When you stake ETH, the ETH visibly leaves your wallet — and most explanations stop right there, at "it starts earning rewards." That gap is where the confusion lives. People stake, then notice things nobody warned them about: the rewards don't start immediately, the balance updates in odd increments, and getting out takes longer than getting in. None of this is a malfunction. Staked ETH moves through a defined lifecycle with distinct stages, and each stage exists for a reason the protocol's designers can articulate.
This post walks through that lifecycle — deposit, activation, active duty, rewards, and exit. Choosing how to stake (solo, delegated, pooled, or through an exchange) is a separate question, covered in how to stake Ethereum. Here the question is narrower: once ETH is committed, what does the protocol actually do with it?
Stage One: The Deposit
Staking begins with a transaction to the deposit contract — a smart contract on Ethereum's execution layer that has held ETH since 2020 and has no withdrawal function. That's not an oversight. The deposit contract is a one-way bridge: ETH goes in on the execution layer, and the consensus layer — the part of Ethereum that runs proof of stake — observes the deposit and credits a validator with the balance. Funds come back through an entirely different mechanism at the end of the lifecycle, which is part of why exits feel asymmetric to deposits.
Two things are fixed at deposit time. The first is the validator's public key, which will sign its duties. The second is the withdrawal credentials — the address that will eventually receive rewards and the returned stake. These are deliberately separate. The validator key works constantly and lives on a machine that's online all day; the withdrawal credentials can sit in cold storage. If the validator key is compromised, the attacker can get the validator penalized, but they can't redirect the funds.
If you're staking through a pool or an exchange rather than depositing 32 ETH yourself, this stage still happens — it just happens at the protocol's level, with the pool's contract or the custodian's wallet making the deposit. Your ETH backs a validator either way. The lifecycle below is running underneath every staking product, whatever the interface shows you.
Stage Two: The Activation Queue
A new validator doesn't start working the moment the deposit confirms. It enters an activation queue, because the protocol caps how many validators can join (or leave) per epoch — a limit usually called the churn limit, which scales with the size of the validator set. The cap is a security decision: if stake could flood in or out instantly, an attacker could rapidly shift the composition of the validator set, and the chain's guarantees about finality assume that composition changes slowly.
The practical consequence is that activation time varies with demand. When few people are joining, it's hours. During heavy inflows — mid-2023 is the reference episode — the queue stretched to over a month. Your ETH is committed during the wait and earning nothing, which surprises people who expected staking to behave like a savings account deposit. It's closer to joining a workforce: there's a hiring pipeline, and it's rate-limited on purpose.
Stage Three: Active Duty
Once activated, the validator works on a fixed rhythm. Every epoch — about six and a half minutes — it's expected to publish one attestation: a signed vote on what the head of the chain is and which checkpoints should be finalized. Attestations are the bulk of the job and the bulk of the rewards. They're small, constant, and unglamorous.
Occasionally the validator is selected to do something bigger. Block proposals are assigned pseudo-randomly, weighted by stake — a single validator might propose a handful of blocks per year. Proposals pay disproportionately, because the proposer collects priority fees and MEV from the transactions in the block, on top of the protocol reward. There are also sync committee assignments, rarer still and similarly overpaid relative to attestations. The result is that validator income is lumpy by design: a steady drip from attestations, punctuated by occasional proposal payouts that depend partly on luck and partly on what's happening on-chain that day.
The protocol tracks each validator's effective balance — the stake that actually counts for duties and rewards. Since the Pectra upgrade, that balance can compound up to 2,048 ETH rather than being capped at 32, so rewards left in place now increase a validator's weight instead of idling.
What about failure? Going offline isn't a catastrophe — an offline validator leaks roughly what it would have earned, so a day of downtime costs a day of rewards, give or take. Slashing is the severe penalty, reserved for provable equivocation: signing two conflicting blocks or contradictory attestations. A slashed validator loses a portion of its stake and is forcibly ejected, and the penalty scales up if many validators are slashed at once — a correlation rule aimed squarely at large operators running many validators on shared infrastructure. In practice, slashing events trace back to misconfiguration, almost always the same one: the same validator keys running in two places at once.
Stage Four: Getting Out
Exiting reverses the lifecycle, with the same throttling. The validator broadcasts a voluntary exit, joins an exit queue governed by the same churn logic as activation, stops duties once processed, and then waits for its balance to become withdrawable. The actual transfer happens via a sweep — the protocol cycles through validators continuously, pushing withdrawable balances out to their withdrawal addresses. The same sweep is what delivers partial reward withdrawals during active duty, which is why staking rewards arrive on a cycle rather than the instant they're earned.
Under normal conditions the full exit path clears in hours to days. Under mass-exit conditions it's designed to take much longer — that's the point. A system secured by stake can't allow the stake to vanish overnight.
What's Changing
Pectra, shipped in 2025, modified this lifecycle in three places worth knowing. EIP-7251 raised the maximum effective balance to 2,048 ETH, enabling consolidation and reward compounding. EIP-6110 moved deposit processing onto the execution layer, cutting the lag between depositing and entering the queue from many hours to minutes. And EIP-7002 made exits triggerable from the withdrawal credentials — so the party who owns the funds can always initiate an exit, even if the party running the validator won't cooperate. The lifecycle's stages are unchanged; several of its delays and dependencies shrank.
Confirmation, Invalidation, Timing
Confirmation that this picture holds: activation and exit queues continuing to clear within churn-limit expectations through volatile periods, slashing remaining rare and traceable to operator error, and post-Pectra consolidation proceeding without consensus incidents.
Invalidation: a slashing event caused by client software bugs rather than misconfiguration would break the "penalties are predictable" claim. Exit queues extending to months under stress would mean the liquidity assumptions around staked ETH need rewriting.
Now: the lifecycle is fully operational end to end — deposits, withdrawals, and credential-triggered exits all live. Next: watch how validator consolidation changes queue dynamics and the slashing correlation math. Later: further protocol changes to issuance or churn parameters are under research discussion, not scheduled.
Boundary
This describes what the Ethereum protocol does with staked ETH. It isn't a recommendation to stake, a yield estimate, or a comparison of staking providers. Liquid staking tokens add a contract layer on top of this lifecycle that has its own risks, covered separately. Tax treatment of rewards at each stage varies by jurisdiction and is out of scope. The mechanism is the static part; whether participating makes sense for anyone in particular depends on things this post deliberately doesn't address.
