchange rate. Risk contagion: This led to the de-pegging of DAI (over 50% of DAI's collateral is USDC), triggering over 3,400 automatic liquidations on Aave, with a total value of $24 million. This event indicates that even well-intentioned, regulated stablecoins face concentration risk and dependence on the stability of the traditional banking system.

November 2025: Stream Finance (xUSD)

Loss scale: direct loss of $93 million, total ecosystem exposure of $285 million. Operating mechanism: creation of unsecured synthetic assets through recursive lending (actual collateral magnified by 7.6 times); 70% of funds invested in opaque off-chain strategies managed by anonymous external managers; no proof of reserves. Current status: xUSD trading price ranges from $0.07 to $0.14 (down 87%-93% from the pegged price), with nearly no liquidity; withdrawals indefinitely frozen; multiple lawsuits have been filed; Elixir has completely ceased operations; the entire industry is withdrawing from interest-bearing stablecoins.

All cases exhibit a common failure pattern: unsustainable high returns (Terra at 19.5%, Iron at 1700% annualized, Stream at 18%), all providing returns disconnected from actual income generation; circular dependencies (UST-LUNA, IRON-TITAN, xUSD-deUSD), all experiencing a mutually reinforcing failure pattern of “a loss for one is a loss for all”; lack of transparency (Terra concealed the cost of Anchor subsidies, Stream placed 70% of operations off-chain, Tether's reserve composition has been consistently questioned); partial collateral or self-issued backing (relying on volatile assets or self-issued tokens, creating a death spiral under pressure — collateral value plummets precisely when support is most needed); oracle manipulation (frozen or manipulated price feedback leads to failure of liquidation mechanisms, turning price discovery into trust discovery, with bad debts accumulating until the system is insolvent). The core conclusion is evident: stablecoins are not stable. They may appear stable before a crash, and the transition from stability to collapse can occur in just a few hours.

Oracle Failures and Infrastructure Collapse

At the beginning of the Stream crash, the oracle issue immediately became apparent. When the actual market price of xUSD dropped to $0.23, many lending protocols hardcoded the oracle price at $1.00 or higher to prevent cascading liquidations. This decision, aimed at maintaining stability, led to a fundamental disconnection between market reality and protocol behavior. This hardcoding is a deliberate policy choice, not a technical failure. Many protocols use manual updates for oracles to avoid liquidations triggered by temporary fluctuations, but when the price drop reflects actual insolvency rather than temporary market pressure, this approach can lead to catastrophic failures.

The protocol faces a dilemma: using real-time prices may encounter manipulation and cascading liquidations during volatility (the events in October 2025 caused significant losses); using delayed prices or time-weighted average prices (TWAP) cannot respond to real insolvency, leading to the accumulation of bad debts (in the Stream Finance incident, the oracle showed a price of $1.26 while the actual price was $0.23, resulting in $650,000 in bad debts for MEV Capital alone); using manual updates introduces centralization and discretionary intervention, and may cover up insolvency by freezing the oracle. All three methods have led to losses of hundreds of millions or even billions of dollars.

Infrastructure capacity under pressure environment

In October 2020, Harvest Finance encountered an infrastructure collapse - after a $24 million attack, users rushed to withdraw, and TVL dropped from $1 billion to $599 million. The lessons at that time were obvious: oracle systems must consider infrastructure capacity under stress events; liquidation mechanisms must set rate limits and circuit breaker mechanisms; exchanges must maintain redundancy capacity ten times the normal load. However, the events of October 2025 proved that the institutional level still did not learn this lesson. When millions of accounts faced simultaneous liquidations, billions of dollars in positions were liquidated within an hour, and the order book went blank because all buy orders were consumed and the system was overloaded and unable to generate new buy orders, the degree of infrastructure collapse was comparable to that of the oracle. Technical solutions have long existed, but have not been implemented for a long time - because these solutions would reduce efficiency under normal circumstances and require investment that could have been turned into profit.

If you cannot clearly identify the source of your returns, then you are not earning profits but rather incurring costs for someone else's gains. This principle is not complicated, yet billions of dollars are still being deposited into black box strategies — because people prefer to believe comforting lies rather than unsettling truths. The next Stream Finance is currently in operation.

Stablecoins are not stable. Decentralized finance is neither decentralized nor secure. Returns from unknown sources are not profits, but theft with a countdown. These are not subjective opinions, but empirical facts confirmed after paying a huge price.