Delphi: Why Ethena is the Focus of Our Investment in This Bull Market

Original Title: 《 Ethena Thesis --- The Internet Bond》

Original Author: José Maria Macedo, José Maria Macedo

Original Translation: Rhythm Worker, BlockBeats

As one of the "golden hands" in the last crypto bull market, Delphi gained fame by betting on the GameFi pioneer Axie Infinity and the algorithmic stablecoin Terra Labs (composed of governance token Luna, algorithmic stablecoin UST, and DeFi protocol Anchor), while also suffering from the backlash of the UST collapse. Even team member Larry lost most of his funds in the crypto space due to the Terra ecosystem project Anchor.

Despite this, Delphi Labs CEO José Maria Macedo has once again chosen the "algorithmic stablecoin" Ethena on behalf of himself and Delphi Ventures, stating that Ethena is one of their most steadfast and unwavering investments in this bull market.

In this article, Delphi explains why Ethena's design is clever, how USDe differs from earlier algorithmic stablecoins, what improvements it has over previous similar protocols, and most importantly— the risks of Ethena. Will USDe encounter a moment like UST (Terra's stablecoin)? Rhythm will translate the full text as follows.

For both Delphi Ventures and myself, Ethena is one of our most confident investments this cycle. I believe:

sUSDe will provide the highest dollar yield in cryptocurrency on a large scale;

USDe will become the largest stablecoin outside of USDC/USDT in 2024;

Ethena will become the highest-earning project in all of cryptocurrency.

In this article, I will introduce what Ethena is, why it is interesting, and break down the risks I currently see.

Opportunity

It is undeniable that stablecoins remain one of the killer applications of cryptocurrency.

The market has repeatedly indicated its desire for stablecoin yields. The challenge is to generate it in an organic and sustainable way.

Ethena is able to provide this yield, with stablecoins as a byproduct. Stablecoins generate yield, while the capital used to mint stablecoins generates yield.

Specifically, the funds used to support the stablecoin are staked in ETH and shorted in perpetual contracts, maintaining a delta-neutral exposure, with both sides of the position typically providing yield (i.e., staking ETH for PoS yield and shorting ETH for funding fee income).

sUSDe Yield = stETH Yield + Funding Fee Rate (currently 35.4%)

In this way, Ethena effectively combines the two largest sources of "real yield" in cryptocurrency: ETH staking (approximately $3.5 billion/year) and perpetual contract funding fee rates (approximately $37 billion in OI between ETH/SOL/BTC, with an average yield of about 12%).

This is the realization of Arthur Hayes' original idea of "synthetic USD." While similar delta-neutral stablecoin projects (like UXD) have been attempted before, they have never been able to leverage the liquidity of centralized exchanges (CEX).

Ethena and the Stablecoin Trilemma

Before delving into the design and its risks, it is necessary to provide a brief summary/history of stablecoin design and their position in the stablecoin trilemma.

There are three popular forms of stablecoins: over-collateralized, fiat-backed, and algorithmic stablecoins. Each addresses different aspects of the stablecoin trilemma (i.e., decentralization, stability, and scalability/capital efficiency), but ultimately fails to solve all three problems.

Fiat-backed (USDC, USDT)

• Stability: Authorized participants (i.e., market makers) can mint and redeem them at arbitrage prices, ensuring they remain pegged.
• Scalability: They are 1:1 collateralized, thus offering scalability + capital efficiency.
• Decentralization: Highly centralized, meaning holders face counterparty risk (bank solvency, asset seizure, etc.) and censorship risk, as legal entities may be coerced and bank accounts may be frozen.

Over-collateralized (DAI)

• Stability: Anyone can mint and redeem the underlying collateral and arbitrage, creating stability.
• Scalability: There are challenges in scalability, as it primarily exists as a byproduct of leveraged demand. Aave and other products further surpass this functionality.
• Decentralization: Highly decentralized compared to alternatives, although it relies to some extent on centralized stablecoins and government bonds as collateral.

