All 22 Ethereum Hard Forks Explained: From Frontier in 2015 to Fusaka in 2025

Sure, Ethereum brings a lot of innovation, but so do other networks. What sets it apart is the extensive testing done to design the best possible developer blockchain. The Merge might sound like a big step forward, but it’s really the result of countless smaller steps. Thousands of proposals over the years, and one fork after another.

Here’s the evolution of Ethereum, starting from beta until the post-Merge version, as well as its direction for the upcoming Ethereum 2.0.

Stick around until the end to learn about one of the newest exciting developments of the Ethereum hard fork genre!

What is an Ethereum Hard Fork?

An Ethereum hard fork is a permanent, backward-incompatible upgrade to the Ethereum protocol. Every node on the network must update its software to follow the new rules. Nodes that do not update stay on the old chain. Ethereum is a public blockchain: decentralized, autonomous, and not owned by any single team or company.

A good example of this is the Internet. Many companies provide Internet connections, website software, and online services. But there’s no server that singlehandedly controls the Internet. If you want to shut down the Internet or change its software, you’d need to change every single device node.

Ethereum follows a similar logic. An ethereum hard fork is an important software update that divides the blockchain in two. Those who don’t update the blockchain software stay in the old version while those who do move to the newly forked blockchain.

Like the Internet, new blockchains need enough nodes to be secure and decentralized. Once enough users update, the Ethereum fork goes live. Developer teams fork the blockchain either to upgrade Ethereum or solve vulnerabilities (A bit like updating from Windows 10 to 11. The same happened on the transition from Web1 to Web2, and currently to Web3).

Ideally, every device would abandon the old version and move to the new one. That never happens because no one can force every node to update, and not everyone will want to. Hence why Ethereum forks temporarily start with fewer nodes (e.g., +75% of the initial number), which lowers its security. But doing nothing isn’t safe either, as people eventually switch from outdated versions, which will lose nodes and security.

The “unforked” version of Ethereum is Ethereum Classic (ETC). It’s not nearly as successful for the mentioned reasons: it had lower adoption and increased vulnerabilities. Also note that you can’t stake ETC, as it’s proof-of-work (PoW) and doesn’t inherit Ethereum’s updates.

Hard Fork vs. Soft Fork

Not every Ethereum upgrade is a hard fork. A soft fork is backward-compatible. Old nodes can still validate blocks produced under the new rules, even without updating. A hard fork is not compatible in either direction. Old nodes reject new blocks. New nodes reject old blocks. The chain splits unless the entire active network upgrades together.

Every major Ethereum upgrade since Homestead has been a coordinated hard fork. The community agreed on the upgrade, updated their nodes, and moved forward together. The one exception that created a lasting split was The DAO Fork in 2016, which left behind Ethereum Classic.

List of Ethereum Hard Forks

Ethereum isn’t a blue-chip cryptocurrency for nothing. Few investors know about the many forks that made the blockchain what it is today. It’s not necessarily the most innovative, but definitely, the one that has tested the most, with thousands of proposals coming from the developer community. Here’s its evolution, from the beta version until post-Merge Ethereum:

All 22 Ethereum Hard Forks at a Glance

👉 Quick takeaway: Ethereum has shipped 22 protocol upgrades since its 2015 launch. The most consequential are The DAO Fork (2016), which reversed a hack; London (2021), which introduced fee burning via EIP-1559; The Merge (2022), which ended proof-of-work; and Dencun (2024), which introduced blob data to cut Layer 2 fees by 80%+.

1st: Frontier

Frontier is the first ever version of Ethereum launched on July 15th of 2015. It started as a testnet with basic smart contracts aimed only at experienced developers. The goal was simple: launch Ethereum, observe, and troubleshoot.

Once the foundations were set, new updates followed just 1 month after. The actual cryptocurrency didn’t appear on most exchanges until Frontier Thawing.

2nd: Frontier Thawing

Frontier Thawing was live after reaching the 200,000th block on September 6th of 2015. It’s also called Frontier because it’s mostly bug fixes and tweaks from the previous version:

• Moving gas limit from 5K to 3M to guarantee that transactions go through
• Setting a minimum gas limit of 21K, which is still the default used today

Ethereum 2.0. goes all the way back to Frontier Thawing. It introduced the “difficulty bomb,” which one year later, would ease a transition from Proof-of-Work (PoW) to Proof-of-Stake (PoS).

Ethereum had a 60M circulating supply and stayed around $1 until 2016.

3rd: Homestead

Homestead was the first fork created after the 1,150,000th block on March 14th of 2016. It’s the second major update that applied three Ethereum Improvement Proposals (EIPs) from the developer community. It expanded with smart contract functions on the Ethereum Virtual Machine (EVM) and forward compatibility (which makes Ethereum easier to upgrade).

