Ethereum

Changelog.

• v0.3.0 (2026-10-14; replaces v0.2.1, publish hash 9c91b597925b021d46b4cb234daede392e83f03d31e190bbfb7d942ce274124b): plain-English rewrite for learners. Adds an "At a glance" block, plain explanations of terms specific to this subject, labeled analogies, and "why it mattered" notes for each upgrade. Sources unchanged. No claim changes.
• v0.2.1 (2026-10-14; replaces v0.2, publish hash c40c35228549cf32240a0a228e6d49fed3b0f4bdbf6f0ce4317bfc2309246476): the page record now shows the published status, version, publish date and proof. Body wording: the policy line now names the Learning Hub; so does the line about the other dives. No claim changes.
• v0.2 (2026-10-06; replaces v0.1, sha256 4904395ee43eafb7d1df2ea9cff302dcc003b9aea96c3db3ad28bfd44352e6ea): sourced the smart-contract definition to ethereum.org's introduction to smart contracts, using its wording; described EIP-8038 as updating the cost of accessing state, not only reading it; applied "most" to flagged issues rather than contracts and named the outreach as aimed at the most-affected builders; removed the client-share percentages (a usage figure) and kept the 33% guideline and the page's out-of-date warning; removed an unsupported statement that no newer primary measurement exists; updated four redirecting ethereum.org source URLs to their final addresses; described EIP-8061 as higher exit and consolidation churn and added the ethereum.org Glamsterdam page as the source for faster exits; updated the Learning Hub label.

At a glance

• Ethereum has secured itself with proof-of-stake instead of mining since The Merge on September 15, 2022 [1].
• Rule changes are written as Ethereum Improvement Proposals and adopted when node operators run new software, with no token vote [5].
• Every transaction pays a protocol-set base fee, which is burned, plus an optional tip [3].
• Since Dencun in 2024, layer-2 networks can post data to Ethereum as cheaper, temporary "blobs" [1].
• The next upgrade, Glamsterdam, has no mainnet date; Ethereum.org expects Q4 2026, "date not yet confirmed" [8][9].

1. What it is

Ethereum and ether

Ethereum is a public blockchain that runs smart contracts, which ethereum.org defines as programs that run on the Ethereum blockchain [12]. Its native currency is ether (ETH), which pays for every transaction [3]. Vitalik Buterin published the introductory white paper in 2013, and the network went live as "Frontier" on July 30, 2015 [1].

How the rules change

Nobody owns Ethereum, and its rules change through an offchain process rather than a token vote [5]. Proposals are written up as Ethereum Improvement Proposals (EIPs), implemented by independent client teams, and bundled into network upgrades [5]. Node operators then choose whether to run the new software [5].

Analogy, not the mechanism itself: it is like a language with no central academy: new rules spread only if enough people start using them.

The Merge

Since September 15, 2022, Ethereum has secured itself with proof-of-stake instead of mining [1]. That switch is known as The Merge [1].

A base for layer 2s

Ethereum is also the settlement layer for many layer-2 (L2) networks, which execute transactions elsewhere and post their data back to Ethereum [4]. Two of the Learning Hub's other dives, Arbitrum and Robinhood Chain, are L2s of this kind.

2. How it works

Execution and consensus clients

An Ethereum node runs two pieces of software [2]. The execution client handles transactions and the Ethereum Virtual Machine (EVM) [2]. The consensus client handles proof-of-stake [2]. A validator also runs a third piece, the validator client [2]. Since The Merge, the two layers upgrade together, which is why upgrades carry paired names such as "Pectra" (Prague plus Electra) [1].

Analogy, not the mechanism itself: the execution client is like a store's cash registers, and the consensus client is like the head office that agrees on the official daily books.

Many client teams

Several teams write these clients independently [5]. At the execution layer they include go-ethereum, Nethermind, Besu, Erigon and Reth [5]. At the consensus layer they include Prysm, Lighthouse, Nimbus, Teku, Lodestar and Grandine [5].

Slots, epochs and attestations

To become a validator, a user deposits 32 ETH into the deposit contract and joins an activation queue [2]. Time is divided into 12-second slots, and 32 slots make an epoch [2]. In each slot one validator is chosen at random to propose a block, and a committee of validators votes on it [2]. These votes are called attestations [2].

Checkpoints and the inactivity leak

A block becomes final through checkpoints [2]. When validators holding at least two-thirds of all staked ETH vote for a pair of checkpoints, the earlier one is finalized [2]. Reverting a finalized block would require an attacker to lose at least one-third of all staked ETH [2]. If the chain fails to finalize for more than four epochs, an "inactivity leak" drains stake from validators who are not voting with the majority, so the chain can recover [2].

Slashing penalties

Validators that break the rules are slashed [2]. The two slashable offenses are proposing two blocks in one slot and making contradictory attestations [2]. The penalty grows with the number of validators slashed around the same time, from under 0.1% of stake for an isolated case up to the whole stake in a mass event [2].

Fees and the base-fee burn

Every operation costs gas, and gas is paid in ETH [3]. A transaction pays a base fee set by the protocol plus an optional priority fee, or tip, for the validator [3]. The base fee is burned [3]. It moves up or down by at most 12.5% per block, depending on whether the previous block was above or below its target size, which is half the gas limit [3]. This fee design came from EIP-1559, part of the London upgrade of August 5, 2021 [1].

Analogy, not the mechanism itself: the base fee works like surge pricing that adjusts a little after each busy or quiet period, except that the base charge is destroyed rather than paid to anyone; the optional tip goes to the validator.

