Liquid Restaking Wars: the Next Lido

Tranche warfare: Which LRT Protocol will capture the most market share? What are the risks associated with liquid restaking?

Shoutout to those that helped the ideation of this piece, excited to work together in the future. Thank you anon.

The pandora’s box of restaking has been opened, and with it, a whole host of questions around mass slashing events and centralization. In true crypto fashion, the first product built on top of EigenLayer is a way to tokenize this restaked position, unlocking the utility of the capital securing Ethereum and other services: liquid restaking. After seeing the prominence of liquid staking tokens (LSTs) and the dominance of players like Lido, all eyes are on liquid restaking token (LRT) providers, as speculators search for the next Lido. With over 15 LRT providers, which one will capture the most users? What should these speculators look for as signal in such a saturated space? The answer to these questions largely center around one theme: risk management.

Side note: this piece will not go into the basics of restaking. For that, I recommend reading my previous writing on restaking before reading the following on liquid restaking.

How (Liquid) Restaking Works

In short, restaking involves rehypothecating the ETH used to secure the Ethereum network to secure other networks. What this means is the same ETH that is staked to Ethereum, slashable by Ethereum, is also staked to another service, slashable by that service. The ETH that is staked is now not only agreeing to the rules of Ethereum, but also to the rules of another service. The validator’s ETH gets slashed if it violates the rules of Ethereum or the rules of the other added service. These added slashing conditions come with added rewards as the new added service that the ETH is staked to will pay out staking rewards.

The new service that the ETH stake is securing could be an L1, a data availability layer, an oracle, etc. These restaked services are called actively validated services (AVSs). Within each AVS, there are different node operators. This means that Ethereum restakers must decide which AVS to restake to and which operator they want to delegate their stake to. Each AVS will have a different risk profile, and each operator on that AVS will have a different risk profile.

Ethereum stakers are the supply and AVSs are the demand. AVSs demand stake from Ethereum stakers and pay stakers in the form of restaking rewards.

Liquid restaking is functionally similar to liquid staking. Liquid staking unlocks the liquidity of staked Ether by tokenizing it. stETH and rETH represent some amount of ETH staked to the Ethereum network, and by unlocking this liquidity, it can be used in DeFi.

Risk Fungibility

Due to the different risk profiles for each operator, risk is no longer fungible. This is different from traditional staking where all nodes are doing the same thing by participating in the same execution and consensus layer. All Ethereum validators are proposing blocks on Ethereum, making it easy to apply a standard framework to evaluate their risk and performance. There is relatively little risk in staking which is why LSTs are so fungible. It doesn’t matter who the Lido operator is, all stETH gets the same yield and is treated the same. However, an AVS could be an oracle, sequencer, bridge, pre-commit sponsor, DA layer, or even a validator on a completely separate L1. This wide range of services creates different slashing conditions that don’t have a standardized method of evaluation. In addition to the slashing risks of different AVSs, there are technical risks such as bugs.

The first layer of added risk is in the AVS, and the second layer of risk is in the node operator. With more sophisticated slashing conditions, non-deterministic conditions, and even potentially increased hardware requirements, who the operator is will play a much larger role in the amount of risk the restaker takes on.

The increased risks associated with each AVS and operator demands more rewards, which will be restaking rewards that the AVS will pay out in their native token. However, there is no way to evaluate return relative to risk. There is no way to determine if the rewards given by AVSs fairly compensate the restakers for the amount of risk they take on.

Risk Tranching through LRTs

LRT protocols are intermediaries between Ethereum holders and restaking. While the average user could decide which operator and AVS to stake to, it is much easier for the user to have the LRT protocol allocate stake across operators and AVSs and pay the LRT some percentage of rewards. The popularity of LSTs like Lido are in part because the user doesn’t have to decide for themselves how to delegate stake across operators, they can have Lido do it. Lido also arguably does a better job at delegating stake than users would. LRTs will perform the same job as Lido but with considerably more risk. They will need to thoughtfully determine which AVSs to onboard and which operators to use.

