Case Study 1: Lido's 33% Problem — When Liquid Staking Threatens Decentralization

Background

In the summer of 2023, the Ethereum community confronted a scenario that many researchers had warned about but few expected to materialize so quickly: a single liquid staking protocol was approaching the threshold at which it could theoretically prevent the network from achieving finality.

Lido Finance, launched in December 2020 just days after the Beacon Chain's genesis, had grown from a niche solution for illiquid staked ETH into the dominant force in Ethereum's staking ecosystem. By May 2023, Lido controlled approximately 31.8% of all staked ETH — uncomfortably close to the 33.3% threshold that Byzantine Fault Tolerance theory identifies as critical. At 33%, a single entity can prevent finality. At 50%, it can censor transactions. At 66%, it can unilaterally finalize blocks. Lido was approaching the first of these thresholds, and its growth trajectory suggested it would not stop there.

The situation raised fundamental questions about the tension between capital efficiency and decentralization — questions that remain unresolved and that illuminate the deepest challenges of Proof of Stake governance.

How Lido Grew

Lido's growth was not accidental. It was the product of rational economic behavior by ETH holders facing a genuine problem.

Before the Shanghai/Capella upgrade in April 2023, staking on Ethereum was a one-way commitment. Depositing 32 ETH meant locking it indefinitely, with no guarantee of when withdrawals would be enabled. For large ETH holders — including DeFi protocols, treasuries, and institutional investors — this illiquidity was unacceptable. Lido solved the problem by accepting ETH deposits, staking them through a curated set of professional node operators, and issuing stETH — a liquid token that tracked the value of staked ETH plus accumulated rewards.

stETH quickly became one of the most deeply integrated tokens in DeFi. It was accepted as collateral on Aave and MakerDAO, traded in deep pools on Curve Finance, and used in yield strategies across dozens of protocols. This DeFi composability created a flywheel: the more DeFi integrations stETH had, the more useful it was, the more people deposited with Lido, and the more DeFi protocols integrated stETH.

By the time withdrawals were enabled in April 2023, Lido's dominance was entrenched. Users who had deposited with Lido during the lockup period largely stayed — the convenience of liquid staking and the DeFi yield on top of staking yield made solo staking or even other liquid staking providers less attractive.

The 33% Threshold: What It Actually Means

To understand why 33% is critical, recall how Casper FFG finality works. A checkpoint is justified when two-thirds of the total stake (by weight) attests to it. If a single entity controls 33% of the stake, it can prevent any checkpoint from reaching the two-thirds threshold simply by withholding its attestations.

In concrete terms, this means:

• Finality would stop. No new checkpoints would be finalized. The chain would continue producing blocks (LMD-GHOST still operates), but those blocks would lack the strong irreversibility guarantee of finalization.

• The inactivity leak would activate. After four epochs without finality (~25.6 minutes), the protocol would begin penalizing all non-attesting validators. If Lido's validators were the ones withholding, they would face quadratically increasing penalties.

• Economic destruction would follow. The inactivity leak is designed to eventually restore finality by draining the stakes of non-participating validators. In the scenario of a deliberate Lido attack, billions of dollars in staked ETH would be progressively destroyed.

Lido's defenders argued that this scenario was implausible for several reasons. Lido's validator operations were distributed across approximately 30 independent node operators, each running their own infrastructure. These operators were bound by contracts and reputation — they would not simply follow malicious instructions from Lido's governance. Furthermore, an attack would destroy the value of the LDO governance token and the stETH derivative, making it economically irrational.

Critics pointed out that the node operator set was selected by LDO governance, which itself was highly concentrated. A small number of LDO holders (including the Lido DAO's founding team and early investors) held significant voting power. More importantly, the risk was not just deliberate attack — it was also regulatory coercion, smart contract vulnerability, or coordinated node operator failure.

The Community Response

The debate came to a head in June 2023, when the Ethereum research community published a series of analyses and proposals.

Danny Ryan, an Ethereum Foundation researcher and one of the architects of The Merge, published a widely cited post titled "The Risks of LSD" (Liquid Staking Derivatives), arguing that any staking protocol exceeding certain thresholds posed an existential risk to Ethereum. He called for self-limitation — a voluntary cap on Lido's market share.

