Is Ethena a systemic risk or a savior for DeFi?

Author: DC | In SF

Compiled by: Block unicorn

Ethena is the most successful protocol in DeFi history. About a year ago, its total locked value (TVL) was less than $10 million, but now it has grown to $5.5 billion. It is integrated in various ways with multiple protocols, such as @aave, @SkyEcosystem (IE Maker / Sparklend), @MorphoLabs, @pendle_fi and @eigenlayer. There are so many protocols collaborating with Ethena that I had to change the cover multiple times when recalling another partner. Among the top ten protocols by TVL, six are either collaborating with Ethena or are Ethena itself (Ethena ranks ninth). If Ethena fails, this will have a profound impact on many protocols, especially AAVE, Morpho and Maker, which will be insolvent to varying degrees in terms of functionality. At the same time, Ethena's growth of tens of billions of dollars has significantly increased the overall usage of DeFi, similar to the impact of stETH on Ethereum DeFi. So, is Ethena destined to destroy the DeFi we know, or will it usher in a new renaissance for DeFi? Let's dive into this issue.

How does Ethena actually work?

Although it has been launched for more than a year, there is still widespread misunderstanding about how Ethena works. Many claim it is the new Luna and then refuse to elaborate further. As someone who has warned about Luna, I find this view very one-sided, but I also believe that most people lack a sufficient understanding of the details of how Ethena operates. If you think you fully understand how Ethena manages delta-neutral positions, custody and redemption, please skip this section, otherwise this is important reading material for a thorough understanding.

Overall, Ethena benefits from financial speculation and the cryptocurrency bull market, but in a more stable manner than BTC. As cryptocurrency prices rise, more and more traders want to long BTC and ETH, while fewer and fewer traders are willing to short. Due to the supply-demand relationship, the short traders pay fees to the long traders. This means that traders can hold BTC while shorting the same amount of BTC, thereby achieving a neutral position, where the gains and losses of the long and short positions offset each other, while the traders still earn interest income. Ethena's operation is entirely based on this mechanism; it leverages the lack of sophisticated investors in the crypto market, who are more interested in earning yield rather than simply going long on BTC or ETH.

However, a major risk of this strategy is the custody risk of exchanges, as evidenced by the collapse of FTX and its impact on first-generation delta-neutral managers. Once an exchange goes bankrupt, all funds may be lost. This is why, no matter how efficiently and securely mainstream managers manage capital, they were greatly impacted by the collapse of FTX, the most obvious example being @galoiscapital, and this was not their fault. Exchange risk is one of the important reasons why Ethena has chosen to use @CopperHQ and @CeffuGlobal as custodians. These custodial service providers, as trusted intermediaries, are responsible for holding assets and assisting Ethena in interacting with exchanges, while avoiding exposing Ethena to the custody risk of exchanges. The exchanges, in turn, can rely on Copper and Ceffu, as they have legal agreements with the custodial institutions. The net profit and loss (i.e., the amount Ethena needs to pay to the long traders, or the amount the long traders owe to Ethena) is settled periodically by Copper and Ceffu, and Ethena systematically rebalances its positions based on these settlement results. This custodial arrangement effectively mitigates exchange-related risks while ensuring the stability and sustainability of the system.

The minting and redemption of USDe / sUSDe is relatively simple. USDC or other major assets can be used to purchase or mint USDe. USDe can be staked to generate sUSDe, which will earn yield. sUSDe can then be sold to the market by paying the corresponding swap fee, or redeemed for USDe. The redemption process usually takes seven days. USDe can then be exchanged 1:1 for the underlying assets (corresponding to $1 in value). These underlying assets come from the asset reserve and the collateral used by Ethena (mainly BTC and ETH/ETH derivatives). Given that some USDe are not staked (many of which are used for Pendle or AAVE), the income generated by the assets supporting these unstaked USDe helps to enhance the yield of sUSDe.

So far, Ethena has been able to handle large amounts of withdrawals and deposits relatively easily, although the USDe-USDC slippage can sometimes be as high as 0.30%, which is relatively high for a stablecoin, but far from reaching a significant de-peg level or posing a danger to lending protocols, so why are people so concerned?

Well, if there is a large withdrawal demand, say 50%...

How could Ethena "fail"?

Given that we now understand Ethena's yields are not "fake" and how it operates at a more granular level, what are the main real concerns about Ethena? Basically, there are a few scenarios. First, the funding rate could turn negative, in which case, if Ethena's insurance fund (currently around $50 million, sufficient to withstand a 1% slippage/capital loss on the current TVL) is not enough to cover the losses, Ethena will ultimately lose money instead of profiting. This scenario seems relatively unlikely, as most users would likely stop using USDe when yields decline, which has happened in the past.

Another risk is the custodial risk, i.e., the risk of Copper or Ceffu trying to operate with Ethena's money. The fact that the custodians do not have full control of the assets mitigates this risk. The exchanges have no signing authority and cannot control any wallets holding the underlying assets. Copper and Ceffu are "commingled" wallets, meaning that all institutional users' funds are mixed in hot/warm/cold wallets, with multiple risk mitigation measures such as governance (i.e., control) and insurance. From a legal perspective, this is a bankruptcy-remote trust structure, so even if the custodians go bankrupt, the assets held by the custodians do not belong to the custodians' property and the custodians have no claim over these assets. In practice, there are still risks of simple negligence and centralization, but there are many safeguards to avoid this, and I would consider this a black swan event.

The third, and most commonly discussed, risk is the liquidity risk. To manage redemptions, Ethena must sell both its derivative positions and spot positions simultaneously. If there are violent price fluctuations in ETH/BTC, this could be a difficult, expensive and potentially very time-consuming process. Currently, Ethena has prepared billions of dollars to be able to redeem USDe 1:1 for dollars, as it holds a large amount of stable positions. However, if Ethena's share of the total open interest (i.e., all outstanding derivatives) becomes larger and larger, this risk becomes more severe and could lead to a drop in Ethena's net asset value (NAV) by a few percentage points. However, in this case, the insurance pool would likely fill the gap, and this alone is not enough to cause a catastrophic failure of the protocols using it, which naturally leads to the next topic.

What are the risks for protocols using Ethena?

Here is the English translation of the text, with the specified terms retained and not translated: The risks of Ethena can be broadly divided into two core risks: USDe liquidity and USDe solvency. USDe liquidity refers to the actual cash available, willing to purchase USDe at a benchmark price of $1 or up to 1% below that benchmark. USDe solvency means that even if Ethena may not have cash at a given moment (e.g., after a prolonged period of withdrawals), it can still obtain that cash by liquidating assets given enough time. For example, if you lent your friend $100,000 and they have a $1 million house. Indeed, your friend may not have the ready cash, and they may not be able to produce it tomorrow, but given enough time, they will likely be able to raise the money to pay you back. In this case, your loan is healthy; your friend just lacks liquidity, meaning their assets may take a long time to sell. Insolvency essentially means that liquidity should not exist, but limited liquidity does not mean insolvency. When Ethena collaborates with certain protocols (e.g., EtherFi and EigenLayer), they will only face significant risks if Ethena becomes insolvent. Other protocols, such as AAVE and Morpho, may face significant risks if Ethena's products lack liquidity for an extended period. Currently, the on-chain liquidity of USDe/sUSDe is around $70 million. While aggregators can provide quotes to exchange up to $1 billion of USDe for USDC at a 1:1 ratio, this is likely due to the current high demand for USDe, as it is based on intended demand, and this liquidity may dry up when Ethena faces large-scale redemptions. When liquidity dries up, Ethena will face the pressure of managing redemptions to restore liquidity, but this may take time, and AAVE and Morpho may not have enough time. To understand why this might happen, it is important to understand how AAVE and Morpho manage liquidations. When debt positions on AAVE and Morpho become unhealthy, i.e., exceed the required loan-to-value ratio (the ratio of the loan amount to the collateral), liquidation occurs. Once this happens, the collateral is sold to repay the debt, with fees charged, and any remaining funds returned to the user. In short, if the value of the debt (principal + interest) approaches a predetermined ratio compared to the value of the collateral, the position will be liquidated. When this happens, the collateral will be sold/converted into the debt asset. Currently, many people use these lending protocols, depositing sUSDe as collateral to borrow USDC as debt. This means that if liquidation occurs, a large amount of sUSDe/USDe will be sold for USDC/USDT/DAI. If all of this happens simultaneously, accompanied by other severe market volatility, USDe will likely lose its peg to the US dollar (if the scale of liquidation is very large, say around $1 billion). In this scenario, there could theoretically be significant bad debt, which would be acceptable for Morpho, as the treasury is used to isolate risk, although some income-generating treasuries would be negatively impacted. For AAVE, the entire core pool would be negatively affected. However, in this purely liquidity-driven potential scenario, adjustments to the liquidation management approach may be made. If liquidation could lead to bad debt rather than immediately selling the underlying assets into an illiquid market and letting AAVE holders bear the difference, the AAVE DAO could take on the responsibility of the tokens and positions without immediately selling the collateral. This would allow AAVE to wait until the price and Ethena's liquidity have stabilized, allowing AAVE to earn more during the liquidation process (rather than a net loss) and allowing users to receive their funds (rather than nothing due to bad debt). Of course, this system would only work if USDe returns to its previous value; otherwise, the bad debt situation would be worse. However, if there is an as-yet-undiscovered high-probability event that could cause the token value to go to zero, liquidation is unlikely to be better than waiting to realize more value, perhaps a 10-20% difference, as individual holders become aware and start selling positions faster than parameter changes. This design choice is crucial for assets that may face liquidity issues in bubble markets and could also be a good design choice for stETH before the Beacon Chain enables withdrawals, and if successful, it could also be an excellent way to increase the AAVE treasury/insurance system. The risk of insolvency is relatively reduced but not zero. For example, suppose one of the exchanges Ethena uses goes bankrupt. Of course, Ethena's collateral is safe with the custodian, but it suddenly loses its hedge and now must hedge in a potentially turbulent market. The custodian could also go bankrupt, as @CryptoHayes pointed out to me in a conversation in Korea. Regardless of the protective measures around the custodian, there could still be severe hacking or other issues, as crypto is still crypto, and there is still potential risk, even if these risks are highly unlikely to occur and may be covered by insurance, but the risk is not zero. What are the risks of not using Ethena? Now that we have discussed the risks of Ethena, what are the risks of not using the protocols that do not use Ethena? Let's look at some statistics. Half of Pendle's TVL (at the time of writing) is attributed to Ethena. For Sky/Maker, around 20% of their revenue is in some way attributed to Ethena. Approximately 30% of Morpho's TVL comes from Ethena. Ethena is now one of the primary drivers of AAVE's revenue and new stablecoins. Those well-known platforms that do not use Ethena or interact with its products in some way have essentially been left behind. There are some interesting parallels between Ethena's adoption and Lido's adoption within protocols. Around 2020 and 2021, the competition to be the largest lending protocol was more intense. However, Compound was more focused on minimizing risk, perhaps to a ridiculous extreme. AAVE integrated stETH as early as March 2022. Compound started discussing adding stETH in 2021 but did not do so until they proposed a formal proposal in July 2024. This timing coincided with when AAVE started to surpass Compound. Although Compound is still relatively large, with a total locked value of $2 billion, it is now just over one-tenth the size of AAVE, when it once dominated. To some extent, this can also be seen in the relative approaches of @MorphoLabs and @AAVELabs to Ethena. Morpho started integrating Ethena in March 2024, while AAVE did not integrate sUSDe until November, an 8-month gap during which Morpho saw significant growth, while AAVE lost relative control of the lending space. Since AAVE's integration of Ethena, its TVL has increased by $8 billion, and the yields for product users have increased significantly. This has led to an "AAVETHENA" relationship, where Ethena's products create higher yields, incentivizing more deposits, which in turn drives more lending demand, and so on. Ethena's "risk-free" rate, or at least its "normal" rate, is around 10%. This is more than double the current FFR (risk-free rate) value of approximately 4.25%. Bringing Ethena, especially sUSDe, into AAVE functionally increased the equilibrium lending rate, as AAVE's "benchmark" rate now inherits Ethena's benchmark rate, even if not fully 1:1, but rather in the ballpark. This is similar to what happened when AAVE previously introduced stETH, where the ETH borrowing rate roughly matched the stETH yield.

Here is the English translation:

In short, not using the Ethena protocol may face the risks of lower yields and lower demand, but it is necessary to avoid the risk of serious de-pegging or collapse of the USDe price, which may be negligible. Systems like Morpho, due to their independent structure, may be better able to adapt and avoid potential collapses. Therefore, it is understandable that systems based on larger liquidity pools, such as AAVE, take longer to adopt Ethena. Now, while most of the content is a review of the past, I would like to offer some perspectives more focused on the future. Recently, Ethena has been working hard to integrate DEXs. Most DEXs lack short demand, that is, users who want to short the contract. Generally, the only type of user who can do this on a large scale and continuously is delta-neutral traders, of which Ethena is the largest. I believe that perpetual contract platforms that can successfully integrate Ethena while maintaining a good product can escape competition in a very similar way to Morpho by working closely with Ethena to escape their smaller competitors.