Users will likely prefer this over selecting AVSs and operators themselves. Adoption of LRTs over interacting directly with EigenLayer is also preferable from a fungibility perspective because all users of the same LRT protocol will be taking on the same amount of risk. For example, if Ether Fi hand selects a group of AVSs and which operators to delegate to, all eETH will have the same risk profile. If each LRT provider uses a separate set of AVSs, operators, or even allocates different proportions between the different AVSs and operators, each LRT will have a different risk profile. eETH will be different from ezETH which will be different from rstETH, making Ether Fi, Renzo, and Restake Finance all serve as different risk tranches. Based on the risk profile each LRT protocol creates, different properties might emerge (more on that later).

In this sense, when looking at the business model of LRT protocols, they are similar to banks. Liabilities are user deposits where the LRT has to account for callable nature of these liabilities, and the assets of an LRT are the subset of operators and AVSs they decide to restake to. The LRT becomes the agent in managing operator risk, AVS risk, and duration risk with unbonding periods, while the depositor is the principal.

LRTs serve as good risk tranches for retail investors who are looking for fungibility, utility, and capital efficiency, but institutional restakers might have different needs. For regulatory and safety reasons, institutional restakers might only want to restake to select AVS services and use specific operators. Institutions will likely be unable to restake to privacy AVSs for regulatory reasons and unable to restake to degen AVSs like a memecoin centric service for reputational reasons. They will also likely be required to KYC their operators and only use select professionals.

Generally speaking, there are two main outcomes for how restaking will work. The first outcome is that these restaking services will be bundled. A third party, such as an LRT provider, will perform due diligence on AVSs and operators and allocate accordingly based on the desired level of risk, the desired return, partnerships, etc. This will create risk fungibility by individuals all using the same AVSs and operators, enabling the creation of LRTs. Each LRT will likely serve as a different risk tranche with different risk profiles and restaking rewards.

The second outcome is where individuals will personally select what AVSs to restake to and which operators to use. This will likely only be applicable for institutions who need more control and KYC over their counterparties, but it will remove fungibility of risk between positions. This lack of fungibility likely won’t be an issue for these institutions as they will be less interested in unlocking the liquidity in their restaked positions to go farther out on the degen curve. If they do need to underwrite their positions to unlock liquidity, they will have few barriers working with other established centralized lenders.

Cascading Depeg Risk

Staking is akin to a perpetual duration bond. Anyone who has stETH has essentially lended a node operator their ETH, and in return, the lender gets paid interest (staking rewards). When the lender (depositor) wants their ETH back, the node operator must return them their ETH in exchange for the stETH the node operator gave to the lender. However, PoS has an unbonding period to prevent network attacks, resulting in the possibility for delayed redemptions. This unbonding period creates duration risk that the node operators must manage. One solution is to have 10% of the lended ETH held liquid instead of staking it. This way up to 10% of total ETH lent to the operator can be redeemed immediately, but if there is a “bank run”, a duration mismatch could occur which could be catastrophic (ex. Silicon Valley Bank). Restaking takes this duration risk even further because now the unbonding period is reliant on:

1. AVS details
2. EigenLayer details (7-day escrow period)
3. L1 details

This means LRTs might be unredeemable for the underlying ETH if there is a run on the “bank” (run on the protocol). A rush to exit might be sparked by a slashing event where the value of ETH held by an LRT protocol is lower than the value of the LRTs outstanding. Lido’s way to mitigate this effect is through bunker mode. Bunker mode pauses withdrawal requests until the negative event consequences are resolved. The intention behind bunker mode is to prevent sophisticated stakers from exiting before unsophisticated stakers, leaving the unsophisticated stakers holding the bag when the music stops. This socializes the impact of a mass-slashing event more than would otherwise, reducing the probability of a depeg event. However, stETH holders can still swap to ETH in liquidity pools. Thin liquidity may cause the stETH:ETH peg to fall below 1. In such a situation where LRTs trade at a discount, a cascading liquidation event could occur if many people borrow against their LRTs. While Lido can’t control all market forces, bunker mode mitigates the probability of depeg events and catastrophic externalities from slashing. Something similar will likely be implemented in LRTs.

How does this slashing risk appear for different LRTs? If one LRT is taking on significantly more risk than another one, surely they can’t be treated the same, right? They likely won’t. If an LRT has a significantly higher restaking yield than another, potentially indicating higher risk, lending markets like Aave might offer lower LTV ratios. For example, if eETH comes with higher risk and higher yield than rstETH, perhaps I will only be able to borrow $500 against $1,000 of eETH while I can borrow $800 against my $1,000 of rstETH. This will be a decision for lending markets as they balance the demand for liquidity with the risk of bad debt. The main areas that a depeg event would be dangerous is if there are large quantities of LRTs using leverage, doing carry trades, or are looping.

Centralization Risk

Stake Centralization

Centralizing pressure appears in several ways in restaking and liquid restaking. Going from lowest to highest in the value chain, the first instance of centralizing pressure is in different restaking rewards. In traditional staking, each ETH staked gets the same amount of rewards. With restaking, there are different APYs based on how well restakers can pick AVSs and operators. This is centralizing in the same way MEV is centralizing — the top players will slowly get more and more capital and more control over the network. This is a chronic distortion risk, as it occurs slowly over time. However, delegation eliminates this centralization risk by creating a new problem. By allowing individuals to delegate to the restaker (or LRT) with the highest APR, unsophisticated restakers can get the same APY as sophisticated restakers. The new problem this creates is an acute distortion risk related to slashing. Slashing risk becomes outside the control of the restaker and is in the hands of the delegated party (LRT).

Some might argue that even centralizing pressure to one LRT provider is bad. Even though the stake is sourced from many individuals, a single LRT provider should not be delegated that much stake. This is true — while centralization of delegation is an improvement from centralization to a single party, it is still dangerous in terms of mass slashing events, network attacks, etc. However, centralization under an LRT might not be a problem if there are other ways to get best-in-class yield. For example, if a restaker can get 10% APY through Ether Fi and 10% APY as a home restaker without taking on additional risk, they would be indifferent. It is possible that some restaking services will pay higher restaking rewards to at-home restakers, reducing the incentive to delegate and incentivizing decentralization. Some applications need economic trust through the amount of stake, while other applications need decentralization trust. If redundancy and decentralization is important, the market will naturally trend towards paying higher rewards and incentivizing decentralization.

Operator Centralization

Due to the added complexity of AVSs, most operators will likely be professional node operators. Simply Staking, Consensys, and Coinbase are examples of centralized entities that will likely be some of the most widely used operators for restaking. Trusted professionals like these will likely be some of the first operators that restakers and LRTs will trust. In general, centralized entities are better at underwriting and assessing risk than decentralized entities are. However, centralization amongst operators will only be a temporary problem. As time passes and being an operator for different AVSs becomes more widely understood, it will be easier to evaluate the risks associated with being an operator on different AVSs. Lido is a perfect example of this as they started with 19 different operators in Q4 2021 and as of Q4 2023 used 35 different operators. The same will likely play out, where stake will be decentralized across individually centralized operators.

A natural follow up question to the idea of LRTs deciding which AVSs and operators to restake to is: is there any decentralization in LRTs, or is the LRT effectively an on-chain bank? As of now, the answer is yes, it appears the architecture of the hottest new LRTs are centralized. This might not be a problem. With many different options for liquid restaking, it is easy for users to take their stake elsewhere if the LRT provider is not acting in the best interest of depositors. However, it would still be desirable to have a decentralized process in true hyperstructure fashion. This is where node operators come in.

Decentralized LRT Architecture

To properly assess the risks associated with restaking to an AVS and operating as a node for that AVS, the underwriter will have to be somewhat of a professional. One way to do effective risk management is through a centralized entity such as Coinbase, Figment, or an LRT provider. Another way is to consult with node operators who better understand AVS risk and operator risk.