Superphiz, a prominent Ethereum community voice, proposed "social slashing" — the idea that the Ethereum community should socially coordinate to penalize protocols that exceeded decentralization thresholds, even without protocol-level enforcement.

Lido's governance debated a self-limiting proposal internally. In June 2023, a snapshot vote was held asking whether Lido should self-limit to a maximum percentage of staked ETH. The proposal was defeated, with approximately 99.8% of LDO voting against self-limitation. This result was widely criticized as evidence that economic self-interest would always override decentralization concerns in a token-governed protocol.

The Ethereum Foundation explored protocol-level solutions, including:

1. Capping the benefits of scale: Proposals to reduce attestation rewards for validators associated with entities above certain thresholds.
2. Increasing solo staker incentives: Ideas for rewarding validators that could prove they were independently operated (difficult to verify in practice).
3. Distributed Validator Technology (DVT): Supporting protocols like SSV Network and Obol that enable multiple independent operators to share responsibility for a single validator, reducing the operational advantage of large staking providers.

The Aftermath

Lido's market share peaked at approximately 32.4% in late 2023 and has since moderated, settling in the 28-30% range by early 2025. Several factors contributed to this partial decline:

• Withdrawal availability: Once withdrawals were enabled, some users who had used Lido purely for liquidity switched to solo staking or other options.
• Competition: New liquid staking protocols (including Coinbase's cbETH and smaller entrants) captured some marginal growth.
• Community pressure: Social campaigns encouraging staking diversity had a modest but measurable effect.
• DVT maturity: Distributed Validator Technology began enabling smaller operators to compete with Lido's operational efficiency.

However, the fundamental dynamics that drove Lido's growth remain in place. stETH's DeFi integration is still unmatched. The convenience of liquid staking still outweighs the principles of decentralization for most ETH holders. And no protocol-level mechanism has been implemented to cap liquid staking dominance.

The Deeper Lesson

The Lido episode illustrates a tension at the heart of Proof of Stake that has no easy resolution.

PoS was designed to be more accessible and decentralized than PoW — and at the individual validator level, it is. Anyone with 32 ETH and a consumer computer can validate. But the economic dynamics of liquid staking create powerful centralizing forces. Capital efficiency (the ability to earn staking yield while simultaneously using stETH in DeFi) is a genuine value proposition, and the market rewards the most liquid and best-integrated staking derivative.

This is a pattern seen across financial markets: the efficiency gains of intermediation create natural monopolies. Just as traditional finance saw the emergence of dominant clearinghouses, custodians, and index fund providers, Ethereum's staking ecosystem is gravitating toward dominant intermediaries. The question is whether the blockchain's governance and protocol mechanisms are sufficient to constrain these dynamics in ways that traditional regulation constrains monopolies in legacy finance.

As of this writing, the answer is: not yet. The community has awareness of the problem, a shared vocabulary for discussing it, and several promising research directions. But no binding solution has been implemented. Lido remains the largest single staking entity on Ethereum, and the next protocol to gain traction could face the same centralizing dynamics.

Discussion Questions

1. Self-governance failure: The LDO vote against self-limitation passed with 99.8% against. Does this demonstrate a fundamental flaw in token governance — that economic self-interest will always override public goods considerations? Or is there a governance design that could produce a different outcome?

2. Protocol-level caps: Should the Ethereum protocol enforce a maximum percentage of stake that can be controlled through a single liquid staking smart contract? What are the implementation challenges? Could such a cap be circumvented?

3. Regulatory analogy: Traditional financial markets have antitrust laws and concentration limits for custodians. Are these relevant models for decentralized staking, or does the open-source, permissionless nature of blockchain make them inapplicable?

4. The DVT path: Distributed Validator Technology allows multiple independent operators to share a single validator. Could DVT fundamentally change the liquid staking centralization dynamic? What challenges does it face in achieving adoption?

5. The stETH systemic risk: stETH is used as collateral across dozens of DeFi protocols. If a Lido failure (smart contract hack, mass slashing, governance attack) caused stETH to depeg from ETH, what would the cascading effects be? How does this systemic risk compare to traditional finance's "too big to fail" problem?