Algorithmic Stablecoins

• Scalability: Algorithmic stablecoins have high capital efficiency and scalability, as they can be minted without external collateral and typically distribute some form of yield to participants when demand exceeds supply.
• Decentralization: They are also decentralized, as they often rely solely on crypto-native collateral.
• Stability: However, they fail miserably in stability, as they are only supported by endogenous collateral, leading to reflexivity and ultimately collapsing through a death spiral. Every attempted algorithmic stablecoin has suffered this fate.

So What About USDe

In my view, USDe is the most scalable fully collateralized stablecoin ever. It is not fully decentralized yet, nor is it likely to be, but it is at a very interesting point of balance.

Stability

USDe is fully collateralized by a delta-neutral position, offset by staked ETH and shorted ETH in perpetual contracts. Authorized participants can redeem the stablecoin for the underlying collateral, which should provide stability. That said, this is a new design that clearly has risks (which will be detailed later). It is also less likely to be as stable as fiat-backed stablecoins, as the redemption cost for these stablecoins is free, while the redemption cost for USDe will depend on the liquidity conditions at the time (i.e., the cost of closing the short position).

Scalability

This is where USDe truly shines, for two main reasons. First, like fiat-backed stablecoins, Ethena can mint against collateral at a 1:1 ratio. However, unlike fiat-backed stablecoins, Ethena can generate meaningful organic yield at scale for its holders. Specifically, USDe can be staked into sUSDe for protocol yield, which is a combination of stETH yield and funding fee rate.

sUSDe Yield = stETH Yield + Funding Fee Rate (currently 35.4%)

Crucially, this yield can be: a) scalable, b) counter-cyclical to treasury yields.

Regarding scalability: Ethena effectively combines the two largest sources of "real yield" in cryptocurrency:

ETH staking: $3.5 billion output per year;

Funding fee rates in perpetual contracts: approximately $35 billion in positions between ETH and BTC (released this week), with an average yield of about 11% over the past three years.

As we have seen over the past three months, this number can be even higher during bull markets, with average funding rates around 30%.

Ethena can also add other assets over time, such as BTC ($25 billion in positions) and SOL (possibly including jito SOL), to further expand the supply scale.

Regarding counter-cyclicality: Over time, treasury yields may decline, and as people move further along the risk curve, demand for crypto leverage should rise. As yields supported by US treasuries decline (due to rising leverage demand), Ethena's yields should remain elevated.

Decentralization

Decentralization is a multi-dimensional spectrum that ultimately depends on the weight of each dimension. Personally, I believe Ethena is positioned between fiat-backed and over-collateralized stablecoins in terms of decentralization.

It is more resistant to censorship than fiat-backed stablecoins, as it does not rely on traditional banking rails, which ultimately depend on the Federal Reserve through proxy banks and can be shut down overnight. Arthur described this well in his recent article.

However, it does face some counterparty risk from CEX. Specifically, Ethena holds collateral in exchanges with institutional-grade custodians in an MPC wallet, then maps the collateral to CEX using Copper, Ceffu, and Cobo. Settlements occur every 4-8 hours, and profits during the settlement period help mitigate trading risks.

More importantly, unlike over-collateralized stablecoins that can be minted/redeemed on-chain without permission, Ethena relies on calling off-chain servers to calculate the CEX with the most efficient funding rates and mint USDe. This is undeniably a centralized factor that makes it susceptible to censorship.

Profitability:

Unlike most other crypto projects, Ethena's profitability is also very high. It has become the most profitable dApp in the cryptocurrency space, surpassing all DeFi, with only Ethereum and Tron generating more revenue in 30 days.

Ethena's profitability comes from all the yield it generates. Currently, these funds are deposited into an insurance fund, but ultimately, it is expected that these funds will be distributed to stakers at a certain ratio.

Assuming a distribution rate of 10%, Ethena's protocol revenue is:

Total Yield * (1 - 90% * (1 - sUSDe Supply / USDe Supply))

It is worth noting that due to Ethena's activities, its current profitability is high, with a staking rate of only about 30% due to the incentive of locking USDe.

