We would like to thank Justin Drake from Ethereum Foundation, Amir Forouzani from Puffer, Rok Kopp from EtherFi, Zhuling from Bedrock, Alan Curtis from Rio, Brianna and Sreeram from EigenLayer, Pablo Villalba from Diva, Daniel Dizon from Swell, Kratik Lodha from Renzo, Matthias Ang from SSV, and Max from Obol, for their generous comments and contributions.
Contents of this article:
The gravitational force pulling in all of crypto.
Restaking has become a gravitational force in crypto. And at its essence, restaking represents modular security.
Infrastructure such as EigenLayer, Babylon, and Cosmos ICS allow the same tokens to stake on multiple protocols, thereby decentralizing trust beyond their sovereign chains. Through LRTfi, the economic security of one chain, can compose with the economic bandwidth of other chains. The security of multiple chains is interweaving into one meshed network.
On the positive side, restaking lowers the threshold of launching middleware. Several AVSs are the most highly anticipated launches this year. Because slashing is so rare, economic security may be over-collateralized as things currently stand, and can find additional utility beyond their sovereign chains.
On the negative side, restaking has seen some controversy due to its externalities. An attack on a foster chain can lower the economic security of its parent chain. Vitalik has openly conveyed that Ethereum should not be expected to fork if for some reason validators face externalities from restaked middleware:
Across ecosystems, we are seeing innovation in restaking in the 1:1, 1:N and N:N models. Tokens with strong fundamentals, such as Bitcoin, Ethereum, and Solana will likely find middleware willing to adopt its economic security.
Babylon
Babylon allows the restaking of BTC to PoS chains. This feat is enabled by two key innovations: merged mining and ingenious use of Bitcoin opcodes.
Through merged mining, the same miner can mine for both Bitcoin and Babylon blocks, as long as they use the same signature, at no additional energy cost. Babylon blocks are sent to Bitcoin blocks as a transaction, thereby making Bitcoin the ultimate DA layer. To enable bonding, withdrawal, and slashing, Babylon uses Bitcoin opcodes to support four transactions, ‘staking transaction’, ‘unbonding transaction’, ‘slashing transaction’, and ‘unstaking transaction’.
EigenLayer
In Eigenlayer, Ethereum and ERC20 tokens are restaked on middleware services, dubbed as AVS (actively validated services).
A ZK proof is generated to validate that a particular EigenLayer validator belongs to the associated validator on the Beacon Chain. An EigenPod is instantiated for each validator. Once a state root proof is verified along with a ZK proof provided through an Oracle, the validator can fully restake their ETH to AVS and start to receive rewards. Several interesting edge cases occur, if a validator is slashed on the Beacon Chain, anyone can then call the EigenPod.verifyBalanceUpdate and the pod Owner pays the negative shares.
LRTs showcase modular security. They allow the same token from one chain’s economic security to participate in other chains’ economic security and bandwidth. As a novel asset class, LRTs have distinct and advantageous features:
LRT protocols, such as EtherFi (eETH), Rio, Bedrock (uniETH), Puffer (pufETH), KelpDAO (rsETH), Renzo (ezETH), and Swell (rswETH) have demonstrated the appeal of restaking by bootstrapping billions of deposits within a matter of weeks. Of the breadth of protocols, they mainly differ along the vectors of native vs LST restaking, withdrawal enablement, signer protection, and reward mechanism.
EtherFi, is a native restaking protocol, with DVT implemented and withdrawals enabled, which are features aimed at optimizing user experience. Puffer offers native restaking and is unique in its usage of the secure signing module. Bedrock, launched by professional node operators, offers native restaking, allows withdrawals and differs in its non-rebasing reward mechanism. Rio, Kelp, and Renzo are also native restaking protocols with their own network of validators and EigenPods.
EigenLayer altogether attracted over $6B in TVL and some of the strongest teams to build on its AVSs, pointing to promising developments in a Cambrian explosion of decentralized non-Nakamoto chains.
In Cosmos, replicated security distributes 25% back to ATOM holders for the security it provides. While the overall yield is similarly hard to estimate given how early this space is, ATOM holders would slowly accumulate consumer zone tokens and accrue governance voting power.
In Solana and Bitcoin, restaking is still at an early stage, and we are excited to see these trends develope and mature.
