The computer that cut its power bill by 99 percent

Changing the consensus rule of a live blockchain worth hundreds of billions, without stopping it, is one of the more impressive pieces of coordinated engineering of the last decade. It also settled an argument: proof of stake works at scale, and the energy objection to smart contract platforms is answered.

What actually changed

Before the Merge, miners spent electricity to compete for the right to add blocks. After it, validators lock up 32 ether each and are selected to propose and attest to blocks, losing part of their stake if they misbehave. Security moved from an external cost, electricity, to an internal one, capital at risk.

Energy consumption fell by more than 99 percent, which removed the most effective criticism the network faced with institutions and regulators. Issuance fell sharply too, because validators need far less compensation than miners did. Combined with a fee-burning mechanism introduced in 2021, ether's net supply has at times shrunk.

The scaling decision and its consequence

Ethereum chose not to scale the main chain much. Instead it optimised for being a settlement and data-availability layer, with transactions executed on layer 2 networks that post compressed results back to it. Later upgrades added dedicated cheap data space for exactly that purpose, and fees on layer 2 fell by an order of magnitude.

It worked. Most user activity now happens on layer 2. That success created the current open question, because if activity leaves the main chain, so does the fee revenue that burns supply and funds security. The counter-argument is that layer 2 networks pay for data and that demand grows with usage. Both sides of that argument are visible in the fee data and neither has won.

What staking became

Running a validator requires 32 ether, a reliable machine and an uptime discipline. Most holders do not do this. They stake through a service, which pools deposits and issues a token representing the staked position, or through an exchange. The result is that a large share of staked ether sits with a small number of operators, which is a centralisation the network's designers spend significant effort mitigating.

The yield, a few percent a year in ether terms, is compensation for locking capital and accepting penalty risk. Our staking calculator separates the two things people confuse: token count growth, which the protocol controls, and value, which the market does.

What runs on it

The honest inventory: stablecoins, decentralised exchanges, lending protocols, tokenized treasury funds and a large amount of speculative activity. Stablecoins and tokenized real-world assets are the categories institutions actually use, and both are growing. The non-financial applications that were promised in 2017 remain mostly unbuilt.

That is not a failure so much as a clarification. Ethereum turned out to be a settlement system for programmable money, which is a large thing to be.

The competitive position

Faster and cheaper chains exist and have real usage. Ethereum's defence is not speed; it is the depth of the ecosystem, the amount of value already there and a decade of running without a halt. The risk is that layer 2 fragmentation makes the user experience worse than a single fast chain, which is a design problem the ecosystem is actively working on and has not solved.

What to watch

Watch the share of total fees paid by layer 2 networks for data, which decides the long-run economics. Watch validator concentration among the largest staking providers. And watch whether tokenized funds and treasuries keep choosing this network as they scale, because institutional issuance is the clearest vote of confidence available.