Ethereum staking: 2.46% before anyone takes a cut

Ethereum staking is the clearest example in crypto of a yield that is real, protocol-funded and thoroughly unexciting. There is no borrower who might default and no marketing budget that might be cut. The network issues ETH to validators for attesting to blocks, and on 16 September 2026 that came to a 2.46% annual rate. Everything interesting about the product happens around that number: how long you wait to start, how long you wait to stop, and how much of the 2.46% survives the trip to your account.

This page deals with Ethereum specifically. The general mechanics of validators, commissions and unbonding live on our crypto staking pillar, the receipt-token model on liquid staking, and the cross-asset numbers onstaking rewards compared.

Where the 2.46% comes from

A validator earns from two separate places, and confusing them is the root of most bad estimates. Consensus-layer rewards are newly issued ETH paid by the protocol for timely attestations, sync-committee duty and block proposals. They are split by weight out of 64: timely source 14, timely target 26, timely head 14, sync committee 2 and proposer 8. A validator that never proposes a block and never sits on a sync committee collects about 54/64 of the base reward, which is the normal experience for most of the time.

The formula behind that base reward divides by the square root of the total active balance. The consequence is the most important structural fact about ETH staking economics:per-validator consensus issuance falls as more ETH is staked, and doubling total stake cuts it by roughly 29%. This, not any policy decision, is why the rate has compressed from its earlier levels as participation climbed to 35.31% of supply.

Execution-layer rewards are different in kind. Priority fees and MEV go to whoever proposes the block, are not issued by the protocol, and rise and fall with on-chain activity. MEV-Boost typically adds 10–30% on top of base validator rewards at 2026 parameters, which works out at roughly +0.28% to +0.83% of APR. KuCoin puts all-in returns for a well-run node at 3.3–3.8%. CoinShares argues that when on-chain demand is weak, execution-layer rewards can reach around 80% of total returns. Those two views bracket a wide and volatile range rather than settling on a number, and any calculator that quotes execution rewards to two decimal places is overstating what is knowable.

The queue is the part that actually costs you something

A new validator cannot start attesting the moment its deposit lands. Activations are rate limited by the churn limit — 256 ETH per epoch in each direction — and everyone waiting forms a queue. On 16 September 2026 that queue held 1,831,267 ETH, about 31 days and 19 hours. Capital sitting in the queue earns nothing.

The queue's behaviour through 2026 is the story of the year. It went from near zero to 3,114,842 ETH and 54 days on 24 January 2026, then to 3,589,414 ETH and 62 days by 20 May, and has since drained to today's 1.83M ETH. The exit side has sat at or near zero throughout, which is a complete inversion of mid-2025 when exits ran to nine-day waits. The drivers cited for the entry pressure are ETH treasury companies — BitMine alone has around 1.84M ETH staked out of roughly 4.2M held — Grayscale staking 73.57% of its ETH, and staking-enabled US exchange-traded products.

Once you do exit, there is a further wait: a roughly 27-hour withdrawability delay after leaving the exit queue, then the withdrawal sweep, currently running at about 7.9 days. Practically, the round trip today is a month to get in and about a week and a half to get out. That asymmetry is temporary and it has pointed the other way before, which is the argument for not treating the current queue as a permanent feature.

What the upgrades changed for a staker

Three upgrades in eighteen months have altered the mechanics of getting in, getting out and being punished. None of them changed the reward formula.

The most under-appreciated of these for retail readers is EIP-7002. Before it, exiting a delegated validator required the operator who held your signing keys to cooperate. Now the withdrawal-credential address can trigger the exit directly. It does not make a staking-as-a-service arrangement risk-free, but it removes a specific dependency that used to sit between you and your own capital.

Fee drag: the difference between a good and a poor route

The gap between the four routes is almost entirely fee. Starting from the same 2.46% gross network rate, here is what each one leaves, alongside what you actually have to commit.

Observed net APYs on the liquid staking tokens line up with that arithmetic: stETH at 2.22–2.30%, WBETH at 2.18–2.35%, rETH at 2.15–2.54%, cbETH at 2.47% and mETH at 2.05–2.73%. The 30 to 50 basis point spread across the majors is essentially pure fee differential rather than a difference in how well the underlying validators perform.

Slashing: rare, but not the risk you should weigh most

Slashing on Ethereum punishes equivocation only. Going offline is not slashable; it costs you rewards and a small penalty. Proposing two blocks at the same height, casting two conflicting attestations for the same target epoch, or a surround vote are.

Empirically it is rare. Fewer than 500 validators have been slashed out of more than 1.2 million active since the Beacon Chain launched in December 2020. The largest recent correlated event was on 10 September 2025, when 39 validators were slashed. The cause was operational, not adversarial: an Ankr maintenance action and a cluster migrated from Allnodes two months earlier where a leftover secondary signer produced duplicate signatures across SSV Network distributed-validator clusters. One 2,020-ETH validator lost about 0.3 ETH, roughly $1,300 at the time.

The penalty has three parts. An initial penalty, cut by Pectra from 1/32 to 1/4096 of effective balance. A correlation penalty applied at the roughly 18-day midpoint, calculated as the smaller of your balance and three times your balance multiplied by the total slashed effective balance in the surrounding 36-day window, divided by total active balance. And a forced exit with withdrawability around 36 days out. For an isolated incident the correlation term rounds to nothing; in a mass event it can approach your entire balance. That tail, not the base rate, is the thing to price.

Net issuance, and what 2.46% is worth in real terms

Nominal APR is not the return. Staking rewards are newly issued ETH, and if the supply grows faster than your balance you have lost ground in ownership terms. The correct adjustment is(1 + nominal) ÷ (1 + inflation) − 1, not simple subtraction.

Ethereum's net issuance was 0.88% as reported by Staking Rewards — net of the EIP-1559 burn, and capable of going negative in periods of high demand, which makes ETH the only major asset in our cross-asset comparison whose inflation rate is demand-dependent. Running the formula: 1.0246 divided by 1.0088 gives a real yield of+1.57%. Using the higher 2.63% cross-check it is +1.73%, and at the 2.55% midpoint of the reported range it is +1.66%.

Now put the fee back in. A 10% protocol fee on that gross rate costs 25 basis points, which is about 15% of the real yield. A 35% exchange commission costs roughly 89 basis points and leaves a real yield of around 0.71% — less than half of what a zero-fee validator keeps. This is the arithmetic that should drive route selection, and it is why the difference between a 10% and a 35% provider matters far more on Ethereum than the difference between one exchange's marketing copy and another's.

It is also worth keeping the number in proportion. ETH traded above $4,900 at its 2025 high and around $1,700 in mid-June 2026. A 2.46% annual rate is roughly nine days of typical volatility. Staking is a sensible thing to do with ETH you intend to hold anyway; it is not a reason to hold ETH. We develop that distinction onpassive income with crypto, and the counterparty questions on crypto earn risks.

Where the rules currently stand

US permissions for staking improved through 2025 and 2026 but still rest on agency interpretation rather than statute. The SEC's Division of Corporation Finance stated on 29 May 2025 that solo, self-custodial, delegated and custodial protocol staking are not securities transactions, with Commissioner Caroline Crenshaw dissenting; a further staff statement on 5 August 2025 addressed liquid staking receipt tokens under conditions. On 17 March 2026 a joint SEC–CFTC Commission-level interpretation published at 91 FR 13714 placed Ether among digital commodities. Restaking remains excluded from all of them. The detail sits on crypto yield regulation, and geographic availability on where earn products are available.