Slashing puts a twist on an otherwise seemingly straightforward restaking platform. With slashing, sometimes neutrality can be at risk. As an example, a validator could ransom attack its stakers, “Hey if you don’t pay me ETH, I will cause a slashing event”. Game theoretically, the staker would pay up to the full ransom amount. Of course, there are solutions to mitigate this, which we’ll come back to.
While EigenLayer’s current slashing mechanism is not finalized and may be delegated to individual AVSs, its withdrawal mechanisms are and act as a layer of defense. Essentially what would need to happen is when a validator commits a slashable offense, its ETH collaboral on the Beacon Chain would need to be decreased accordingly. By design, validator withdrawal credentials point at EigenPods, so that EigenLayer can complete the accounting including any penalties before a validator can withdraw its funds.
Cosmos Replicated Security has its own set of slashing rules and perhaps sets a benchmark for the restaking vertical. Downtime is punishable by 10 minutes of removal from the validator set and 0.01% of the validator’s fund. A slap on the wrist. Slashing is punishable by tombstoning, a permanent removal of the validator, and 5% of the funds. To prevent impact on Cosmos Hub by faulty or compromised consumer chains, a slash throttle is in place so that not too many Cosmos validators can be slashed at once. This is important because a sudden drop in the staked amount could open up the network to 51% attacks if the staked amount drops below TVL.
On EigenLayer, withdrawal credentials for native ETH staking are held by the EigenPod. So that when a withdrawal event occurs, Eigenlayer can finish all of its accounting, net of yields and penalties, before the validator can withdraw its staked ETH. There is a withdrawal period of 7 days. However, with LST restaking, the mechanism is different and potentially more fluid.
Cosmos has a variety of restaking mechanisms. In Stride, withdrawal is instant. In Persistence and Quicksilver, unbonding is dependent on the native chain. In Cosmos Hub replicated security, ATOM holders will govern the inclusion of consumer zones. If ATOM holders wish to withdraw from a consumer chain, they could unstake their ATOM which requires a 21-day cool-down period. If Cosmos Hub wishes to stop support for a consumer chain, a two-week voting period requiring a 40% quorum will need to take place.
With heavy competition amongst liquid re/staking protocols, composability is a key feature that LST and LRT protocols lean on to gain market share. DeFi users prefer highly composable LSTs, such as Lido’s stETH, which affords deeper liquidity. These LSTs/LRTs can be deployed across DeFi protocols like DEXes and lending platforms, to boost returns significantly. Therefore, competitors challenge the dominance of stETH in DeFi by focusing their efforts on BD to build out a strong DeFi ecosystem for their LST/LRT.
The chart below shows the effect of additional yield generated from utilizing LST in various DeFi strategies; however, as more protocols are utilized to increase yield, the associated risks also increase.
As shown in the above example, it is possible to increase the capital efficiency of stETH by almost 2x through Curve and AAVE. Capital efficiencies of >2x can be achieved by further stacking yields using a protocol like Sturdy Finance to collateralize on a DEX-like curve. It is possible to get even higher yields on LSTs in DeFi protocols on popular L2s like Optimism and Arbitrum due to the higher trading activity and incentive rewards on L2.
By composing LSTfi/LRTfi across DeFi protocols, the user also faces the following types of risk depending on the protocols used:
Slashing risk — Potential slashing risk affects all LST/LRT protocols, although the risk is very low as discussed in Part I and node operators are mostly professional. However, loss of capital due to slashing can happen. For example, Lido has had multiple slashing events with some of its operators.
Smart contract risk — Utilizing LSTs/LRTs across multiple DeFi protocols compounds the smart contract risk since the user is exposed to additional vectors for exploits. As examples, the Curve finance exploit caused many synthetic ETH pools to be drained and the Sturdy finance exploit was attributed to price manipulation of LP tokens to drain the protocol, leading to bad debt.
Depeg risk — Even though Beacon Chain withdrawals greatly mitigated depeg risks, the underlying LST/LRT token contract could still be exploited. A depeg can lead to mass liquidations. After stETH depegged following the Luna and Celsius collapse in May 2022, cascading liquidations in leveraged stETH positions caused the price of ETH to plummet. Depeg of LST tokens can amplify losses for LPers due to impermanent loss and on concentrated liquidity AMMs
Restaking risk — Restaking allows validators who run actively validated services to be slashed if they operate incorrectly or act maliciously. The greater risk of slashing depends on how many services are run per validator. Unintended slashing caused by bugs could lead to cascading slashing across multiple services and validators
There has been an explosion of DeFi protocols tapping into massive pools of liquidity from LSTs/LRTs to unlock new use cases, such as synthetic stablecoins, yield optimization, and yield tokenization.