But those updates weren’t enough to correct all code vulnerabilities. The DAO was the first high-profile project to launch in April and fund-raise $150M in ETH. Two months later, it was exploited through a reentrancy vulnerability and 3.6M ETH, worth roughly $60M at the time, was drained.

Ethereum was at $12.50 and 75M of circulating token supply until the next fork.

4th: The DAO Fork

In response to the incident, Ethereum introduced this fork to move funds from the old smart contract to a new one. From July 20th of 2016 and after block 1,920,000, anyone who lost funds could exchange 100 DAO tokens for 1 ETH. But not everyone agreed with this change.

15% of people who voted against the fork perceived it as network manipulation. They wanted an immutable Ethereum, so they created Ethereum Classic (ETC) separated from Ethereum (ETH). The founders and dev team moved to this second new blockchain.

5th: Tangerine Whistle

On October 18th of 2016, Tangerine Whistle came in response to the previous month’s DoS spam attacks. Proposals EIP-150 and EIP-158 led to this hard fork, effective from block 2,463,000. At that time, Ethereum averaged $12 and crossed 84M in circulating supply.

Tangerine Whistle wasn’t about new features but troubleshooting urgent issues. While it did mitigate those cyber-attacks (by increasing gas fees for IO-heavy operations), major upgrades didn’t happen until Byzantium.

6th: Spurious Dragon

Spurious Dragon is an expansion of Tangerine Whistle and a second response to DoS attacks. It went live on November 22th of 2016 after selling 2,675,000 with a token price of ~$10 and a supply of 85M. Four Ethereum proposals improved the protocol’s code size limit, transaction cost, state size, and security.

After Spurious Dragon, Ethereum’s development ran smoothly until the next upgrade a year later.

7th: Byzantium

Ethereum forked to Byzantium after block 4,370,000 on October 16, 2017. By this point, Frontier Thawing’s difficulty bomb was starting to make itself felt. PoS Ethereum was still years away, so Byzantium delayed the bomb by roughly 18 months and reduced block mining rewards from 5 ETH to 3 ETH. The nine EIPs in this fork also improved cryptographic primitives and laid groundwork for Layer 2 development.

8th: Constantinople/ St. Petersburg

The Constantinople would occur on February 28 of 2019 after block 7,280,000. But after some discussion, the dev community decided to postpone it to correct potential vulnerabilities. Essentially, the St. Petersburg fork would draw back one proposal (EIP-1283) and replace Constantinople’s fork on the same date. It’s two forks at once, except only St. Petersburg was effective.

There were 104M ETH tokens, and the price averaged $135 until the next update announcement.

Interestingly enough, the blockchain community that once remained inflexible finally forked in September 2019: Ethereum Classic. ETC underwent the Atlantis fork, which mirrored the improvements of Spurious Dragon and Byzantium. Yet, these upgrades weren’t enough to secure Ethereum Classic as much as Ethereum, and cyber-attacks continued until 2020.

9th: Istanbul

Istanbul was scheduled for block 9,069,000 and predicted for December 8th of 2019. It would be the largest upgrade until October 2020 (Staking contracts). Istanbul contributed to Ethereum by improving DoS-attack resilience, Layer-2 solution performance, EVM gas cost optimization, and interoperability with the ZCash network.

Ethereum had a 108M token supply and a $165 average price.

10th: Muir Glacier

Muir Glacier occurred at block 9,200,000 on January 1st of 2020. It was a response to the difficulty bomb, which threatened the usability of Ethereum. This update delayed the “bomb” for a year once again to keep block times between 10 and 20s.

This wasn’t the last time they would delay, but hopefully, it would be enough to standardize proof-of-stake. Ethereum had a 108M circulating supply and a $125 average price.

11th: Staking Deposit Contract

Ethereum implemented staking after block 11,052,984 and October 14th of 2020. It was a quick yet essential update to speed up the upcoming Beacon Chain (ETH2). Users could send amounts to this contract and start staking as soon as the new chain launched.

Ethereum had a 108M token supply, and prices moved above $370.

12th: Beacon Chain Genesis

Beacon Chain Genesis required a staking minimum to secure the blockchain. This was achieved by November 2020, which led to the first block validation on December 1st. You could finally stake on ETH2 Genesis, even though Ethereum was still mining-based.

Ethereum crossed $600 in price and 112M in circulating supply. This update happened while DeFi was booming, so Ethereum saw a rapid increase in traffic and gas fees.

13th: Berlin

The Berlin update was scheduled for block 12,244,000 and predicted for April 15th of 2021. Ethereum teams implemented 4 EIPs that would improve gas costs and add new transaction types. Ethereum had a 114M token supply and $2400 average price.

