Proof of work and proof of stake, compared on their own terms

The energy question follows directly from the cost model. Proof of work converts electricity into security, and the security budget is therefore tied to the price of power and the price of the asset: a miner will keep hashing while the expected reward exceeds the cost of the electricity. This is why the mechanism is often described as anchoring the chain to a physical input, and why its critics treat the same property as waste. The honest framing is that proof of work buys a specific kind of assurance, and the electricity is what it costs.

Proof of stake removes the external expenditure and replaces it with a capital requirement. Participation is gated by how much of the asset a validator is willing to lock, not by access to cheap power or specialised machines. That lowers the barrier in one direction and raises it in another: a validator needs to acquire the asset rather than a machine, and the return on the bond competes with every other use of that capital. Ethereum's documentation notes that issuance is divided among validators and that the total depends on how much ether is staked, so the yield falls as more capital joins.

The two designs also differ in what an attacker must acquire. To attack a proof-of-work chain profitably, an adversary needs a majority of hash rate, which is hardware and energy that can be rented or bought on a market. To attack a proof-of-stake chain, an adversary needs a majority of the staked asset, which is a position that must be accumulated in public markets and that loses value if the attack succeeds. Neither requirement is trivially satisfiable, and the two are not directly comparable in cost.

The mechanisms produce different settlement behaviour. Bitcoin's chain is extended by whichever miner finds the next valid block, and a transaction is considered settled as it accumulates confirmations; the assurance is probabilistic and grows with depth. Ethereum's proof-of-stake chain reaches finality through attestations, in which validators vote on checkpoints and a supermajority of the staked weight finalises them. The specification's attestation machinery, including committees, aggregation and the slashing conditions that punish conflicting votes, exists to make that finality enforceable.

The validator economy is where the two systems diverge most sharply in practice. A miner's revenue is the block subsidy plus fees, and it is spent on hardware and power that has a resale value and an alternative use. A validator's revenue is issuance plus priority fees, and its costs are the opportunity cost of the bond, the infrastructure to run a node, and the risk of penalties for being offline or for misbehaving. Ethereum's documentation records that validators can be penalised for going offline and slashed for breaking the rules, with the penalty removed from circulation rather than paid to anyone.

Neither mechanism is a free lunch, and neither is a settled scientific question. Proof of work has the longer operating record and a security argument that rests on physics and markets. Proof of stake has a far smaller energy footprint and a penalty mechanism that acts inside the protocol, and it depends on the assumption that the staked asset retains value and that the validator set remains sufficiently distributed. A reader weighing them should ask which assumption they find more credible, because that is where the real disagreement lies.