Concerns on the Future of Staking

Published in

CoinShares Research Blog

8 min readSep 25, 2023

Introduction

There has been a lot of discussion regarding Ethereum Staking and the potential future implications on the network as a whole at different staking levels (as a percentage of total ETH supply). At the moment, ~22%¹ (27M ETH) of the current ETH Supply is being staked by validators, and the queue for people looking to join stands at roughly 10 days (from a peak of ~45 days in June 2023). Although the data for time stamped queue length doesn’t go back far enough to visualise this, it should be noted that in the days leading up to the Shanghai upgrade (which enabled staking withdrawals), the Exit Queue was briefly larger than the Entry Queue (17 vs 8 days) (Apr 2023).²

People looking to stake their ETH have a multitude of options, boiling down into 2 main categories: They can solo stake their ETH themselves (at the cost of setting aside 32ETH), or turn to liquid staking providers, which provide mechanisms for people to stake in a less capital intensive way, often times with no minimum ETH amount (read more on liquid staking here).

Of all staking providers, Lido has by far the largest market share (~32%), with the next most popular being Coinbase, with roughly 8%.

Centralisation Risk and What it Means for Security

Though the queue is full (albeit falling rapidly), the amount of ETH staked has been rising consistently, and Lido has shown no signs of slowing its dominance over other providers. Some have pointed to the fact that Lido is about to cross over into controlling 1/3rd of all staked ETH, which has raised some alarm bells in terms of centralisation risk.

Firstly, it does not bode well for a decentralised blockchain to have such a large concentration of validators under one provider, as ideally, block production should happen in a decentralised manner. Further to this, should Lido succumb to a governance attack, and a malicious actor be in control of the Lido validator nodes, >33% is the necessary amount of staked ETH as a percentage of the total staked supply in order to prevent the chain from finalising (finality delay). Through malicious attestation or lack thereof, an attacker could cause serious issues for the network as a whole, like double finality, where, after being chosen as a block producer, the attacker could equivocate (generating two or more conflicting proposals), vote twice with its >33% control, and cause the chain to fork into two separate blockchains, where the implications would be severe.

The risk of Lido being attacked is low (operator nodes are spread between ~30 companies), but should not be taken lightly. There has been a lot of discussion from the Ethereum community, that something should be done to avoid any possibility of a governance/other attack, but this has been met with radio silence on any official plans by Lido to get this under control. There has even been some pushback from Lido staff, who seemingly don’t believe this is a genuine threat. (Seraphim is the DeFi Expansionist at Lido Finance)

Another risk to the decentralisation of Ethereum is the “cartelization” of Lido, a term that has been thrown around discussion forums to describe activity that could further cement its monopoly status. By reaching critical consensus thresholds (like the 33% mentioned above), Lido Node Operators could in theory, participate in block-timing manipulation and coordinated MEV extraction, which would lead to outsized rewards (in relation to other providers)³, and thus a cartel-like dominance in terms of distributed rewards. This would mean that there would be a clear competitive advantage and users would have little to no incentive to stake elsewhere.

Part of Lidos dominance in market share comes from its first mover advantage (deep liquidity and long term proven stability) as well as its attractive APR % in comparison to other liquid staking providers. There is a reasonable expectation that as the segment matures and more competitors gain scale, there will be a race to the bottom in an attempt to grab market share. So far, there is little data to suggest this hypothesis, as new validators are still flocking to Lido heavily.

Dealing with Future Scenarios Around Liquid Staking

The next area for discussion is the growing number of validators in general and the implications that this may have on the performance and underlying characteristics of the network. Around 2 weeks ago, Dankrad Feist, Ethereum Core Developer, released a blog post in which he outlines his concerns on the future of liquid staking.

Although as previously mentioned, the validator queue is falling rapidly and expected to reach 0 over the next few weeks/months, there is a non-zero probability that the demand for staking will rise again and the queue could remain full for the foreseeable future. If the queue remains full, according to Dankrard, that would mean “50% of all ETH would be staked by May 2024, and 100% by December 2024”.

From a technical perspective, a growing number of validators causes an increasing strain on the peer-to-peer network, as gossiping between nodes increases proportionally with the number of active validators. If we reached a point where there were, let’s say, 2 million active validators, node operators may find themselves in a situation where they need to upgrade their hardware to accommodate this increase in message propagation. Generally speaking, if Ethereum wants to be a highly scalable network, that would not be the way forward.

A further issue that could arise if the staking queue remains full and we reach much higher levels of staked ETH is what Dankrad refers to as an “untested economic regime” — the staked ETH market cap becomes larger than the unstaked ETH market cap. This, again, has a very unlikely, yet non-zero probability of occurring, and so solutions to avoid this scenario should be explored. It is important to remember that there is an inverse, decaying relationship between the percentage of ETH staked and the % APR a validator receives.

It would be expected that as the number of validators grows, we reach a point in which validators would begin to exit as the opportunity cost of tying up capital for yields under 2% would not be worth it. One risk to this is the ease of entering and exiting a liquid staking provider — only solo stakers are truly tying up a significant amount of capital. Especially with the economies of scale that benefits providers like Lido, the associated hardware costs may not be a disincentive to stop staking as yield decreases.

In view of the multi-year high on benchmark Federal Funds Rates, there is an added dynamic to the opportunity cost of tying up capital in ETH staking. Given you can get anywhere from ~4.0–5.5% on Treasury Bills, it seems we are reaching some form of equilibrium in relation to staking demand (as evidenced by the queue rapidly falling). Should government backed yields fall (as indicated by the dot plot this could be some time next year)⁴, we could certainly see a new pickup in the demand for staking, although it should be stated that there is a contrast in risk profiles for the typical ETH staker and the typical T-bill investor.

The Solution (In the Short Term)

A long term solution is unclear — in the same vein that liquid staking dynamics were not fully understood when staking was introduced, any solution that will be included in a hard fork must consider what dynamics it could change further down the road.

A short term solution, however, is in the works. Dapplion proposed EIP-7514, which would modify the validator churn limit from an infinitely increasing amount to a hard capped maximum (of either 16, 12, 8, 6 or 4). Simply put, this reduces the amount of new validators that can join per epoch (6.4 minutes), and thus greatly increases the amount of time it would take to reach milestones, such as 50%, 75% or 100% of staked ETH. This would be assuming a scenario in which the activation queue remains full and the demand for staking becomes unaffected by the “self-balancing” decay relationship above.

As seen above⁵, modifying the churn limit to a hard capped figure slows down the growth of active validators significantly. If things stay as is, the projected churn limit will continue rising as the active validator set rises, which would speed up how fast we reach higher levels of staked ETH (as a % of total).

This solution does not address any of the underlying issues we have discussed, but gives the Ethereum developer community more than enough time to come up with a solution that actually addresses staking dynamics. Some of the potential long term solutions have been suggested by Dankrad and other Ethereum community members, such as:

MEV burn
Increasing validator maximum effective balance
Making staking rewards less appealing (by reducing issuance rewards)
Making the barrier for entry lower for liquid staking providers (to incentivize decentralisation)

All of these would fundamentally change one or more dynamics of the network as a whole, and could trigger large second order effects, so they need to be planned and considered carefully.

Overall, it seems unlikely that the queue will remain full, and that we will reach levels higher than those observed in other PoS blockchains of ~40–45%⁶ staked as a percentage of the total supply, which if reached would theoretically not have any of the potential negative impacts discussed above, apart from maybe in terms of scalability (an increase in gossip messages and the networks ability to accommodate this). In further posts, we will discuss proposed long term solutions in greater detail, as well as upcoming changes to Ethereum that will fundamentally impact the way the network operates.