14th: London

The London fork activated at block 12,965,000 on August 5, 2021. It introduced five EIPs. The most consequential was EIP-1559, which replaced the old first-price auction fee model with a base fee that adjusts automatically block by block. That base fee is burned, not paid to miners. Any tip above the base fee goes to the miner. This made transaction costs more predictable and introduced a deflationary mechanism into ETH’s token economics that remains active today.

15th: Altair

The Altair update followed on October 27th of 2021 after epoch 74,240 (around block 13,500,000). Back then, Ethereum was working on a future “merge,” as it was still using PoW and PoS at once. The difficulty bomb would ease the transition to PoS, and Altair was the last big update needed to set an official Merge date.

Originally, Altair should have been released with London (just like Constantinople had with St. Petersburg). But mining difficulty ramped up too early, which postponed it until October. Altair was the first Beacon Chain upgrade, making it more reliable by increasing slashing penalties on inactive validators.

Ethereum had a 117M token supply and a $4,000 average price.

16th: Arrow Glacier

Arrow Glacier was scheduled for block 13,773,000 and activated on December 9, 2021.. Like previous “glacier” updates, Arrow Glacier pushed the difficulty bomb further and mimicked the same improvements made on Byzantium and London. Ethereum had a 119M token supply and a $4,100 average price.

By now, we can deduct a few observations that will make future forks easier to understand:

• As for names, every glacier-related name (including Frontier Thawing) is associated with the difficulty bomb, also known as Ice Age. It’s called this way because it exponentially increases difficulty and block time. Its mathematical function makes the difficulty unnoticeable until a year later (give or take). Ideally, PoS Ethereum should release before the Ice Age bomb goes off, making PoW Ethereum unusable.
• After Istanbul, most forks use the name of the city that hosted the annual DevCon event.
• Until the Merge, the official Ethereum history page mixes updates from the Mainnet and the Beacon Chain. Beacon Chain doesn’t follow previous conditions, and it also uses another block counting system (epoch instead of block number).

Both Ethereum chains would remain separated for another 3 years until the Paris fork.

17th: Gray Glacier

The Gray Glacier update followed after block 15,050,000, expected by June 30th of 2022. As typical, it delayed the difficulty bomb by several months, although less than usual. It’s not a major upgrade, but an urgent one in response to a premature difficulty bump.

ETH prices fell to $1,000 while circulating supply increased past 121M.

18th: Bellatrix

Bellatrix took place on September 6th of 2022 after epoch 144,896 (around block 15,485,000). It was the second Beacon Chain upgrade and the first phase of the Merge. It’s often omitted for the second phase, as Bellatrix was just a preparation upgrade. It includes changes in the validator penalty values, expanded settings for the transition, and Merge instructions for ETH developers.

Ethereum had a 122M token supply and a $1,500 average price.

19th: Paris (The Merge)

Paris (The Merge) is the most anticipated Ethereum update. After many bomb delays and security revisions, it’s becoming official on September 15th of 2022 (after block 15,537,394). The Ice Age will come to the PoW chain, and Ethereum will finally integrate it as a new PoS network.

A common misconception is that The Merge would bring massive upgrades. That’s not the case, as the teams had already made those updates since early 2022. What we’ve seen until September is a test period to find out how sustainable and secure the upgrades were. And because there were no incidents, the teams considered it secure to finally release The Merge.

Ethereum is now on its way to becoming way more efficient thanks to proof-of-stake. As long as it’s not flooded with traffic, developers should soon notice better speeds and costs on their dApps. For many, it’s the most important upgrade since Homestead.

After The Merge, Ethereum had a 122M token supply and a peak near $2,000. The Merge cut ETH issuance by roughly 90% by eliminating proof-of-work mining rewards entirely. It was one phase in a long sequence. Three more major forks followed: Shanghai/Shapella in April 2023, Dencun in March 2024, and Fusaka in December 2025. Each one built on the last.

What Comes After The Merge

The Merge was not the finish line. It was the starting point for a new kind of roadmap.

Three major upgrades have activated since September 2022. Each one targeted a different problem. Here is what happened and why it matters.

20th: Shanghai / Shapella (April 12, 2023)

Shanghai paired with the Capella upgrade on the consensus layer to form what is commonly called Shapella. The combined fork did one thing that stakers had been waiting years for: it enabled validator withdrawals. For the first time, ETH locked in staking contracts since 2020 could be withdrawn. Validator exits and partial withdrawals both became possible from this point forward.