This dynamic also highlights why the success of USDe as a stablecoin is beneficial. The more USDe is used as a stablecoin, the less USDe is staked, and the higher Ethena's profits.

Risks

The most common FUD I have seen people focus on is the risk of funding rates, specifically what happens if funding rates remain negative for an extended period? Will we see a de-pegging or collapse similar to UST?

In this regard, it is worth noting:

• Historically, funding rates have typically been positive.
• There is an insurance fund (IF) to cover periods of negative funding.
• Most importantly, even in the worst-case scenario, where funding is negative for an unprecedented period and the IF is completely exhausted, USDe is fully collateralized by external assets and has a certain degree of "anti-reflexivity" built into its design, making it very different from UST.

Historical Situation of Funding Rates

Historically, funding rates have been positive, especially when considering Ethereum staking yields. Over the past three years:

• On an OI-weighted basis, funding has averaged positive 8.5%.
• The funding rate has only been negative on 11% of days when excluding staking ETH yields.
• The maximum consecutive days of negative funding was 13, with a total of 110 days of negative funding.

There may be reason to believe that funding rates will remain structurally positive over the long term. Some exchanges (Binance, Bybit) have benchmark funding rates of 11%, meaning that if funding rates fall within a certain range, they will default back to 11%. These exchanges account for over 50% of the positions. Even when we look at TradFi, CME Bitcoin futures are larger than Binance, with the current spread around 15%. Generally speaking, as a proxy indicator of capital costs, futures yield spreads are positive the vast majority of the time.

Insurance Fund (IF)

When funding rates do turn negative, there will be an insurance fund to subsidize sUSDe yields and ensure they do not drop below 0 (i.e., will never be negative).

A portion of the protocol revenue will be redistributed to the IF to ensure its organic growth over time. Ethena Labs has provided a $10 million donation to kickstart the IF.

Currently sitting at $27 million, all protocol revenue is sent there (approximately $3 million per week at the current run rate).

The Ethena team and Chaos Labs have conducted in-depth research to determine the optimal size of the insurance fund. Their recommendation is to match $20 million to $33 million of insurance fund for every $1 billion of USDe.

Anti-Reflexivity

Now, let’s assume a scenario where funding rates are negative and exceed stETH yields, and this persists long enough to exhaust the insurance fund.

In this case, the principal balance of the stablecoin will slowly drop below $1, as funding rates are paid through the collateral balance. While this sounds bad, the risk here is very different from adjustable risks, as the collateral will slowly decrease over time rather than rapidly crashing to 0. For example, Binance's maximum negative funding rate of -100% means a daily loss of 0.273%.

As Guy pointed out, this exogenous funding rate actually embeds "anti-reflexivity" or negative feedback loops into the design.

Negative yield → Users redeem stablecoins → Shorts close → Funding rates recover above 0

The redemption of stablecoins helps balance funding rates and restore equilibrium to the system. This is in contrast to algorithmic stablecoins, where redemptions lower the price of the corresponding tokens and create feedback loops that constitute the so-called "death spiral."

Additionally, two other points are worth noting:

1) When funding rates turn negative, de-pegging may not happen suddenly, but rather gradually as yields decline over time. If the same yield can be obtained from treasuries (or RWA products), why hold USDe?

2) The insurance fund is a design choice intended to optimize the user experience for sUSDe holders by smoothing yields and preventing them from worrying daily about principal loss.

Ethena could choose to pass negative yields onto holders, as Cobie suggested below, which would encourage people to redeem funds more quickly in response to funding changes, thereby strengthening the negative feedback loop.

Other Risks

While I do not believe negative funding rates are a particularly large risk, there are certainly many other risks to consider.

After all, this is a brand new mechanism that can provide very high yields. No yield comes without risk, and the higher the yield, the more skeptical one should be.

Here is a non-exhaustive list of risks and mitigations that I see:

1. Historical funding rate data does not include Ethena itself. If USDe becomes large enough relative to the overall positions, it could: a) significantly lower the average funding rate b) exacerbate funding rate volatility, potentially leading to violent liquidations, poor execution, and potential USDe de-pegging.

