Proof of Stake (PoS): How It Works, Slashing, Finality and Comparison with Proof of Work
Proof of stake (PoS) is a family of blockchain consensus mechanisms. Validators lock up the network's native token as collateral, and the protocol chooses them to propose and vote on blocks, usually in proportion to their stake. Honest work earns rewards. Provable misbehaviour can be punished by destroying part of the stake, which is called slashing. This article covers how PoS works, using Ethereum as the main example: slots, epochs, finality and penalties. It also covers the history of PoS from 2012, how it compares with proof of work on energy and attack cost, the main design variants, known weaknesses and how staking is treated by regulators.
What is proof of stake?
Proof of stake (PoS) is a type of consensus mechanism: the set of rules that lets the computers in a decentralized network agree on which blocks of transactions are valid. In proof of work (PoW), miners compete by spending computing power. In PoS, participants called validators lock up the network's native token as collateral. The protocol gives them the right to propose and approve blocks. Ethereum's documentation describes PoS as "a way to prove that validators have put something of value into the network that can be destroyed if they act dishonestly" [1]. The US Securities and Exchange Commission (SEC) staff give a similar description. Node operators stake assets so that the protocol software can select them to validate transactions. They earn newly created tokens and transaction fees as rewards. The staked amounts are locked up under the protocol's terms but still belong to the staker [12].
How proof of stake works
Staking and validator selection
A validator first commits capital. On Ethereum this means depositing ETH into a smart contract and running three pieces of software: an execution client, a consensus client and a validator client [1]. The minimum needed to take part in consensus is 32 ETH. Since the Pectra upgrade of 7 May 2025, one validator can hold between 32 and 2,048 ETH of effective balance [6]. Before that, the maximum was 32 ETH, so large stakers had to run many separate validators [7].
Block proposal and attestation
Ethereum divides time into 12-second slots. Each epoch contains 32 slots. In every slot, one validator is chosen at random to propose a block. A randomly chosen committee of validators then votes on whether the block is valid. These votes are called attestations [1].
Finality
Finality means a block cannot be reversed without a very large economic cost. On Ethereum, the first block of each epoch serves as a checkpoint. If a pair of checkpoints receives votes from validators holding at least two-thirds of all staked ETH, the checkpoints are upgraded. Once two checkpoints are linked by such a supermajority, the earlier one becomes finalized [1].
Rewards, penalties and slashing
Validators earn rewards for doing their duties correctly. Slashing is reserved for provably dishonest behaviour, such as proposing two blocks for the same slot or sending contradictory attestations. A slashed Ethereum validator loses part of its stake straight away. It then faces a 'correlation penalty' around day 18, which is larger when many validators are slashed at the same time, and it is removed from the network on day 36 [1]. If the chain fails to finalize for more than four epochs, an 'inactivity leak' starts. It gradually drains the stake of validators that are not voting, until the remaining active validators again hold a two-thirds majority [1].
| Ethereum PoS parameter | Value |
|---|---|
| Slot length | 12 seconds [1] |
| Slots per epoch | 32 [1] |
| Minimum stake per validator | 32 ETH [6] |
| Maximum effective balance (since Pectra) | 2,048 ETH [6] |
| Finality threshold | At least two-thirds of total staked ETH [1] |
| Inactivity leak trigger | No finality for more than four epochs [1] |
History of proof of stake
- August 2012: the paper describing PPCoin (later Peercoin) was published. Its authors were listed as Scott Nadal and Sunny King, and 'Sunny King' is a pseudonym. Peercoin launched on 12 August 2012 and used both proof of stake and proof of work. It is widely described as the first cryptocurrency to use PoS [10].
- 2017: Aggelos Kiayias, Alexander Russell, Bernardo David and Roman Oliynykov presented "Ouroboros: A Provably Secure Proof-of-Stake Blockchain Protocol" at CRYPTO 2017 (pages 357–388). The authors describe it as the first PoS blockchain protocol with rigorous security guarantees [9].
- 1 December 2020: Ethereum launched the Beacon Chain. This was a separate PoS chain that ran alongside the main network [3].
- 15 September 2022: in 'the Merge', Ethereum's main network joined the Beacon Chain and permanently stopped using proof of work [3].
- 7 May 2025: Ethereum's Pectra upgrade raised the maximum effective balance per validator from 32 to 2,048 ETH [6][7].