With stablecoins being the most popular asset for web3 users seeking stability, many protocols mint decentralized stablecoins from LST collateral.
Lybra Finance is similar to a CDP protocol like MakerDAO. Lybra’s high collateral ratio of 150% enables it to distribute the staking rewards from the overcollateralized LSTs to all holders of the stablecoin and offer about 25% more yield than the base LST yield (eg. Lybra’s eUSD offered ~7% apr when the ETH LST yield was5%). On the other hand, Prisma Finance works with DEXes like Curve and Convex Finance to reward users for staking its stablecoin in DEX LPs.
However, these protocols are not very capital efficient because of the high collateral ratio and lack of integrations with other DeFi protocols for now due to their early nature. By using perps to hedge the price of the LST collateral for its stablecoin, Ethena offers a more innovative approach that could be the key to becoming a top decentralized stablecoin. Their approach provides higher capital efficiency, reduced liquidation risk, and could even challenge the dominance of centralized fiat-backed stablecoins.
Increased yield for LSTs/LRTs is key to attracting TVL. Synthetic ETH and tokenized LST indices are the main forms of yield optimization we are seeing in the market.
Synthetic ETH protocols take in LSTs or a basket of LSTs as collateral to mint a synthetic version of ETH. Protocols like Alchemix and Zeroliquid offer up to 50% LTV to borrow synth ETH and use the yield on LSTs to pay down the debt on self-repaying loans. Cat-in-a-box allows users to repay loans with the yield or generate additional yield depending on the LTV (higher health gives boosted yield). The main use case for Synthetic ETH is to LP them in DEXes for additional trading fee income. It also gives an arbitrage opportunity for users on the peg of the assets. However, it should be noted, that synthetic ETHs minted from these protocols are often volatile and trade below ETH peg which can lead to high impermanent loss for LPers
For users who want to avoid the hassle of using multiple DeFi protocols, LST index products like Yearn’s yETH and unshETH are simple and relatively safe ways to boost yield. LST indexes are a basket of LSTs pooled together into an AMM to facilitate easier deposits/withdrawals and rebalancing. The yield optimization is achieved by setting the weights of the LSTs in the index, with higher-yielding LSTs like sfrxETH and swETH representing a larger part of the index. In addition to LST incentives, users also receive yield from swap fees within the AMM as people withdraw/deposit LSTs.
Interest rate swaps in TradFi are used to lock in and hedge against volatile interest rates for borrowers. Protocols like Pendle and Flashstake enable similar functionality on LSTs/LRTs in web3.
Pendle allows DeFi users to split yield-generating assets like LSTs into two components, the principle token which can be redeemed 1:1 for the underlying asset upon maturity and the yield token which receives the yield of the underlying asset. This allows users to perform various strategies on LSTs such as locking in fixed yield on LSTs for stability, arbitrage on the yield and principle tokens for profit, or generating more yield by providing liquidity to pools of principle and yield tokens which have minimum impermanent loss due to their price correlation.
Flashstake, another yield tokenization protocol, allows users to deposit yield-bearing LSTs into the protocol to mint time-based tokens (TBTs). These TBTs are a special ERC-20 token that represents the total yield of the assets deposited in a vault. The most common use case for LSTs with Flashstake is to instantly redeem the TBTs from staking an LST to collect the upfront yield from LSTs.
Preserving protocol integrity, one line of code at a time
Composability of LSTs/LRTs is a centralizing force. The more protocols that offer composability to certain LSTs, the more demand for those LSTs; it’s a continuous feedback loop that concentrates liquidity. To make matters worse, solo stakers may at one point find their yield dip below hardware costs. If solo stakers withdraw from the protocol, decentralization suffers.
For restaking, risks around externalities exist. If a protocol with restaking causes a massive slashing event, Ethereum staked amount could fall drastically thereby opening it up for attacks. Vitalik has expressed that Ethereum would not fork to bail out large restaked AVSs.
Luckily, projects are making strides toward preserving decentralization and upholding neutrality. EigenLayer hosts governance votes to include more LSTs into its protocol, thereby increasing decentralization. Lido is adding permissionlessness to its protocol by introducing a staking router to allocate stakes across a variety of validator sets. LRTs pools are exploring the usage of DVT to minimize externalities introduced by restaking. These are directions we are delighted to see.