An example of decentralized risk management can be seen in Maple Finance. Maple operates as an onchain credit facility issuing undercollateralized/uncollateralized stablecoin loans to real world borrowers. These borrowers are typically businesses in countries where the local currency is not as stable as the US dollar, which makes individuals want to borrow in USD(T). To assess the credit-worthiness of borrowers, Maple utilizes a handful of credit professionals (delegates) who can apply to underwrite loans. These delegates negotiate loan terms with borrowers, perform due diligence, and they operate on first loss capital to make sure they are financially incentivized to issue good loans. In the context of liquid restaking, professional ode operators could serve as delegates who can accurately assess risk and perform technical analysis to understand the code, the clients, the delegate will know how to run the node, and will know how much the node will cost.

An LRT DAO could be half operator led, where operators perform the technical analysis, and the other half of the DAO is financial professionals. The financial analysis performed will involve determining the proper allocation amongst operators and AVSs, and the restaking sharpe ratio discussed in my previous article. This LRT structure involves governance and a decentralized set of delegates which is different from the centralized LRT approach discussed previously where there is minimal governance over what AVSs and operators to restake to.

Governance processes on both the technical and financial sides of the DAO could occur in a manner similar to Lido’s Easy Track. Easy Track motion is a lightweight voting process for decisions that don’t spark the need for wider DAO involvement. To qualify for Easy Track, proposals must be below some minimum objection threshold. The most common uses are:

• Node operators increasing staking limits
• Funds being allocated to LEGO program (Lido Ecosystem Grants Organization)
• Funds being allocated into reward programs

The Next Lido

While it is unclear which LRT protocol will capture the most market share, there are certainly things to look for that might be signal. When looking at Lido’s history and what made it capture the most market share, one of the important factors was that it felt safe. It was as ubiquitous to ETH as possible and depositors felt safe using Lido. One aspect of why this was possible was because they did not decentralize their validator set too early. They were fairly centralized to begin with and it wasn’t until February 2023 that the project introduced Staking Router improvements in their v2 upgrade, paving the way for a more decentralized node operator set. Additionally, getting the support from politically influential people in the Ethereum community was important. The words of researchers and developers in the Ethereum community carry a lot of weight when it comes to technical implementations. In restaking, this might look like members from the EF or Ethereum research team tweeting about an LRT protocol expressing support, or even researchers from EigenLayer.

It’s more likely than not that the most successful LRT will not be the one dumping the most degen yields to its holders, but rather the one that has the safest implementation. Ethereum and EigenLayer researchers will likely say that LRTs are dangerous, but if you’re going to make them, we want the adults in the room. One way enhanced safety could appear is as an insurance fund. If an LRT takes a percentage of their restaking rewards and sets them aside as an insurance fund, this could mitigate the impacts of a slashing event.

Another important part of Lido’s dominance was its early integration with Aave. This started a flywheel of liquidity that gave users more flexibility with their stETH. In the same way, LRT integrations with both blue chip DeFi protocols like Aave and up-and-coming projects like Pendle could serve as catalysts for greater market share.

Concluding Thoughts

There are many unknowns when it comes to LRTs and how they will be implemented. What will yield be? Who will have insurance funds? What other DeFi integrations exist? As of now, it appears that EtherFi is winning with $3.25 billion in TVL, and Renzo in second with $1.67 billion in TVL. While first movers’ advantage may serve this project well, this is a dynamic industry and there is no telling what the leaderboard of LRTs will look like. EtherFi’s recent token launch success shows that there is an appetite for LRTs (or airdrops), but it is still hard to paint an accurate picture of what the ecosystem will look like a year from now.

Sources

Broadhead, Ronan. “A Look into Etheruem’s Future Staking Market Dynamics.” Medium, Collab+Currency, 10 Nov. 2023, medium.com/collab-currency/a-look-into-etheruems-future-staking-market-dynamics-16fdfd6bba62.