Related to this, the yield of stETH may continue to decline over time, further damaging the economics and worsening the above issues. This is definitely a risk. Here are some mitigations:

a) The unstaking of sUSDe has a 7-day delay, which will alleviate market panic when there is a large supply to be staked;

b) Even in the worst-case scenario, this de-pegging would not have too severe an impact on the protocol's solvency, as the spread would lead authorized participants to redeem. This would primarily harm users redeeming at a loss and, more importantly, harm protocols/users using leverage on USDe.

2. The liquidity of LST collateral is relatively poor, which could lead to significant penalties and/or de-pegging. A sufficiently violent de-pegging could lead to Ethena being liquidated and incurring losses.

However, given the limited or no leverage used by Ethena, only an unprecedented de-pegging would lead to liquidation. According to Ethena's own research, this would require LST to de-peg from ETH by 41-65%, while the highest de-pegging ratio for stETH in 2022 was about 8%.

Ethena is now also diversifying its LST exposure, further mitigating this issue, with only 22% of collateral being LST, while ETH currently accounts for 51%. When funding rates reach +30% in a bull market, the 3/4% stETH yield becomes less significant, so Ethena may hold more ETH in a bull market and more stETH in a bear market.

3. Ethena faces credit risk from CEX short counterparties. A counterparty collapse could mean: a) Ethena ends up being net long rather than delta-neutral b) USDe de-pegs based on its P&L exposure to specific counterparties.

However, Ethena settles with CEX every 4-8 hours, so they are only affected by the differences between two settlement periods. While this could be significant during rapid and severe market fluctuations, it is completely different from having all capital affected.

It is also worth noting that, as we saw with the USDC incident last May, all stablecoins carry a certain degree of counterparty risk.

4. Once USDe begins to be looped with leverage, all of the above risks could be amplified and become systemic.

This would certainly lead to some panic, cascading liquidations, and de-pegging. These situations could be more destructive for users and protocols composed of USDe rather than Ethena itself. However, in extreme cases, it could also harm Ethena.

The only way to re-peg is to redeem the underlying asset and close the short position. If liquidity is thin, this step could lead to significant losses.

5. Ethena Labs and the associated multisig have control over the assets (currently a ⅔ multisig with Ethena, Copper, and an independent third party).

Theoretically, they could exert influence over the assets off-chain or otherwise harm the assets, with USDe holders having no legal rights and needing to resolve this issue in a court without precedent.

6. Ethena could also face regulatory bans and be required to freeze assets, thereby indirectly controlling a bunch of ETH/stETH.

7. Finally, there may be many unknowns.

Ethena operates as a tokenized hedge fund on the backend. This is difficult, with many moving parts and ways things can go wrong. Do not invest more than you can afford to lose.

Everything in cryptocurrency carries risk, as we have repeatedly discovered through difficult processes. In my view, it is important to be as transparent about risks as possible and allow individuals to make their own decisions.

I want to say that the Ethena team has overall done a great job in this regard, providing the most comprehensive documentation and risk disclosures I have seen in early projects.

For my part, I have had many personal positions exposed to Ethena since before the shard activities, purchasing a bunch of USDe/sUSDe Pendle YT and investing through Delphi Ventures. As you may now know, this is one of the projects I am most excited about this cycle.

I still believe stablecoins represent a $10 billion opportunity. Ethena presents a very interesting perspective on the stablecoin trade-offs and is hard to match at this scale and beyond.

I also believe Guy is one of the best founders we have supported, transforming Ethena from an idea into the fastest-growing dollar-denominated crypto asset ever, with a TVL of $1.5 billion.

This time, he has assembled a star team to realize his vision and surrounded himself with some of the best supporters in the field (top CEXs, VCs, market makers, etc.), and I look forward to seeing how he develops in the coming years.

Thanks to Yan Liberman for helping me brainstorm and organize this article, to 0xDef1, Jordan, and Conor Ryder for their reviews, and to Guy Young for answering all my silly questions.