Ethereum restaking sees reduced market value and risks

Staking ETH earns a yield because the staked capital is doing a job: it backs a promise to follow Ethereum’s consensus rules, and it can be destroyed if that promise is broken. Restaking asks an obvious follow-up question. If that capital is already locked up and already carries a credible threat of destruction, why can it only back one promise? Restaking protocols let a staker point the same collateral at additional services, accept those services’ penalty rules on top of Ethereum’s, and collect extra rewards for doing so. The idea is elegant, the mechanism is real, and the period since penalties became enforceable has been an unusually honest test of whether the economics hold up.

The problem restaking was built to solve

Any new piece of blockchain infrastructure that needs distributed validation faces the same cold-start problem. An oracle network, a bridge, a data-availability layer or a sequencer all needs a validator set with real money at stake before anyone will trust its output. Historically that meant launching a token and hoping its market capitalisation grew large enough that attacking the network cost more than it paid, which is slow, expensive, and produces security only as durable as the token price.

Restaking inverts the sequence. Instead of building a new security budget from scratch, a service rents access to an existing one. In EigenLayer’s vocabulary these services are actively validated services, or AVSs, and they buy externally attributable, slashable economic security without ever launching validator infrastructure of their own. The services now built this way span data availability, oracles, bridges, verification of AI outputs, off-chain compute and rollup sequencing.

How the mechanism actually works

There are two entry points. Native restaking targets operators of real Ethereum validators, who route their withdrawal credentials through a smart contract construct called an EigenPod. The ETH stays staked on the beacon chain and keeps earning consensus and execution rewards, but the restaking contracts gain a claim on it if an AVS penalty is triggered. The second route uses liquid staking tokens: a holder of stETH, rETH or cbETH deposits the token directly, trading the technical burden of running a validator for exposure to an additional protocol’s smart contracts.

Most restakers do neither job themselves. They delegate to operators, who run the infrastructure and, critically, choose which AVSs to support. That delegation is where the risk profile is actually set: a restaker inherits their operator’s uptime record, key management practices and AVS selection without necessarily inspecting any of it. A further layer sits on top in the form of liquid restaking tokens, which package the position into something tradable and reusable as DeFi collateral, adding depeg and liquidation exposure to everything underneath.

When the penalties became real

For the first phase of the category’s life, restaking was effectively a rewards program with a hypothetical downside. Slashing conditions were described in documentation but not enforceable on mainnet, so capital flooded in to farm points with no live mechanism capable of destroying it. Slashing went live on April 17, 2025, with operator sets and unique stake, meaning an operator’s allocation to one AVS can be penalised without automatically dragging in their commitments elsewhere. Redistributable slashing followed in July 2025, allowing slashed assets to be redirected rather than simply burned, which matters for services where a fault produces an identifiable victim who should be compensated.

The market repriced accordingly, and that is the single most important fact about restaking in 2026. EigenLayer’s total value locked peaked near $20 billion during the 2024 points era; data pulled from DefiLlama in July 2026 puts base-layer TVL at roughly $5 billion. Symbiotic, the main permissionless competitor, sits near $329 million on the same measure, Babylon holds around $3.3 billion in BTC-denominated terms, and Karak has collapsed from 2024 highs above $700 million to a residual figure. Yields compressed alongside. Base Ethereum staking currently returns something in the low three percent range, and AVS rewards add a modest increment rather than the double-digit numbers implied by early points speculation.

What researchers are actually worried about

The academic literature is more specific than the popular framing. A peer-reviewed empirical study of liquid restaking protocols by Terenzi and Ferretti, presented at the BRAINS conference in November 2025, found that many AVSs are secured by highly centralized validator sets, and argued that the combination of collateral reuse, concentrated delegation and layered smart contract architecture creates the conditions for risk amplification and correlated failure. The failure mode is not one service slashing one operator. It is many operators sharing infrastructure dependencies and therefore failing together, at which point pooled security turns out to have been considerably less pooled than advertised.

A separate paper by Sevim and Ferreira Torres, revised in August 2026, stress-tested the bridge exposure created when liquid restaking tokens expand across multiple chains and concluded that at current size it does not constitute systemic risk to the staking ecosystem, while flagging the growing interconnection as something requiring ongoing monitoring. Vitalik Buterin’s warning, published in May 2023 before any of this was live, drew the line differently: dual use of validator staked ETH is fundamentally fine, but attempting to recruit Ethereum’s social consensus to bail out an application’s failures is not.

The loss that actually happened

The largest restaking-related loss to date came from none of these mechanisms. On April 18, 2026, attackers compromised the single verifier node that KelpDAO’s LayerZero bridge relied on and forged a cross-chain message, minting roughly 116,500 rsETH out of nothing, worth about $292 million and near eighteen percent of the token’s circulating supply. The stolen tokens were deposited into Aave as collateral and borrowed against, leaving the lending protocol facing estimated losses between $123 million and $230 million depending on how the damage was allocated. No base restaking contract failed and nobody was slashed. The money left through a one-of-one configuration in a wrapper layer, precisely the class of risk the research literature keeps pointing at.

What it means for an ETH holder

For anyone holding ETH and deciding whether restaking is worth it, the honest summary is that the extra yield is now small, measurable and genuinely earned, while the extra risk is layered and difficult to price. Native restaking through an EigenPod adds the fewest new dependencies. Restaking a liquid staking token adds one protocol layer. Holding a liquid restaking token and then using it as collateral elsewhere adds three or four, and the April incident demonstrated that the outermost layer is where losses have actually occurred. The relevant question is not whether restaking works, because it demonstrably does, but whether the incremental percentage point or two justifies the number of separate things that must all keep functioning correctly. That calculation looks very different for a validator operator with in-house infrastructure than for a retail holder buying a wrapped token on a secondary chain, even though the advertised yield figure is the same.