Dencun is the upgrade every subsequent Ethereum release has been built on top of. It opened the network's "Surge" phase, made rollups economically viable, and set the template for the blob-scaling work that Pectra and Fusaka continued.

It also created the value-accrual debate that still defines ETH in 2026. Here is what Dencun shipped, how the network responded across two years of live data, and what the upgrade ultimately cost as well as delivered. 👇

What is the Ethereum Dencun (Cancun-Deneb) Upgrade?

The Ethereum Dencun upgrade was a coordinated hard fork that went live on 13 March 2024 at epoch 269568. Its name merges the two halves of the release: Deneb, a star governing consensus-layer changes, and Cancun, a former Devcon host city covering execution-layer changes.

The split matters because the two layers do different jobs. Cancun reworked how the Ethereum Virtual Machine handles data and memory, while Deneb refined validator operations, attestation timing, and staking mechanics on the Beacon Chain.

Dencun followed the Merge and Shanghai as the third pillar of Ethereum's post-proof-of-stake roadmap. It marked the formal start of the Surge, the phase where Ethereum commits to scaling through rollups rather than by expanding base-layer throughput directly.

What is EIP-4844 and Proto-Danksharding?

EIP-4844, better known as Proto-Danksharding, was the reason Dencun mattered. It created a new blob-carrying transaction type that lets rollups post compressed batch data to Ethereum in a dedicated lane priced separately from ordinary gas.

Each blob holds roughly 128 KB of data committed via KZG polynomial commitments and stored on the consensus layer, invisible to the EVM. Crucially, blobs are pruned after approximately 4,096 epochs, or 18 days, so nodes never carry them permanently.

That expiry is the whole trick. Before Dencun, rollups wrote data as calldata that every node stored forever, which was both expensive and unbounded. Blobs preserve availability long enough for fraud and validity proofs, then disappear, keeping hardware requirements flat as usage grows.

Proto-Danksharding also laid the cryptographic groundwork for full Danksharding, where blob counts scale far higher through data availability sampling. We cover the mechanics, blob fee market, and economics in depth in our dedicated EIP-4844 guide.

Ethereum Dencun EIP List by Layer

Dencun shipped nine EIPs, formally catalogued in the EIP-7569 hardfork meta. Two of them, EIP-4788 and EIP-4844, required changes on both sides of the protocol, which is why the layer counts overlap rather than divide evenly.

The Cancun Execution-Layer Proposals

• EIP-4844: Introduced blob-carrying transactions and a separate blob fee market, giving rollups cheap, temporary data availability instead of permanent calldata storage.
• EIP-1153: Added transient storage opcodes that behave like RAM, clearing after each transaction and sharply reducing gas costs for reentrancy locks.
• EIP-5656: Introduced the MCOPY instruction for efficient memory copying, cutting execution costs for contracts that manipulate large data structures.
• EIP-6780: Restricted SELFDESTRUCT to the same transaction as contract creation, closing a long-standing source of unexpected contract termination.
• EIP-7516: Added the BLOBBASEFEE opcode, letting smart contracts read current blob pricing and account for data costs programmatically.

The Deneb Consensus-Layer Proposals

• EIP-4788: Exposed the Beacon block root inside the EVM, creating a trust-minimised oracle for staking pools, restaking protocols, and bridges.
• EIP-7044: Made signed voluntary exits perpetually valid, so delegated stakers can pre-sign exits and withdraw without relying on node-operator permission.
• EIP-7045: Widened the attestation inclusion window to the end of the following epoch, strengthening LMD-GHOST security assumptions and confirmation rules.
• EIP-7514: Capped the validator churn limit at eight per epoch, slowing validator-set growth as a temporary brake on stake-related node bloat.

Core developer Tim Beiko authored the meta specification coordinating this set. The bundle reflected a deliberate strategy: ship one transformative proposal alongside smaller, low-risk improvements that had been queued since the Shanghai release.

Key Benefits the Dencun Upgrade Delivered

Dencun's improvements clustered around scalability, security, and validator experience, with each EIP addressing a constraint that had surfaced in the two years following the Merge.

The measurable benefits Dencun introduced include the following:

• Rollup scalability: Blobs gave Layer 2 networks a dedicated data channel, so rollup batches no longer compete with ordinary transactions for the same limited blockspace.
• Fee collapse: Layer 2 costs fell by 90-99% almost immediately, turning transactions that cost dollars into ones costing fractions of a cent.
• Bounded state growth: Automatic blob pruning after 18 days meant capacity could expand without permanently inflating the storage burden on node operators.
• Contract security: Restricting SELFDESTRUCT removed a mechanism that had enabled unexpected contract destruction and complicated the security assumptions of upgradeable systems.
• Staking flexibility: Perpetually valid exit messages let delegated stakers retain genuine withdrawal control, reducing counterparty risk in pooled and institutional staking arrangements.
• Cross-layer trust: Beacon block roots in the EVM allowed contracts to verify consensus state directly, benefiting liquid staking and restaking protocols.

