Proof of stake secures a blockchain with collateral instead of electricity: validators lock up the network's own coins, take turns proposing blocks and vote on each other's work. Honest validators earn newly created coins, and a validator caught signing two contradictory messages loses part of its deposit. Ethereum, the largest network run this way, supplies most of the numbers below. The crypto hub covers the wider market, and what is Ethereum describes the network these rules protect.
How does proof of stake work?
Validators lock coins with the protocol, which picks one at random, weighted by stake, to propose each block while committees drawn from the rest check it and vote.
On Ethereum the cycle runs on a fixed clock:
- Bond. A validator deposits at least 32 ETH, the activation minimum in the consensus rules. Since the Pectra upgrade on May 7, 2025, one validator can carry up to 2,048 ETH.
- Propose. Time runs in 12-second slots. For each slot the protocol draws one proposer using a shared random value called RANDAO, with odds proportional to each validator's effective balance.
- Vote. Every validator votes once per 6.4-minute epoch of 32 slots, backing the head of the chain and a pair of checkpoints. Once two-thirds of the stake backs the same pair, the earlier checkpoint and every block before it become final, a process how blockchain works traces in detail.
- Leave. Exits wait in a queue, and a withdrawal sweep then returns the ether to the owner's address.
Ethereum adopted these rules in the Merge on September 15, 2022, when its proof-of-stake Beacon Chain, running since December 1, 2020, took over from miners. The switch cut the network's energy use by about 99.95%, according to ethereum.org. On November 2, 2026, validators had staked about 44.3 million ETH, or 36% of all ether, across 815,282 active validators.
Why does proof of stake need slashing?
Signing a block costs a validator almost nothing, so without penalties it could back every competing fork at once; slashing makes that provably expensive.
Vitalik Buterin's proof-of-stake FAQ calls this the nothing-at-stake problem. Early designs such as Peercoin paid for blocks and penalized nothing, so the rational move during a fork was to build on every branch at once, just in case. A miner has no such option, because splitting hash power between branches splits its chance of a reward. Slashing closes the gap: anyone can submit two conflicting messages signed by the same key, and the protocol then destroys part of the signer's stake and forces it out. On Ethereum the slashable offenses are proposing two different blocks for one slot and casting votes that contradict each other.
The punishment scales with how many others cheat at the same time. Since Pectra, a slashed validator loses 1/4,096 of its effective balance at once, its ether cannot be withdrawn for about 36 days, and halfway through that wait comes a second charge:
If validators holding 5% of all stake were slashed in the same window, each would forfeit 3 × 5% = 15% of its balance in that second charge alone. A third of the stake signing conflicting checkpoints together would lose everything, which is why reversing a finalized block is so costly. Being offline is not slashable; it costs missed rewards, and more only if the chain stops finalizing. The cost of one slashing to a 32 ETH validator, step by step, is in what is staking.
How does Ethereum set staking rewards?
The protocol pays a yearly rate that falls with the square root of all ether staked, so total issuance grows as staking rises while each validator's rate shrinks.
The 166.3 comes from the consensus specification: a base reward factor of 64, times 225 epochs a day and 365.25 days, divided by the square root of 1 billion gwei per ether, or 64 × 225 × 365.25 ÷ 31,623 ≈ 166.3. The result is a ceiling that assumes every validator performs every duty on time, and it counts consensus rewards only.
Worked example: Reward ceilings at two levels of staking
- With 44.3 million ETH staked, as on November 2, 2026: √44,300,000 ≈ 6,656, so the ceiling is 166.3 ÷ 6,656 ≈ 2.50% a year, about 0.80 ETH on a 32 ETH validator.
- New ether paid out at that ceiling: 166.3 × 6,656 ≈ 1.11 million ETH a year.
- With 60 million ETH staked: √60,000,000 ≈ 7,746, so the ceiling drops to about 2.15% while total issuance climbs to about 1.29 million ETH.
- The network rate validatorqueue.com measured over the seven days to November 2, 2026 was 2.45%, just under the ceiling.
Block proposers also keep priority fees and payments from block builders, and those arrive in bursts. In Lido's daily reports, the rate paid to holders of its stETH token was between 2.24% and 2.34% on five of the seven days to November 1, 2026, but 15.08% in the October 30 report, which covered the October 29 surge in gas fees. How US tax treats rewards is in staking taxes, and the spot ether funds that stake are in Ethereum ETFs explained.
What kinds of proof of stake are there?
Designs differ in who produces blocks, whether coins are locked and whether cheats lose stake, and four networks show the range.
| Design | How block producers are chosen | Locked coins and penalties | Example |
|---|---|---|---|
| Coin-age staking | Coins held longer win more often, scored as amount times days held | No slashing; history protected by checkpoints broadcast centrally a few times a day | Peercoin, paper dated August 19, 2012 |
| Validator staking with slashing | Anyone with the minimum runs a validator; proposers are drawn at random, weighted by stake | Deposit locked until exit; conflicting signatures are slashed | Ethereum, 32 ETH minimum |
| Delegated proof of stake | Stakers vote, and the top vote-getters produce every block | Producers can be voted out at regular re-rankings | Tron, 27 Super Representatives |
| Pool delegation | Holders assign stake to pools, which win slots in proportion to stake | Coins stay spendable; no slashing | Cardano, using the Ouroboros protocol |
The Peercoin paper gives the coin-age idea a simple shape: an output holding 200 coin-years can expect to find a block twice as fast as one holding 100. Security thresholds also vary. Cardano's documentation says Ouroboros stays secure while honest participants hold most of the stake, whereas Ethereum needs two-thirds of the stake voting together before anything becomes final.
How is proof of stake different from proof of work and staking?
Proof of stake is a network rule; staking is the act of locking coins under it, and proof of work is the rival rule built on electricity.
| Term | What it is | How it relates to proof of stake |
|---|---|---|
| Proof of work | Miners spend electricity searching for a valid block hash | The alternative rule; Bitcoin uses it, and Ethereum did until 2022 |
| Staking | Locking coins with a validator, pool or exchange to earn rewards | The activity proof of stake makes possible |
| Liquid staking | Staking through a smart contract that issues a tradable token for the deposit | A wrapper whose token can trade below the coins behind it |
| Restaking | Pledging already-staked ether to secure other services | Adds a second set of slashing rules on top of the network's own |
| Lending | Handing coins to a borrower for interest | Not staking; the return depends on the borrower repaying |
The liquid staking discount is not hypothetical. In mid-June 2022, before Ethereum allowed withdrawals, stETH fell to about 0.93 ETH as large holders sold it to meet withdrawals and margin calls, according to Genesis Trading's monthly review; withdrawals have worked since the Shapella upgrade of April 12, 2023. Mining is covered in the proof of work entry, the code behind staking pools in smart contract, and both networks side by side in Bitcoin vs Ethereum. The live ether price sets what any reward is worth in dollars.
What do people get wrong about proof of stake?
The commonest error is treating the staking rate as fixed interest, when it moves with total stake, fees and validator performance and is paid in a volatile coin.
- "Taking part needs 32 ETH." That is the floor for running a validator. Pools and exchanges accept small amounts, and running an ordinary node that does not validate needs no ether at all.
- "The biggest holder decides every block." Selection is random and weighted, so a validator set with 1% of the stake proposes about 1% of blocks over time.
- "Rewards protect against price falls." They are paid in the same coin, so a 2.5% yearly reward barely dents a 50% fall in ether's price.
- "Final means impossible to change." Finality is an economic guarantee: reversing it would cost at least a third of all staked ether, a vast sum, but not an infinite one.