Proof of work vs proof of stake: one pays the power bill, the other puts up the deposit
Proof of work makes the right to write the next block cost electricity and machines; proof of stake makes it cost a deposit that can be destroyed. Both make cheating expensive. They differ in what the cheater loses, how much energy they use and who can take part.
Two villages keep the same kind of notebook, copied in every house. In the first, the right to write the next page goes to whoever digs fastest. Digging costs firewood, and the firewood is burned whether your page is accepted or not.
In the second village, nobody digs. Villagers lock part of their savings in the deposit box by the well. Whoever is picked writes the page, the others check it and sign. Sign two different pages, and the village takes part of your box.
Ben wants to cheat both villages. In the first, he has to outdig everyone and pay for the firewood as he goes. In the second, he has to lock up a third of the box and hand it over when he's caught. That's the whole comparison: proof of work bills you as you go, proof of stake holds your money hostage.
What it is
Proof of work and proof of stake are the two big answers to the same question: who writes the next block, and what stops a cheat? Proof of work, which Bitcoin uses, makes the right cost computing power and electricity. Proof of stake, which Ethereum has used since the Merge on 15 September 2022 (ethereum.org), makes it cost a locked deposit of coins.
Neither is a vote by heads. Both make a vote expensive, so Ben can't win by pretending to be a thousand people. How the generals in "The Byzantine generals problem" learned to agree is the theory; these are the two working answers.
How it works: what Ben has to spend
Here is the bill for rewriting the notebook in each village, with real numbers read on 19 November 2026.
In the digging village (Bitcoin).
- The honest diggers. mempool.space estimated Bitcoin's hash rate at about 992 exahashes per second over the previous three days, at a difficulty of 134,823,837,323,272.7 (mempool.space).
- What Ben needs. More than that, of his own: more machines than all other miners together, running at the same time. Less than half can get lucky over a block or two, but over time the bigger side wins.
- The power bill. ethereum.org's comparison table, built from estimates it accessed in July 2023, puts Bitcoin at about 149 terawatt-hours a year (ethereum.org). Divided by 365, that's about 408 gigawatt-hours a day. Ben would have to match it, every day of the attack.
- What he keeps. The electricity is gone, win or lose. The machines are not. ethereum.org notes that after a failed attack and a fork, the attacker can reuse the same hardware on the new chain (ethereum.org).
In the deposit village (Ethereum).
- The honest box. beaconcha.in showed 44,339,549 ETH of active stake and 784,458 validators in finalised epoch 490,260 (beaconcha.in).
- To jam the notebook. Finality needs votes from two-thirds of all stake. With one-third, an attacker "could prevent the network from reaching finality" (ethereum.org). One-third of 44,339,549 is about 14.78 million ETH.
- To finalise his own pages alone. Two-thirds: about 29.56 million ETH.
- What he keeps. Possibly nothing. If his validators sign conflicting pages, "some or all of their staked ETH can be destroyed" (ethereum.org). The box is the bill.
ethereum.org's own verdict is that a 51% attack is "~20x less expensive on proof-of-work than proof-of-stake" (ethereum.org). Read that as the Ethereum side's estimate, not a neutral referee's.
Why it matters
- Energy. This is the clearest difference. The same ethereum.org table gives proof-of-work Ethereum about 21 TWh a year and proof-of-stake Ethereum about 0.0026 TWh. That's roughly 8,000 times less, the drop of more than 99.98% the page cites from the research firm CCRI (ethereum.org).
- Who can join. In proof of work, anyone can buy a machine and plug it in, nobody's permission needed. But ethereum.org says "mining hardware arms races tend to price out individuals and small organizations" (ethereum.org). In proof of stake, you need the coins: 32 ETH to activate a validator, under the rule in EIP-7251 (EIP-7251). Since the Pectra upgrade of 7 May 2025, one validator can hold up to 2,048 ETH (ethereum.org). Smaller holders use pools.
- Who pays. In proof of work, miners pay electricity companies, and the network pays miners in new coins and fees. In proof of stake, the cost is the money stakers can't spend while it's locked, plus the risk of losing it.
- Track record. Bitcoin's proof of work has run since January 2009. ethereum.org itself lists as a downside that proof of stake is "younger and less battle-tested" and "more complex to implement" (ethereum.org).
Where people get it wrong
- "Proof of stake is proof of work without the waste." It swaps one cost for another. The power bill goes, and so does the link to anything outside the chain: the deposit is paid in the chain's own coin.
- "Proof of work has no punishment." It has one, just a quieter one. A miner who builds on a losing chain has burned the electricity for nothing.
- "Proof of stake means the rich decide." Stake buys the right to propose and vote, under rules every node checks. Big holders do have more weight, as big miners do in proof of work. How spread out the stake really is gets its own article, "Who really controls a proof-of-stake chain?".
- "One of them is simply better." Each is a set of trade-offs: energy against track record, open entry against a deposit you can lose.
Look up the size of the box
Before you trust a small chain with a payment, look up one number. For a proof-of-work chain, its hash rate. For a proof-of-stake chain, its total stake. Then ask how much it would cost to bring a third, or a half, of that.
A big number makes Ben's bill big. A small number means the notebook is guarded by a garden fence, and Ben owns a ladder.
Frequently asked questions
What is the main difference between proof of work and proof of stake?
In proof of work, miners spend computing power and electricity for the right to add blocks. In proof of stake, validators lock up coins as a deposit, and cheating can destroy part or all of it. Both make attacks expensive, in different currencies.
Which uses less energy, proof of work or proof of stake?
Proof of stake, by a wide margin. ethereum.org cites an estimate by CCRI that the Merge cut Ethereum's annualised electricity use by more than 99.98%, from about 21 terawatt-hours a year to about 0.0026.
Is proof of stake less secure than proof of work?
Not by default, but differently. ethereum.org argues an attack is far more expensive under proof of stake because the attacker's deposit can be destroyed. It also concedes that proof of stake is younger and less battle-tested than proof of work.
How much do you need to stake to run an Ethereum validator?
32 ETH to activate a validator. Since the Pectra upgrade on 7 May 2025, a single validator can hold up to 2,048 ETH. People with less ETH usually stake through pools or services.
Can a proof-of-stake chain be 51% attacked?
In principle, yes, but the majority is counted in staked coins, not computing power. On Ethereum, one-third of all stake is enough to stop finality, and an attacker caught signing conflicting blocks can lose some or all of the deposit.
Sources
- ethereum.org, Proof-of-stake vs proof-of-work
- ethereum.org, Proof-of-stake (PoS)
- ethereum.org, The Merge
- ethereum.org, Ethereum energy consumption
- ethereum.org, Pectra
- EIP-7251: Increase the MAX_EFFECTIVE_BALANCE
- mempool.space API, Bitcoin hash rate and difficulty (read on 19 November 2026)
- beaconcha.in, finalised epoch 490,260: active stake and validators (read on 19 November 2026)