What Happened to Layer 2 Fees Since Dencun

The fee effect was immediate and has proved durable. Within days of activation, major rollups reported reductions approaching 99%, and average costs dropped from roughly a dollar to well under a cent on several networks.

Subsequent upgrades compounded the gains. Pectra raised blob throughput from 3 target and 6 maximum to 6 and 9, while Fusaka introduced PeerDAS and staged parameter forks that lifted the ceiling to 21 blobs per block by January 2026.

The result is a fee environment unrecognisable from early 2024. Median Layer 2 transaction fees now sit around $0.02 on Base and a few cents across other major rollups, with simple transfers on some networks costing fractions of a cent.

Mainnet benefited indirectly too. Average Ethereum gas fell from roughly 7.14 gwei in January 2025 to about 0.50 gwei a year later, as execution traffic migrated to rollups and freed base-layer capacity.

How Dencun Changed ETH Supply and the Burn

This is the consequence few anticipated at launch. Because EIP-1559 burns base fees, and because blobs moved the bulk of transaction activity onto cheaper rollups, Dencun quietly removed Ethereum's largest source of fee burn.

The numbers are stark. Daily burn fell from thousands of ETH during busy pre-Dencun periods to as low as 50-70 ETH per day by early 2025. With issuance running near 1,700 ETH daily, the balance that produced deflation simply inverted.

ETH supply consequently crossed back above its Merge-era level and has grown since. Estimates through 2025 and 2026 place net annual inflation between roughly 0.2% and 0.8%, depending on the measurement window, with supply near 121.5 million tokens by March 2026.

That undercut the "ultrasound money" thesis that had underpinned much of the investment case. Justin Drake, the Ethereum researcher who coined the term, publicly paused the narrative in February 2025, noting supply was then expanding at roughly 0.5% annually.

The mechanism was never broken, only bypassed. Activity still generates fees, but it generates them on layers where the burn barely applies, which is precisely the trade-off the rollup-centric roadmap accepted in exchange for cheap, scalable transactions.

ETH Price Performance Since the Dencun Upgrade

Dencun landed at a local peak. ETH crossed $4,000 in the days around activation, its first time above that level since 2021, buoyed by upgrade optimism and mounting speculation about spot ETF approval.

The follow-through disappointed. ETH spent the remainder of 2024 and much of 2025 range-bound before a late surge to an all-time high near $4,950 in August 2025, then reversed sharply through a difficult first half of 2026 as macro conditions turned risk-off.

ETH now trades near $1,774, roughly 55% below its price on Dencun day and well under the August 2025 peak. Layer 2 tokens fared worse still, with Arbitrum down more than 90% from its January 2024 high amid persistent unlock pressure.

The disconnect is the story. On-chain usage climbed to record highs, with daily active addresses surpassing 1.3 million in mid-2026, yet neither ETH nor rollup tokens captured that growth. None of this is financial advice, so do your own research before allocating.

What are the Risks Dencun Introduced?

Dencun solved Ethereum's data-cost problem while creating structural dependencies that core developers are still working through. The risks below reflect what two years of live operation has actually exposed.

The lasting risks stemming from Dencun include the following:

• Fee cannibalisation: Shifting activity to rollups stripped mainnet of its primary burn source, converting ETH from a mildly deflationary asset into a mildly inflationary one.
• Layer 2 dependency: Ethereum's scaling now hinges on rollups whose sequencers, bridges, and governance sit largely outside protocol control, concentrating risk beyond the base layer.
• Sequencer centralisation: Most major rollups still run permissioned sequencers, so users inherit trust assumptions that conflict with Ethereum's own decentralisation standards.
• Fragmented liquidity: Cheap execution across dozens of rollups spread users and capital thinly, creating bridging friction and a worse experience than a single unified chain.
• Validator throttling: The churn cap slowed validator entry, which protected node performance but limited how quickly staking participation could expand.
• Volatile blob pricing: The separate blob fee market swung between near-zero and sudden spikes, complicating cost forecasting for rollup operators until later upgrades added a price floor.

What Dencun Set Up Next

Every Ethereum upgrade since has extended Dencun's design rather than replaced it. The blob mechanism proved sound enough that scaling became a matter of raising parameters instead of rebuilding architecture.

Pectra doubled blob throughput in May 2025 and Fusaka added PeerDAS in December 2025, letting validators sample blob data rather than download it whole. Core developers now target 48 blobs per block through further parameter forks, moving toward the 128-blob endgame of full Danksharding.

Attention has also shifted back to the base layer. Glamsterdam, targeted for the second half of 2026, introduces enshrined proposer-builder separation and parallel execution, an explicit attempt to scale L1 throughput and recapture some of the value that migrated to rollups after Dencun.

Final Thoughts

Dencun was an unambiguous engineering success and a more complicated economic one. It delivered the fee reductions it promised, made rollups viable for everyday use, and gave Ethereum a scaling path that has held up for over two years.

What it also did was expose the gap between network utility and token value. Cheaper blockspace attracted users but redirected the fees that once made ETH scarce, leaving a question the roadmap is still trying to answer. Glamsterdam is the first upgrade designed explicitly to close it.