For now, risks around centralization and externalities exist and must be mitigated. There are several potential solutions that we are excited about:
Execution Tickets to Counter Centralization of Validators — Beacon validators simply proposes inclusion lists and validate such lists, while execution validators can run beefier machines. In this architecture, consensus clients can remain light and decentralized.
Cap Liquid Staking Using EIP4788, ZK-bridges, or Engine API — whereas the current approach to cap staking pools below the Sybil threshold is community-driven, there’s an opportunity to leverage ZK-bridges or Engine API to trustlessly transfer data from the execution layer back to the consensus layer to keep staking pools below Sybil threshold.
Simplify Developer Tooling for Solo Stakers — A DEPIN-style project could be interesting, where a provider sells prepackaged hardware sets with preinstalls of client software to make onboarding new solo stakers as seamless as possible.
Reputation System for Node operators — node operators could be rated on vectors of neutrality and code integrity. This is minimally effective as mass slashings are black swan events and reputation systems are not very effective against black swan events.
Enshrined Smoothing Pool for Validators — we could encourage more solo staking by creating a smoothing pool that generates higher consistent yield on par with quasi-decentralized staking pools. Ethereum Foundation perceive smoothing pool as attack surface. However, third parties could build one as part of the PBS architecture.
AVS and LRT Slashing Throttle— learning from Cosmos, we’d like to see AVS operators and LRT protocols implement real-time slashing throttles to prevent mass slashing events from spilling over to Ethereum. Further, LRT protocols could prioritize AVSs with dual staking and cascade the slashing of AVS tokens first and ETH second.
Enshrinement of LST: the composability of enshrined LST could massively benefit solo stakers to enjoy the economic bandwidth that current LST holders enjoy. On smart contract risk, already the deposits for staking on the Beacon Chain sits on a single smart contract in the Execution Layer, so an additional smart contract that mints LSTs doesn’t introduce much additional risk. However, this does add a large burden on EF for maintenance with an already busy development roadmap.
Case in point on Ethereum Foundation’s keenness to democratize node operations: Portal Network pushes the frontier to modularize and decentralize node operations across three RPC networks: Beacon light client, State network, and History network, so that developers can permissionlessly ping the network for their particular RPC use cases, all in a decentralized fashion.
If you are actively solving for decentralization and neutrality, we’d love to chat with you.
Initiatives such as the Portal Network, MEV burn, and LST enshrinement render confidence, that despite the various challenges in staking’s centralizing tendencies, we’d find a path forward towards sustained decentralization, permissionless, and neutrality.
Staking is the foundation of PoS chains. Decentralization and neutrality must be upheld by virtue of the incentives and disincentives around staking, restaking, and LRTfi.
Staking is the crypto-native benchmark rate. The advent of LSTs/LRTfi opened up composibility based on re/staking yield. Through LSTfi/LRTfi, we get a glimpse of what composability could mean for traditional finance. LSTfi/LRTfi is a hallmark of the benefits of web3 — frictionless, low-minimum, self-custody. These are benefits that could bring tremendous efficiency and open up participation in traditional finance.
Through first principles thinking, we find that various LSTfi/LRTfi strategies would be evaluated along the vectors of composability and capital efficiency, where risk is a boundary condition.
Looking a few years ahead, the Primary Layer around staking is maturing in Solana and Polygon, while Ethereum, Bitcoin and Cosmos staking are evolving. In Ethereum, the end game could look one of two ways: oligopoly if Ethereum values are upheld, where top players reach as close as to 33% without breaching it, or, enshrinement of LSTs if Ethereum values are not upheld. The Secondary Layer of restaking is igniting a race to the top of capital flowing toward the highest yield, especially to LRT protocols. EigenLayer AVSs are some of the most anticipated launches this year, and Bitcoin, Cosmos, and Solana retaking are all at the Cambrian stage. At the Tertiary Layer, capital efficiency is bound to be the highest. Protocols would compete along the vector of highest composability and lowest risk.
An economic stack built on re/staking is all wonderful and great. However, composability and capital efficiency are centralizing forces that could threaten the core values of Ethereum.
Akin to the tragedy of the commons, individual actors would maximize their composability and capital efficiency, while diminishing the common goods of decentralization and neutrality. However, there are proactive solutions that could be built to mitigate centralization and externalities.
Alignment with decentralization and neutrality must be the guiding principle and boundary condition within which the staking industry explores its powerful composability and economic prosperity.
(end of Article)
Sources: