Restaking yield is the composite return on capital that is staked and then committed again as collateral for other services. It is a stack, not a rate: base staking rewards at the bottom, service fees above that, and token incentives or points above those. Each layer has a different funding source and a different lifespan, and only the bottom layer keeps paying once incentive budgets end. A single blended figure hides the one distinction that predicts whether the number survives.
Restaking does not raise the base staking rate. It adds layers on top of it, and the additional return is the price of accepting additional ways to lose the same principal rather than a second income on a second position.
Scroll to see the full diagram
The rate is a stack and the layers are funded differently
The bottom layer is unchanged by restaking. Base consensus rewards and execution-layer income keep accruing exactly as they would for an ordinary staker, and restaking adds nothing to them. Consensys puts the actual trade plainly: participants earn extra yield by validating selected services in return for accepting increased slashing risk.2 That is a risk premium, and it should be quoted as one.
The second layer is service revenue. EigenLayer's rewards documentation describes AVSs submitting rewards to operators and stakers, with the newest mechanism distributing on allocated unique stake or on total delegated stake within an operator set, and supporting both retroactive and forward commitments up to two years.1 Where that money comes from is the question that decides whether the layer is durable: fees paid by users of the service, or tokens the service issued.
The layers above that are the ones a founder should be most careful describing. Across the restaking documentation we have reviewed, the largest advertised components are typically protocol token emissions and points programs, both of which are transfers from existing holders rather than revenue. We do not publish a component breakdown here because we could not resolve one to a live primary source, and an unsourced decomposition in this position would be worse than none.
What is taken before a restaker sees it
The operator takes a cut, and the size is a parameter rather than a convention. EigenLayer lets operators set their fee on AVS rewards anywhere from 0 to 100 percent, with a default split of 10 percent, and permits that split to vary per AVS and per operator set so fees can be aligned to the complexity of each service.1
Read the range rather than the default. A restaker delegating a whole balance to one operator has accepted whatever that operator sets, across every service that operator opts into, and the delegation is all or nothing. The restaking entry covers why that concentrates a two-step principal-agent problem into a single decision.
There is no protocol-level cap in that documentation, which means the operator's published split is a commercial term rather than a protocol constraint. It belongs in the diligence list alongside the client stack, not in a footnote under the rate.
The claim mechanics are part of the rate
Rewards are calculated off-chain and consolidated into a Merkle distribution root posted on-chain weekly on mainnet, with a further one week delay after posting before that root can be claimed against.1 Roots are cumulative, so claiming against the most recent one collects everything outstanding, which at least means a missed week is not a lost week.
One line matters more than the cadence. Rewards stop accumulating when a withdrawal is queued.1 Combine that with the protocol's safety delays and the effective rate on capital that is exiting is not the advertised rate. Any product quoting an annualised figure without stating what happens to accrual during the exit window is quoting a number that does not apply to the period a depositor most cares about.
In our view that single sentence should appear in every restaking product's rate disclosure, and it currently appears in almost none of them.
Sizing the service layer without inventing a number
The arithmetic is simple and the inputs are the hard part. Annualised fee revenue paid by services, divided by the restaked capital securing them, gives the service layer's contribution before the operator's cut. The denominator is observable: DefiLlama's public data API reported EigenLayer, now recorded under the name EigenCloud, at $4,952,961,944 on 3 August 2026 at 13:48 UTC, inside a Restaking category totalling roughly $7.84 billion.3 The numerator is the one to insist on, and it is the one most rate pages omit.
Do that division before accepting any composite figure, and the shape of the answer tends to settle the argument on its own. If the service layer is small relative to the advertised rate, the difference is being paid by token holders, and that is a finding about the business model rather than about the yield.
The risk side scales with the yield side by construction. EigenLayer states that delegated stake becomes slashable once an operator opts into an operator set and allocates unique stake, and makes stakers responsible for confirming their own risk tolerance and monitoring operator allocations as they change.4 Every service that adds a basis point adds a slashing surface. Both appear in the same decision and only one appears on the marketing page. This is reference material for design work, not investment advice, and nothing here is a yield expectation for any asset.
Common questions
Where does restaking yield actually come from?
Four sources with different funding. Base staking rewards from issuance and transaction fees, service fees paid by AVSs, protocol token emissions, and points programs. Only the first two are revenue; the others are transfers from existing token holders. Consensys frames the additional return as the price of accepting increased slashing risk rather than as a second income stream.2 Ask which layer funds which basis point before comparing rates.
How much do restaking operators take as a fee?
Whatever they set. EigenLayer allows operators to set their fee on AVS rewards anywhere from 0 to 100 percent, with a default split of 10 percent, and the split can vary per AVS and per operator set.1 There is no protocol cap in that documentation, so the operator's published split is a commercial term. Since a restaker delegates their entire balance to one operator, it applies across every service that operator joins.
Does restaking increase the base staking rate?
No. Base consensus rewards and execution-layer income accrue exactly as they would for an ordinary staker, and restaking adds separate layers on top rather than raising the floor. What restaking changes is the number of slashing conditions the same capital answers to, since delegated stake becomes slashable once the operator opts into an operator set and allocates unique stake.4
See LRT Tokenomics Guide for how this applies in practice.
Sources
- Rewards Overview
EigenCloud, Eigen Labs, 2026
AVS reward submissions to operators and stakers, the operator split settable from 0 to 100 percent with a 10 percent default and per-operator-set variation, unique stake and total stake weighting with commitments up to two years, weekly mainnet Merkle roots with a one week claim delay, and the statement that rewards stop accumulating when a withdrawal is queued. Cloudflare returns 403 to automated clients; content read by stealth fetch on 3 August 2026. - EigenLayer: Decentralized Ethereum Restaking Protocol Explained
Consensys, 2024
The framing of extra yield earned by validating selected services in return for accepting increased slashing risk. - EigenLayer protocol TVL, public data API
DefiLlama, 2026
Total value locked of $4,952,961,944 read on 3 August 2026 at 13:48 UTC, category Restaking, current entity name recorded as EigenCloud. The Restaking category across all tracked protocols totalled roughly $7.84 billion on the same read. Figures update continuously. - Slashable Stake Risks
EigenCloud, Eigen Labs, 2026
Delegated stake becomes slashable once the operator opts into an operator set and allocates unique stake, and stakers are made responsible for confirming their risk tolerance and monitoring operator allocations on an ongoing basis. Read by stealth fetch on 3 August 2026.
Last reviewed 2026-08
Know the terms but not sure how they apply to your project? That is what an engagement is for. We design, document, and stress-test the whole token economy inside the Tokenomics Data Room.
100+ projects advised. Complete tokenomics in 4 to 6 weeks.