Proof of stake vs proof of work
| Aspect | Proof of stake | Proof of work |
|---|---|---|
| Resource at risk | Staked tokens, which can be slashed [1] | Mining hardware and electricity |
| Punishment for attackers | Stake can be destroyed and the validator removed [2] | Hardware can be reused to attack new forks [2] |
| Barrier to entry | 32 ETH to run an Ethereum validator [2] | Hardware 'arms races' that tend to price out individuals [2] |
| Complexity | Described by Ethereum as "much more complex" than PoW [2] | Simpler design with a longer operating history |
Energy use
The biggest practical difference is energy. In May 2021, the Ethereum Foundation made what it called a rough ('ball-park') estimate. It put the PoS network's power use at about 2.62 megawatts, against about 44.49 TWh per year for PoW Ethereum (a figure it took from Digiconomist). On that basis, PoS would be about 2,000 times more energy-efficient, a reduction of at least 99.95% [8]. After the Merge, ethereum.org gives the reduction as an estimated 99.95% on one page [3]. Another ethereum.org page gives about 99.98%, measured against pre-Merge use of about 78 TWh per year [2]. These are estimates, and the figures differ depending on the method used.
Cost of attack
Ethereum's documentation gives an illustrative example. With 14 million ETH staked at a price of $1,000, an attacker would need more than a third of all staked ETH, or at least $4.62 billion, to disrupt finality. Ethereum estimates that a 51% attack is about 20 times cheaper on PoW than on PoS [2]. These figures change with the token price and the total amount staked.
Variants of proof of stake
- Bonded validators with slashing: validators post their own collateral and can lose it for provable misbehaviour, as on Ethereum [1].
- Delegated or liquid models: token holders hand their block-production rights to a validator while keeping ownership of their tokens. On Tezos, delegates need a minimum stake of 6,000 tez. Any other account can change or revoke its delegate at any time. Staked tez act as security deposits that can be forfeited if the baker (block producer) breaks certain protocol rules [11].
- Research-driven designs: Ouroboros brought formal security proofs and a reward scheme. Its authors show that, under this scheme, behaving honestly is an approximate Nash equilibrium (no participant gains much by deviating) [9].
- Pooled and custodial staking: third-party services, pools and custodians stake on behalf of token holders [1][12].
Security issues and criticisms
Nothing-at-stake problem
Ethereum describes 'nothing at stake' as a conceptual problem in PoS designs that offer only rewards and no penalties. In such systems, validators could back several competing chains at no cost. Ethereum addresses this with finality rules and slashing [5].
Long-range attacks and weak subjectivity
In a long-range attack, early validators keep an alternative history and later try to convince honest nodes to switch to it. Ethereum's finality mechanism prevents this. New nodes start from a recent trusted state, called a 'weak subjectivity checkpoint', and treat it like a new starting block [4].
Stake thresholds
| Share of stake controlled by an attacker | What it could enable on Ethereum |
|---|---|
| 33% | Block finality, which the inactivity leak then works to restore [4] |
| More than 50% | Censor transactions and reorganize short sections of the chain [4] |
| 66% or more | Finalize its preferred chain and rewrite history [4] |
If attackers ever finalized a dishonest chain, Ethereum's documentation says the community could coordinate a response. Options include moving to an honest fork and forcing out or burning the attackers' stake [4].
Centralization
Critics argue that PoS can concentrate influence among large holders. Ethereum's own FAQ warns that liquid staking tokens can lead to "a few big organizations" controlling much of the stake [5]. Its comparison page lists liquid staking among the centralization concerns [2].
Staking in practice and regulation
On Ethereum, you can stake on your own (solo staking), through a pool, or through a staking-as-a-service provider [1]. Staking is optional and is not needed to use the network [5]. The risks include slashing, penalties for being offline, and your tokens being locked for a period.
On 29 May 2025, the SEC's Division of Corporation Finance issued a staff statement. It concluded that certain protocol staking activities do not involve the offer and sale of securities, covering solo staking, self-custodial staking with a third-party operator, and custodial staking. The statement represents the views of SEC staff only and "has no legal force or effect" [12]. Rules on staking and taxes vary by jurisdiction.
References
- officialProof-of-stake (PoS) — ethereum.org · Evidence page
- officialProof-of-stake vs proof-of-work — ethereum.org · Evidence page
- officialThe Merge — ethereum.org · Evidence page
- officialEthereum proof-of-stake attack and defense — ethereum.org · Evidence page
- officialProof-of-stake Frequently Asked Questions — ethereum.org · Evidence page
- officialPrague-Electra (Pectra) — ethereum.org · Evidence page
- newsEthereum Activates 'Pectra' Upgrade, Raising Max Stake to 2,048 ETH — CoinDesk · Evidence page
- officialEthereum's energy usage will soon decrease by ~99.95% — Ethereum Foundation Blog · Evidence page
- academicOuroboros: A Provably Secure Proof-of-Stake Blockchain Protocol (Advances in Cryptology – CRYPTO 2017, pp. 357–388, doi:10.1007/978-3-319-63688-7_12) — University of Edinburgh Research Explorer / Springer · Evidence page
- otherPeercoin — Wikipedia · Evidence page
- officialProof-of-stake — Octez & Protocol documentation — Tezos / Octez documentation · Evidence page
- governmentStatement on Certain Protocol Staking Activities — U.S. Securities and Exchange Commission, Division of Corporation Finance · Evidence page