Proof of stake is a blockchain consensus mechanism in which the right to propose and confirm new blocks is given to validators who lock up the network’s native tokens as collateral, known as stake, instead of to miners who spend computing power. Among consensus mechanisms it is the main alternative to proof of work, and its security rests on an economic principle: validators who follow the rules earn rewards, while validators who attack or misbehave lose part of their stake through slashing.
The idea replaces an external, physical cost, namely electricity and hardware, with an internal, financial cost denominated in the network’s own asset. Ethereum switched to proof of stake in 2022 in an upgrade known as the Merge, and many newer networks use it from the start, often in combination with Byzantine fault tolerant voting.
How proof of stake selects and confirms blocks
Proof of stake confirms blocks in two steps: a pseudo-randomly selected validator proposes a block, and a larger group of validators votes on it with signed attestations. A block that gathers votes representing a sufficient share of the total stake becomes part of the canonical chain.
Selection is weighted by stake, so a validator with more stake is chosen more often and earns proportionally more. The randomness must be unpredictable and hard to manipulate, which is why protocols derive it from combined contributions of many validators rather than from a single source.
| Element | Role | Example from Ethereum |
|---|---|---|
| Validator | Proposes blocks and casts votes | Node operating with deposited ETH |
| Stake deposit | Collateral that can be slashed | Minimum of 32 ETH per validator |
| Slot | Time window for one block proposal | 12 seconds |
| Epoch | Group of slots for voting and checkpoints | 32 slots, about 6.4 minutes |
| Attestation | Signed vote on the chain head and checkpoints | Cast by each validator once per epoch |
| Finality gadget | Makes checkpoints irreversible | Casper FFG with two-thirds supermajority |
On Ethereum a checkpoint becomes final when validators holding at least two-thirds of the staked ETH vote for it in two consecutive epochs, so finality normally arrives after about two epochs, roughly 12 to 13 minutes. Reverting a finalized block would require at least one-third of all stake to be slashed.
Rewards, penalties and slashing
Proof of stake aligns behavior through three economic levers: rewards, inactivity penalties and slashing. Rewards compensate honest participation, penalties discourage downtime, and slashing punishes provably malicious actions such as signing two conflicting blocks for the same slot.
| Incentive | Triggered by | Typical consequence |
|---|---|---|
| Block and attestation rewards | Timely, correct participation | New tokens and transaction fees |
| Inactivity penalty | Being offline or missing votes | Small, gradual loss of balance |
| Inactivity leak | Chain fails to finalize for a long period | Rising losses for inactive validators |
| Slashing | Double signing or contradictory votes | Loss of stake and forced exit |
| Correlation penalty | Many validators slashed at the same time | Much larger loss for coordinated faults |
The correlation penalty matters for operators: an isolated mistake is punished mildly, but if a large share of validators is slashed together, as in a coordinated attack or a shared misconfiguration, penalties rise sharply. This discourages concentrating many validators on identical infrastructure.
Proof of stake compared with other consensus mechanisms
Proof of stake differs from proof of work mainly in what an attacker must acquire. In proof of work an attacker needs a majority of hash rate, which means hardware and energy; in proof of stake the attacker needs a large share of the staked tokens, which would be destroyed by slashing if used to attack.
| Mechanism | Scarce resource | Energy use | Finality |
|---|---|---|---|
| Proof of work | Computing power and electricity | Very high | Probabilistic |
| Proof of stake | Locked native tokens | Very low | Economic finality in many designs |
| Delegated proof of stake | Votes from token holders to delegates | Very low | Fast, with few block producers |
| Proof of authority | Approved validator identities | Very low | Fast, relies on reputation |
| Classical BFT | Fixed set of known nodes | Very low | Immediate, limited validator count |
The switch of Ethereum from proof of work cut the network’s energy consumption by well over 99 percent, the most frequently cited advantage of the model. Proof of stake also enables faster and explicit finality, whereas in Nakamoto consensus a block only becomes increasingly unlikely to be reversed as more blocks follow.
Participating in proof of stake
Token holders can participate in proof of stake by running a validator, delegating stake or using a pooled or liquid staking service. Each route trades control and technical effort against convenience and additional counterparty risk.
- Solo validation: the holder runs validator software and keeps full control of keys, but must ensure high uptime and correct configuration.
- Delegation: on networks that support it natively, holders assign stake to an operator while keeping ownership of the tokens.
- Staking pools: many holders combine smaller amounts to meet the validator minimum and share rewards.
- Liquid staking: a protocol issues a tradable token representing staked assets, adding smart contract risk.
- Exchange staking: a custodial platform stakes on the user’s behalf, which is simple but surrenders control of the keys.
Staking yields are variable and not guaranteed. They depend on total stake, network activity and protocol rules, and they are paid in a token whose market price can fall faster than the rewards accumulate.
Criticisms and known weaknesses
The main criticism of proof of stake is the risk of stake concentration. Because rewards are proportional to stake, large holders and popular staking services can accumulate influence, and a few providers controlling a large share of stake weakens the decentralization the mechanism is meant to protect.
A second weakness is the long-range attack: keys of validators who have since exited could be used to forge an alternative history from far in the past, because signing old blocks costs nothing. Protocols counter this with weak subjectivity checkpoints, recent trusted states that new nodes use when they first synchronize.
Further debated issues include the nothing-at-stake problem, in which validators could vote on several competing forks at no cost, which slashing is designed to remove, and regulatory uncertainty about how staking rewards and staking services are treated in different jurisdictions.
Understanding it simply
Proof of stake decides who may add the next page to a blockchain’s record book. Instead of running power-hungry machines, participants put up their own money as a security deposit.
Those who keep the rules earn a reward. Those who try to cheat lose part of their deposit. It works like a bond that a contractor posts before a job. Honesty pays, and cheating costs more than it could ever gain.
Applied in practice
A small infrastructure company decides to run Ethereum validators for its own treasury. It deposits 32 ETH per validator and operates the software on two servers in different data centers, with monitoring that alerts staff when a validator misses attestations.
To avoid slashing, it never runs the same validator keys on two machines at once, a classic cause of accidental double signing, and it uses a slashing protection database that refuses to sign conflicting messages. Over the following year its validators propose occasional blocks and attest every epoch, earning rewards in ETH that fluctuate with network conditions, while a brief outage during a server migration costs only a small inactivity penalty rather than a slashing loss.