21st: Dencun (March 13, 2024)

Dencun combined the Cancun upgrade on the execution layer with the Deneb upgrade on the consensus layer. The headline change was EIP-4844, which introduced proto-danksharding. This added a new data type called a blob. Blobs give Layer 2 rollups a cheaper place to post transaction data, separate from the main calldata space. The practical result was a steep drop in fees for rollup users. Arbitrum, Optimism, and Base all saw transaction costs fall significantly in the weeks after activation.

The upgrade also included EIP-1153 (transient storage opcodes) and EIP-5656 (memory copy instruction), both aimed at reducing gas costs for smart contract developers.

22nd: Fusaka (December 3, 2025)

Fusaka activated on mainnet on December 3, 2025. It extended the blob-based data availability work started in Dencun. Alongside Fusaka, the Ethereum Foundation confirmed a series of Blob Parameter-Only forks, known as BPO forks, designed to fine-tune blob parameters without requiring a full protocol upgrade. BPO#1 deployed shortly after Fusaka, with additional BPO forks planned into early 2026.

The Current Roadmap

Ethereum’s scaling strategy no longer centers on splitting into 64 shards or on eWASM. Both of those earlier goals were superseded. The current approach uses blob-based data availability and Layer 2 rollups as the primary scaling path. The Surge phase of the roadmap focuses on this combination. The Verge, Purge, and Splurge phases address statelessness, state size reduction, and miscellaneous improvements respectively. No single upgrade delivers all of this. It arrives fork by fork.

What Do These Forks Mean for You?

If you hold ETH, use a Layer 2 network, or run a validator node, each fork changes something concrete. Here is a plain summary of who each major recent upgrade affected.

For ETH Holders

The Merge reduced ETH issuance by roughly 90% by eliminating mining rewards. London (EIP-1559) introduced a base fee burn, which has made ETH deflationary during periods of high network activity. Dencun did not directly change ETH issuance but reduced costs for rollup users, which can drive more on-chain activity.

For Layer 2 Users

Dencun is the most impactful recent fork for everyday users of rollups like Arbitrum, Optimism, and Base. EIP-4844 gave rollups a dedicated blob space for posting data. Before Dencun, rollups posted data as calldata, which competed with regular transactions for block space. After Dencun, blob data is cheaper and does not compete in the same way. Transaction fees on major rollups dropped significantly within days of the March 13, 2024 activation.

For Validators and Node Operators

Shapella (April 2023) let validators exit and withdraw staked ETH for the first time since the Beacon Chain launched in December 2020. Fusaka (December 2025) and the BPO fork series that followed require node operators to update client software each time a BPO fork activates. Missing a BPO update means falling off the canonical chain. Operators should monitor the Ethereum Foundation blog and client team release notes for each BPO activation date.

Decision Framework: Should You Do Anything Before the Next Fork?

• Running a node or validator: Yes. Update your client software before each scheduled fork activation. Check the Ethereum Foundation blog for exact slot numbers.
• Using a Layer 2: No action needed. Fee changes happen automatically after each upgrade.
• Holding ETH on an exchange: No action needed. Custodians handle client updates on your behalf.
• Running a dApp or smart contract: Review each fork’s EIP list for changes to gas costs, opcodes, or transaction types that may affect your contracts.

Frequently Asked Questions

What is the difference between a hard fork and a soft fork on Ethereum?

A hard fork requires all nodes to upgrade. Nodes that do not upgrade cannot follow the new chain. A soft fork is backward-compatible, meaning old nodes can still validate new blocks even without upgrading. Nearly all major Ethereum upgrades since Homestead have been implemented as hard forks.

Does every Ethereum hard fork create a new coin?

No. The only Ethereum hard fork that produced a lasting separate chain with a separate coin was The DAO Fork in 2016, which created Ethereum Classic (ETC). All subsequent forks were coordinated upgrades where the entire active community moved to the new chain.

What is proto-danksharding and how is it different from full danksharding?

Proto-danksharding, introduced in Dencun via EIP-4844, added a new blob-carrying transaction type. Blobs carry data that rollups need but that the Ethereum execution layer does not process directly. Full danksharding would expand this further with data availability sampling, allowing nodes to verify blob availability without downloading all blob data. Full danksharding is a future goal and is not yet scheduled for a specific fork.

What are BPO forks?

BPO stands for Blob Parameter-Only. These are small, targeted forks that adjust blob-related parameters, such as the target and maximum number of blobs per block, without bundling in other EIPs. They allow the Ethereum network to tune blob capacity incrementally. Fusaka activated the first BPO series beginning in December 2025.

How often does Ethereum hard fork?

The cadence has varied significantly. Between 2015 and 2022, Ethereum averaged roughly two to three forks per year. Since The Merge, the pace has slowed to approximately one major fork per year, with smaller BPO forks filling the gaps between major upgrades.