Ethereum's recent upgrades explained: Dencun, Pectra, Fusaka and what comes next
What Ethereum's recent upgrades changed: cheaper layer 2 data with Dencun and Fusaka, smart accounts and staking changes in Pectra, and the roadmap ahead.
Figures checked against 6 official and primary sources on . See the sources · How we check
Key points
- Ethereum changes through planned network upgrades (hard forks) agreed in advance. Recent ones: Dencun (March 2024), Pectra (7 May 2025) and Fusaka (3 December 2025).
- Dencun and Fusaka focused on cheaper data for layer 2 networks, through "blobs" and then PeerDAS, which lets each node store only part of the blob data.
- Pectra let ordinary accounts run smart contract code (EIP-7702) and raised the maximum effective balance per validator to 2,048 ETH.
- ethereum.org lists Glamsterdam for Q4 2026 and Hegotá for 2027, and stresses that the roadmap is community-driven and subject to change.
In this guide
Ethereum doesn’t stand still. Every year or so, its developers ship a network upgrade, a planned hard fork that changes the rules in a way the whole network adopts together. Most users never notice them directly, but they’ve cut layer 2 fees, changed how staking works and added features to ordinary wallets.
This guide explains what the recent upgrades did, in plain English, and what’s planned next. For how forks work in general, see what is a hard fork?
How Ethereum upgrades work
ethereum.org explains that forks are changes to the protocol’s rules, typically coming from Ethereum Improvement Proposals (EIPs). Because no company owns Ethereum, every client must update its software; upgrades are agreed ahead of time so the network moves together. Each upgrade usually has a two-part name, one for the execution layer (where transactions run) and one for the consensus layer (proof of stake), combined into a nickname such as “Pectra”.
Who decides: Ethereum’s off-chain governance
Ethereum’s rules aren’t changed by a vote on the blockchain. ethereum.org describes an off-chain process involving many groups: ether holders, app users and developers, node operators, validators, EIP authors and the protocol developers (or “core developers”) who maintain the client software.
A change usually goes through these stages:
- Proposal: anyone can write an EIP setting out a new feature or process.
- Discussion: the author presents it to the protocol developers, typically on an AllCoreDevs call, after debate in public forums. It may be considered for a future upgrade, sent back for changes, or rejected.
- Iteration: the proposal is refined until it’s ready, then scheduled into a named upgrade.
- Testing: ethereum.org notes that network upgrades are usually activated on testnets before mainnet.
- Activation: client teams release software, node operators update, and the upgrade activates at a set time.
ethereum.org also explains why contentious changes are rare: no single group can force the others. Developers can decline to write code, node operators can decline to run new software, and users can decline to transact. So developers generally avoid proposals where the controversy outweighs the benefits. When that breaks down, a chain can split. See what is a hard fork?
Timeline
| Upgrade | Date | Headline change |
|---|---|---|
| The Merge (Paris) | 15 September 2022 | Switch from proof of work to proof of stake |
| Shapella | 12 April 2023 | Staking withdrawals enabled |
| Dencun | 13 March 2024 | “Blobs” for cheaper layer 2 data |
| Pectra | 7 May 2025 | Smart account features, staking changes |
| Fusaka | 3 December 2025 | PeerDAS to scale blob data |
| Glamsterdam | Q4 2026 (planned) | Proposer-builder separation, block-level access lists |
| Hegotá | 2027 (planned) | Inclusion lists, new transaction type |
See proof of work vs proof of stake for The Merge.
Dencun (March 2024): blobs
Dencun introduced EIP-4844, known as proto-danksharding. ethereum.org says it added data “blobs” that let rollups post data to the main network for a short period at much lower cost, resulting in significantly lower fees for layer 2 users. See what is a layer 2?
Two details explain why blobs are cheap:
- They’re temporary. Blob data can’t be read by smart contracts and is automatically deleted from nodes after 4,096 epochs, about 18 days. That’s long enough for anyone to check it, and keeps nodes from growing endlessly. Rollup operators, users or others can store the data off-chain.
- They have their own fee market, separate from ordinary gas. See Ethereum gas fees explained.
ethereum.org’s figures show how blob capacity has grown: the target was 3 blobs per block in Dencun, raised to 6 in Pectra.