Blobs

L2 rollups post their transaction data to Ethereum so that the data inherits Ethereum's security [4]. Since the Dencun upgrade of March 13, 2024, they can post it as "blobs", a cheaper data type that is kept only for a limited time [1]. This change is EIP-4844, also called proto-danksharding [1].

Analogy, not the mechanism itself: a blob is like a parcel held at a pickup counter for a set time rather than filed in the permanent archive.

PeerDAS

The Fusaka upgrade of December 3, 2025 added PeerDAS [1][7]. Under PeerDAS, a full node no longer downloads every blob [7]. Each node holds a share of the data, and the network can rebuild any blob from half of the pieces [7]. Fusaka also lets clients raise blob capacity through small "blob-parameter-only" forks between major upgrades [7].

3. What changed

Timeline

Pectra (May 7, 2025)

What changed. EIP-7251 raised the maximum effective balance of a single validator to 2,048 ETH [1]. EIP-7002 let a withdrawal address trigger exits directly [1]. EIP-7702 lets an ordinary account run code like a smart contract, which allows features such as batched transactions and gas sponsorship [1].

Why it mattered. Ordinary accounts gained features such as batched transactions and gas sponsorship, and withdrawal addresses can now trigger exits directly [1].

Fusaka (Dec. 3, 2025)

What changed. Besides PeerDAS, Fusaka capped any single transaction at 16,777,216 gas (2^24) [7]. It set a 10 MiB ceiling on block size [7]. It also added a precompile for the secp256r1 curve, which lets wallets use device passkeys [7]. A precompile is a built-in function contracts can call. Client teams shipped a 60 million default gas limit with it [7]. The Ethereum Foundation says Fusaka raised the mainnet gas limit "on the path to 200M and beyond" [10].

Why it mattered. Nodes no longer download every blob, and wallets can use device passkeys [7].

Glamsterdam: what is in it

Glamsterdam combines the Amsterdam execution upgrade with the Gloas consensus upgrade [9]. Its two headline changes are [8][9]:

• Enshrined proposer-builder separation (ePBS, EIP-7732). Today, validators who outsource block building rely on third-party relays outside the protocol [8]. ePBS moves the hand-off and the builder's payment into the protocol [8][9]. Ethereum.org says it widens the window for spreading block data from about 2 seconds to about 9 seconds [8].
• Block-level access lists (EIP-7928). Each block records which accounts and storage slots it touches [9]. Clients can then read state and validate transactions in parallel [9].

Analogy, not the mechanism itself: an access list is like a recipe that names every ingredient up front, so several cooks can fetch them at once.

Why it matters. Ethereum.org and the Foundation describe in-protocol builder payments and parallel validation; none of it is on mainnet yet [8][9].

Glamsterdam: gas repricing

Glamsterdam also reprices gas [9]. EIP-8037 raises the cost of creating new state, and EIP-8038 updates the cost of accessing it [9][11]. Ethereum.org says EIP-8037 targets state growth of 120 GiB a year [8]. The Foundation warns that a small set of existing contracts may break or degrade under the new schedule without updates [11]. Other scheduled changes include ETH transfers emitting a log (EIP-7708) and higher exit and consolidation churn, which ethereum.org says lets stakers exit faster (EIP-8061) [8][9].

Why it matters. Builders of the flagged contracts may need updates [11].

Glamsterdam: dates

Glamsterdam was scheduled to activate on the Sepolia testnet on October 6, 2026 at 13:53:36 UTC, 9:53 a.m. ET [9]. The Foundation's announcement says mainnet and Hoodi dates "have not yet been decided" [9]. Ethereum.org lists mainnet as expected in Q4 2026, "date not yet confirmed" [8].

4. What is open

Mainnet date and final scope

No mainnet date has been set [9]. Ethereum.org says scope "is frozen but can still change before mainnet, as the meta EIP remains in draft" [8]. The Foundation calls a 200 million gas limit a "credible post-Glamsterdam target" [10]. On Sepolia, at least two consensus clients default to 60 million gas after the fork unless validators set 200 million themselves [9].

Contract breakage from repricing

The Foundation found that replaying past mainnet transactions under the new gas schedule flags a small set of contracts that may break or degrade [11]. Most flagged issues can be fixed with a higher gas limit, and outreach to the most-affected builders is under way [11]. How many affected contracts will be updated before mainnet is not known.

Relays until ePBS ships

Until ePBS is live, proposers who outsource block building depend on off-protocol relays and trust between parties [8]. Even after ePBS, builders and proposers "might still choose to use off-protocol relays" for features the protocol does not cover [8].

Client concentration

A bug in a consensus client run by more than a third of nodes could stop finality [6]. A bug in a client with a two-thirds share could finalize an invalid chain [6]. Ethereum.org says that, ideally, no consensus client would reach a 33% share of nodes, and warns that its own client-share snapshot may be out of date [6].

Offchain governance

Upgrades depend on rough consensus among client teams, node operators and users, with no binding vote [5]. That makes contested changes slow by design [5]. It also means "who decides" has no single answer.

Staff changes at the Foundation

In May 2026 the Foundation's Protocol cluster said two coordinators were leaving and a third was going on sabbatical, during preparation for Glamsterdam [10]. The effect on delivery timelines, if any, is not stated.

What this page leaves out

By Learning Hub policy, this dive carries no price, market value, staking-yield or usage-volume figures. Those go stale, and News covers them.

Not financial advice.