Restaking: Risk Stacked on Risk

Restaking is the practice of pledging an asset that already secures one network as collateral for additional services at the same time. The pitch is simple: the capital is already locked, so why not let it do two jobs at once? That framing is also where the trouble starts. Restaking does not duplicate capital — it duplicates the conditions under which that capital can be taken away.

Most explanations stop at the reward side. This one starts with the obligation side: what you promised, to whom, and what happens when two of those promises come due in the same week.

• Restaking adds obligations, not capital. The same deposit now answers to several slashing conditions at once, and each one can act independently.
• Correlation is the real multiplier. Positions that share an operator, client software or oracle feed tend to fail together, so “diversified across ten services” is often one bet in ten costumes.
• Headline restaking yield is gross. It ignores slashing probability, unbonding queues, the cost of hedging, and the discounts that appear when everyone exits at once.
• The exit path matters more than the rate. If you cannot leave within the window you might need the money, the extra yield is compensation for illiquidity you did not intend to sell.

How restaking turns one deposit into several obligations

A restaking position is a stack of contracts, and each layer can impose its own penalty. Reading the stack from the bottom explains why the marginal yield looks attractive and why the marginal risk is larger than it appears.

The first layer: collateral already at work

The base deposit is staked to a proof-of-stake network under that network’s rules: a bonding period, a slashing condition for provable faults such as double signing, and a reward schedule set by issuance and fees. This layer is already a trade — you accept lockup and fault risk in exchange for issuance, and restaking never removes it.

The important detail is that the base layer usually has the longest unbonding period. When people talk about a restaking position being “liquid”, they are often describing a receipt token that trades at a discount precisely when the underlying queue is longest.

The second layer: operators and the services they run

On top of the base deposit, a restaking arrangement lets an operator commit the same stake to additional services — commonly called actively validated services, or AVSs. Each service defines its own duties and its own penalty schedule. Some require the operator to run specific software, some require price or state attestations, some require uptime commitments.

You are not buying ten independent bets. You are buying one operator’s operational competence, split across ten contracts that each hold a claim on the same collateral. If the operator’s infrastructure fails at 03:00, several services may record a fault from the same root cause.

Where restaking risk actually compounds

Restaking risk compounds at the points where two layers share a dependency. The table below maps the layers against what a single failure can reach.

Read the third row carefully. Operator risk is the only row where a single mistake reaches every column at once, and it is the row most often left out of yield comparisons.

Restaking does not pay you for taking more risk in the abstract. It pays you for taking a specific, named risk that the service needs someone to hold — and the fee is usually priced against the cheapest operator willing to hold it.

Working through a restaking loss

Numbers make the stacking visible. Suppose you deposit 100 units of a staking asset. The base network pays 3.0% a year. You add ten services, each paying an extra 0.2%, for a gross 5.0% — 5 units a year instead of 3.

Now a fault occurs. The service that penalises most severely specifies a 20% penalty on the stake allocated to it. Because you allocated evenly, that service holds 10 units, so the penalty is 10 × 20% = 2 units. Against 2 units of extra annual reward from that service, you have lost a decade of its contribution in a single event.

The worse case is the correlated one. Suppose the fault originates in shared operator infrastructure and three services record it. Exposed stake becomes 30 units, the penalty becomes 30 × 20% = 6 units, and the year nets 5 − 6 = −1 unit before you count the base layer’s own risk. The position has produced a negative return while every published rate on the screen stayed positive.

1. Write down the gross reward from each service you are considering, as an annual number in units of the asset.
2. Write down the maximum penalty each service can apply, and whether penalties are capped.
3. Group services by operator, client software, oracle dependency and data source.
4. Assume the largest group fails together, and sum the penalties across that group.
5. Compare that sum with the total extra reward over your intended holding period.
6. Add the exit cost: queue length, receipt-token discount, and the price impact of unwinding any hedge.
7. If the sum of penalties is larger than several years of extra reward, you are not being paid for the tail — you are donating it.

Questions that decide whether restaking is appropriate

Three questions separate a considered restaking position from a yield-chasing one. None of them appear on a leaderboard.

Who controls the operator set

If a small number of operators serve most of the stake, the network’s fault tolerance is lower than the validator count suggests. Check whether delegation is permissionless, whether operators can opt into services without depositor consent, and whether the service can add new penalty conditions after you deposit.

Consent that can be changed unilaterally after the fact is not consent. In many designs, the depositor approves a service once and inherits every future parameter change.

What the exit path costs

Exit cost has three parts: the protocol unbonding period, the service unbonding period, and the market price of any liquid receipt. In calm markets the third is close to par. In stressed markets it is not, and the discount is the market’s estimate of exactly the risk described above.

Compare the exit path with the reason you hold the asset. If the position exists to be sold during a drawdown, a two-week queue converts a hedge into a liability.