Pectra (May 2025): smarter accounts and staking
ethereum.org describes Pectra as improvements for users, layer 2s, stakers and node operators:
- EIP-7702 lets a regular account (not a smart contract) execute code like a smart contract. ethereum.org says this unlocks features such as transaction batching, gas sponsorship (someone else paying your fees), alternative authentication, programmable spending controls and account recovery.
- EIP-7251 raised the maximum effective balance for a single validator to 2,048 ETH and introduced compounding validator accounts, improving capital efficiency.
- EIP-7002 lets a staker’s withdrawal address trigger validator exits or partial withdrawals, giving stakers more control and making node operators more accountable. See liquid staking explained.
- EIP-7691 increased blob throughput for layer 2s.
A safety note on EIP-7702: new features that let accounts run code also mean new kinds of permission requests. ethereum.org warns about “blind signing”, approving transactions you don’t understand. Only approve requests from apps you trust, and read what you’re signing. See token approvals and how to revoke them.
Fusaka (December 2025): PeerDAS
ethereum.org calls PeerDAS (peer-to-peer data availability sampling, EIP-7594) the headline of Fusaka:
- Before Fusaka, every full node had to store every blob. As blob use grew, that became too demanding.
- With PeerDAS, each full node stores only about an eighth of the blob data, enabling a theoretical eightfold scale-up, while the full data can be reconstructed from any 50% of it.
- The aim: more data capacity and smaller fees for layer 2s, without making nodes too expensive to run.
Fusaka also introduced Blob-Parameter-Only (BPO) forks, allowing blob capacity to be raised between major upgrades. ethereum.org explains that each increase is pre-configured in the client software released for Fusaka, so users and validators don’t need to update for each one, only for major upgrades. Fusaka also set a per-transaction gas limit cap of 16.7 million, and raised the default block gas limit to 60 million, increasing capacity on the main network.
Why testing matters: a lesson from Pectra
Upgrades go live on testnets first, and that’s where problems are meant to show up. ethereum.org notes that when Pectra went live on the Holesky testnet in early 2025, minor misconfigurations left that testnet unable to finalize. Fusaka responded with a new requirement (EIP-7910) for clients to report their fork configuration, so testing teams can check that every client is lined up before an upgrade moves from testnets to mainnet.
What’s next
ethereum.org’s roadmap lists:
- Glamsterdam (Q4 2026): enshrined proposer-builder separation, so validators can safely outsource block building, and block-level access lists to speed up syncing and enable parallel execution.
- Hegotá (2027): fork-choice enforced inclusion lists (FOCIL), which let a committee of validators require that valid transactions are included, strengthening censorship resistance; and frame transactions, which would make features like social recovery, spending limits and sponsored gas native to the protocol.
ethereum.org notes that Ethereum’s development is community-driven and subject to change, so dates can move. Longer term, preparing for quantum computers is also on the agenda. See quantum computing and crypto.
What it means for you
- Holding ETH: upgrades happen in the background; your balance doesn’t change.
- Using layer 2s: Dencun and Fusaka were designed to make layer 2 transactions cheaper.
- Staking: Pectra changed what’s possible; check what your provider supports.
- Scams: be wary of anyone who uses upgrade news to tell you to “migrate” or “swap” your coins. No upgrade requires you to send ETH anywhere or share your seed phrase. See fake crypto apps and phishing.
For the basics, see what is Ethereum? and Ethereum gas fees explained.
Frequently asked questions
Do I need to do anything when Ethereum upgrades?
Did these upgrades make Ethereum itself cheaper to use?
What is EIP-7702?
Who decides what goes into an upgrade?
Is blob data stored forever?
When is the next upgrade?
Sources
- The history of Ethereum (forks and upgrades) — ethereum.org (Ethereum documentation)
- Fusaka — ethereum.org (Ethereum documentation)
- Ethereum roadmap — ethereum.org (Ethereum documentation)
- Ethereum governance — ethereum.org (Ethereum documentation)
- Danksharding (proto-danksharding and blobs) — ethereum.org (Ethereum documentation)
- Smart contracts (blind signing) — ethereum.org (Ethereum documentation)
First published . Spotted an error or an outdated figure? Tell us and we will check it against the source.