Brown, Owen, and Sam Wolter. “$MAPLE Investment Report.” Oregon Blockchain Group, DormDAO, static1.squarespace.com/static/53e7dd4fe4b0fb1fc62c318a/t/5a4a6df271c10bd7b34daded/1514827259343/Sample_SS1.pdf. Accessed 29 Mar. 2024.

Drake, Justin. “Justin Drake — Restaking Alignment, Consensus Overload, Protocolizing El.” YouTube, Restaking Summit, 29 Nov. 2023, youtu.be/dNJ5Tj2VDUo?si=VCu3rWqCoO88T_Zw.

EigenLayer. “On Liquid Restaking: Risks & Considerations.” EigenLayer Blog, EigenLayer Blog, 22 Mar. 2024, www.blog.eigenlayer.xyz/liquid-restaking/.

“Guide to Easy Track.” Lido Docs, docs.lido.fi/guides/easy-track-guide/#:~:text=Easy%20Track%20Motion%3F-,%E2%80%8B,threshold%20hasn’t%20been%20reached. Accessed 29 Mar. 2024.

hahahash. “Eigenlayer Ecosystem Metrics | the Ultimate Restaking …” Dune, dune.com/hahahash/eigenlayer. Accessed 29 Mar. 2024.

Ippolito, Mike, and Myles O’Neil. “Liquid Restaking Is Coming Sooner than You Think.” YouTube, Bell Curve Podcast, 22 Nov. 2023, youtu.be/JhYXRih3Ny0?si=dzmAMUCGpFzb1iM6.

Ippolito, Mike, et al. “Converging Cosmos & Ethereum: A Restaking & Shared Security Deep Dive: Sreeram & Zaki.” YouTube, Bell Curve Podcast, 9 Jan. 2024, youtu.be/auDGJXsfvJ4?si=CZheWRvII7Ip0x2M.

Kannan, Sreeram, et al. “Alignment Panel.” YouTube, Restaking Summit, 29 Nov. 2023, www.youtube.com/watch?v=lEpE3Mg2SPk.

Lido. “Introducing Lido v2 — next Step in Decentralization.” Lido Finance, Lido Finance, 12 Sept. 2023, blog.lido.fi/introducing-lido-v2/.

Neuder, Mike, et al. “The Risks of Lrts.” Ethereum Research, 26 Feb. 2024, ethresear.ch/t/the-risks-of-lrts/18799.

Rosic, Ameer. “Lido vs Rocket Pool: Best ETH Staking Solution.” Blockgeeks, 18 Nov. 2023, blockgeeks.com/guides/lido-vs-rocket-pool-best-eth-staking-solution/.

rp_community. “LST Comparison.” Dune, dune.com/rp_community/lst-comparison. Accessed 29 Mar. 2024.

Smith, Dan, et al. “Liquid Restaking Tokens: Crypto’s next Timebomb? | Analyst Round Table.” YouTube, 0xResearch, 10 Jan. 2024, www.youtube.com/watch?v=I0wPkKEvPok.

“Vanom: Lido on Ethereum Validator & Node Metrics.” Hex, app.hex.tech/8dedcd99–17f4–49d8–944e-4857a355b90a/app/3f7d6967–3ef6–4e69–8f7b-d02d903f045b/latest?tab=client-diversity. Accessed 29 Mar. 2024.

Wind, Peter. “Lido vs. Rocket Pool — Which ETH Staking Solution Is Best?” CoinCodex, CoinCodex, 29 May 2023, coincodex.com/article/27799/lido-vs-rocket-pool-which-eth-staking-solution-is-best/.

“‘bunker Mode’: What It Is and How It Works.” Google Docs, Lido, docs.google.com/document/d/1NoJ3rbVZ1OJfByjibHPA91Ghqk487tT0djAf6PFu8s8/edit#heading=h.2ocdo4vk3roe. Accessed 29 Mar. 2024.