Ethereum Upgrade Timeline

Ethereum launched as a proof-of-work chain: miners, not validators, secured it and collected the block reward (5 ETH per block at the start). There were no validators and no staking — the idea of staking ETH would not reach mainnet for another five years.

For a staking-economics record, Frontier is simply year zero: it marks where Ethereum begins, and everything this timeline actually tracks starts later, once the Beacon Chain goes live.

The Beacon Chain started as a separate proof-of-stake chain running in parallel to the still-live proof-of-work mainnet. It carried no user transactions — its only job was to prove that proof-of-stake consensus could run at scale before it took over.

The deposit contract had gone live weeks earlier (November 2020), and reaching genesis took a wave of early stakers: about 21,000 validators at 32 ETH each signed the first block. This is the moment staking rewards began accruing — even though staked ETH was locked, with no way to withdraw it yet.

London introduced EIP-1559, splitting every transaction fee into a base fee that is burned (removed from supply for good) and an optional priority fee, or tip. Under proof-of-work the tip went to miners — but the design quietly pre-built the fee economics that proof-of-stake proposers would inherit at the Merge.

For stakers, London is the reason a validator's execution-layer income is priority tips plus MEV, and never the burned base fee. That burn is also why heavy network usage can make ETH net-deflationary.

The Merge swapped Ethereum's consensus engine from proof-of-work to proof-of-stake in a single block, with no downtime: the execution layer simply began taking its consensus from the Beacon Chain. Issuance fell by roughly 88% overnight because the static block subsidy dropped to zero.

From here, validators earn on two layers at once — consensus-layer rewards (attestations, sync committees, block proposals) plus execution-layer priority fees and MEV. Bellatrix, the consensus-layer preparation fork, had activated nine days earlier at epoch 144896; it is frequently mislabeled as the Merge itself, which is why this timeline pins the real transition to epoch 146875.

Shapella combined Shanghai on the execution layer with Capella on the consensus layer, and finally let staked ETH leave the Beacon Chain — for about two and a half years, staking had been deposit-only.

It unlocked two flows: a full exit that returns a validator's whole balance once it exits, and an automatic skim that sweeps any balance above 32 ETH to the validator's 0x01 execution address every 9 to 12 days. Withdrawals turned staking from a one-way lockup into a liquid, cash-flowing position — and made a wallet-level record of staking receipts possible for the first time.

Dencun (Deneb on the consensus layer, Cancun on the execution layer) introduced blobs via EIP-4844, often called proto-danksharding: a cheap, temporary data lane that cut layer-2 transaction costs by roughly an order of magnitude.

The staking detail that matters is what did not change. Blob fees are burned, not paid to proposers, so a validator's reward is untouched by blob activity. Dencun is the clearest case of an upgrade reshaping Ethereum's economics for users while leaving the validator reward formula alone.

Pectra (Prague on the execution layer, Electra on the consensus layer) was the largest staking-focused upgrade since withdrawals. EIP-7251 raised the maximum effective balance from 32 to 2,048 ETH, letting an operator consolidate many 32-ETH validators into a single 0x02 compounding validator whose rewards auto-compound instead of being swept out.

EIP-7002 added execution-layer-triggered exits and partial withdrawals, so a validator's withdrawal address — not only its signing key — can now start an exit. Together they changed both the shape of validator balances and who holds the power to exit.

Fusaka (Fulu on the consensus layer, Osaka on the execution layer) delivered PeerDAS (EIP-7594), the next step in data-availability sampling: validators verify blob data by sampling a fraction of it rather than downloading all of it, raising blob throughput without raising the hardware bar.

Reward accounting is unchanged — the shift is operational (bandwidth and peering), not economic. Fusaka also introduced Blob-Parameter-Only sub-forks, BPO1 (December 2025) and BPO2 (January 2026), which raise blob targets on a set schedule without a full network upgrade.

Glamsterdam's headline is enshrined Proposer-Builder Separation (EIP-7732, ePBS): the block-building role that MEV-Boost relays perform off-protocol today moves into the protocol itself. An in-protocol builder commits to a payload and pays the proposer directly, and a new Payload-Timeliness Committee duty checks that the payload was published on time.

The consequence for stakers is that MEV-Boost relays become structurally optional, and MEV income is attributed on-chain rather than through a relay's own accounting — a meaningful change to how proposer income is measured and proven. Block-Level Access Lists (EIP-7928) also land, declaring up front which state a block will touch. Both EIPs are fork headliners but still formally in Draft status, and no mainnet activation slot is set.

Hegotá's confirmed headliner is FOCIL (EIP-7805, Fork-Choice-enforced Inclusion Lists): a committee of validators would each name transactions the next block must include, hardening censorship resistance — and adding a new validator duty with its own reward and penalty surface.

Native account abstraction, which would make smart-contract accounts first-class, is under consideration for this upgrade but not confirmed. As a further-out upgrade, treat Hegotá's scope as directional rather than settled.