Ethereum changed its consensus mechanism from proof of work to proof of stake on 15 September 2022. Ethereum.org states the change was never intended to lower gas fees and did not expand network capacity. It cut energy use by an estimated 99.95 per cent and reduced new ether issuance by 88.7 per cent.
Ethereum changed how it agrees on new blocks on 15 September 2022, replacing mining with staking. Ethereum's own documentation is direct about the result: the change "was never intended to lower gas fees", and fees did not fall. What did change was energy use and the rate at which new ether is created.
What was replaced, and what was left alone
A blockchain needs a rule for deciding which version of history is the real one. Before September 2022, Ethereum used proof of work: computers competed to solve a puzzle, and the winner proposed the next block and received newly created ether. After that date it uses proof of stake: participants lock up ether as a deposit, one is selected to propose the block, and dishonest behaviour costs part of the deposit.
That is the whole of it. Ethereum.org calls the event "a change of consensus mechanism, not an expansion of network capacity". The rules about how much work a block can contain, how transactions are priced, and what a smart contract can do were untouched. Account balances, addresses and applications carried on exactly as before.
Energy use is where the effect was enormous. Removing the puzzle removed the reason to run warehouses of hardware, and ethereum.org puts the drop in energy consumption at an estimated 99.95 per cent.
Why fees did not fall
To see why, it helps to know what a fee on Ethereum actually pays for.
Every operation on the network costs gas, a unit measuring computational work. Sending ether costs a small fixed amount of gas; running a complex contract costs far more. The price you pay per unit of gas is quoted in gwei, a unit equal to one billionth of an ether. Your total fee is gas used multiplied by the gas price.
Each block can hold only a limited quantity of gas. When more people want their transactions included than the block can fit, they compete on price, and the gas price rises. The fee is therefore a function of demand for a fixed amount of space.
The consensus mechanism has nothing to do with that limit. Swapping miners for stakers changed who proposes the block and how they are paid, and left the size of the block alone. The same amount of space, contested by the same demand, produces the same fees.
Where the cheaper transactions actually came from
Scaling moved off the main chain deliberately. Layer 2 networks are separate networks that process transactions in bulk and post compressed results back to Ethereum for settlement. Users transact on the layer 2 network at a fraction of the cost, while the main chain stays small, conservative and easy for anyone to verify.
The cost of that arrangement fell sharply in March 2024, when the Dencun upgrade introduced blob transactions through EIP-4844. An EIP is an Ethereum Improvement Proposal, the format in which changes to the protocol are specified. Blobs gave layer 2 networks a cheap, temporary place to post their data instead of writing it permanently into ordinary transaction records, which cut what they pay to settle.
So the fee reductions people associate with the merge came from a different upgrade, eighteen months later, on a different part of the system.
How new ether is created now
Ether comes into existence as a reward for securing the network. Before the change there were two sources; now there is one.
| Issuance | Before 15 September 2022 | After |
|---|---|---|
| Execution layer (mining reward) | 88.7 per cent of all new ether | None |
| Consensus layer (staking reward) | The remaining share | The only source |
| Total, annualised | 4.61 per cent | 0.52 per cent |
Figures as published by ethereum.org. The mining reward went to zero, which cut total issuance by 88.7 per cent in a single step, a reduction of roughly the size of three Bitcoin halvings arriving at once.
Is ether deflationary?
Supply can fall over a period, and whether it does depends on network activity rather than on the calendar.
The mechanism came in a year before the consensus change. The London upgrade of August 2021 introduced EIP-1559, which split the transaction fee in two. A base fee, set by the protocol according to how full recent blocks were, is destroyed. A priority fee, chosen by the sender, goes to whoever proposes the block. A base fee paid to a miner would have circulated back into the market; a base fee that is destroyed leaves the supply permanently.
Two flows therefore run in opposite directions: issuance adds ether, the burned base fee removes it. When the burn over a period is larger than issuance over the same period, total supply falls for that period. When the network is quiet, issuance wins and supply grows. Ethereum.org works the balance point out as a daily average gas price of at least 16 gwei, calculated against roughly 1,700 ether a day of validator issuance at about 14 million ether staked. Those were the conditions when that page was written, and well below the amount staked today, so the break-even moves with the amount staked. That is the point: the net change in supply describes how busy the network has been, not a fixed property of the protocol.
Four things the change did not do
- It did not create a new token. There was never an "ETH2" coin. Ether was not swapped or reissued, and holders had nothing to do.
- It did not unlock staked ether. Withdrawals arrived separately, in the Shanghai and Capella upgrade of April 2023.
- It did not make the main chain meaningfully faster. Ethereum.org describes layer 1 transaction speed as mostly the same. Time is now divided into fixed 12 second slots, each one a chance for a chosen validator to publish a block, rather than blocks arriving at a variable average interval.
- It did not change any application. Contracts, addresses and balances carried across untouched, which was the design goal.
Where to look next
- Ethereum price in Australian dollars - the live figure, struck and sourced.
- Crypto in AUD - the other coins we track against the Australian dollar.
- Glossary - plain definitions of the terms used above.
- Guides - the rest of the explainers.
Sources
- The Merge - ethereum.org, on the 15 September 2022 execution, the intent, energy use and transaction speed.
- How The Merge impacted ETH supply - ethereum.org, on the 4.61 per cent and 0.52 per cent issuance rates, the 88.7 per cent reduction and the 16 gwei balance point.
- Gas and fees - ethereum.org, on gas units, gwei and the block gas limit.
- EIP-1559 - the burned base fee and the priority fee, introduced in the London upgrade.
- EIP-4844 and Dencun - blob transactions and the March 2024 upgrade.
- Staking withdrawals - ethereum.org, on the April 2023 Shanghai and Capella upgrade.
General educational information about how the Ethereum protocol works. Not advice, not a forecast, and not a view on any asset. Tradergy keeps no account and collects no personal data. Reviewed by the Tradergy editorial team, September 2026.
Common questions
Did the Ethereum merge lower gas fees?
No. Ethereum's own documentation states the change was never intended to lower gas fees, because it replaced the way blocks are agreed rather than increasing how much a block can hold. Fees are set by demand for limited space in a block, and that limit did not change on 15 September 2022.
What did the Ethereum merge actually do?
It replaced mining with staking as the way the network agrees on the next block. Energy consumption fell by an estimated 99.95 per cent, new ether created by miners stopped entirely, and total annual issuance dropped from 4.61 per cent to 0.52 per cent. Nothing changed for people holding ether.
Is Ethereum deflationary?
Not by design. Since August 2021 the base portion of every transaction fee is destroyed rather than paid to anyone. When the amount destroyed over a period exceeds the amount newly issued, total supply falls for that period. Ethereum.org works the balance point out at a daily average gas price of at least 16 gwei, on the network conditions shortly after the merge. That level moves as the amount of staked ether changes.
Did the merge make Ethereum faster?
Barely. Ethereum.org describes transaction speed on the main chain as mostly the same. Time is now divided into fixed 12 second slots, each one a chance for a chosen validator to publish a block, rather than blocks arriving at a variable average interval. That is a small change in consistency rather than in throughput.
Did ether holders have to do anything when the merge happened?
No. Ether was not swapped, replaced or reissued, and there was never a second token. Any service asking holders to migrate ether to a new coin at the time of the merge was almost certainly a scam.