The Dencun upgrade was Ethereum’s most consequential 2024 scaling achievement. Named for the combined execution-layer upgrade “Cancun” (EIP specification) and consensus-layer upgrade “Deneb” (beacon chain), it activated on March 13, 2024. The core innovation — EIP-4844 — fundamentally changed how Layer 2 rollups post their transaction data to Ethereum. Before Dencun, L2s posted data as expensive calldata ($$0.01–0.10 per L2 transaction). After Dencun, they post data as cheap temporary “blobs” ($0.0001–0.001 per L2 transaction). The 10–100× fee reduction transformed the economics of Ethereum L2s and directly contributed to Base, Arbitrum, and Optimism becoming viable platforms for high-frequency consumer applications. Understanding Dencun is understanding the cost structure of Ethereum’s modular architecture.
What Dencun Changed: Before and After
The following sections cover this in detail.
Before Dencun: Calldata Congestion
Prior to March 2024, L2 rollups (Optimism, Arbitrum, Base, zkSync, Starknet) posted compressed batches of their transactions as calldata to Ethereum L1:
- Calldata is permanently stored on Ethereum (every full node stores it forever)
- Gas cost: 16 gas per non-zero byte; 4 gas per zero byte
- At high ETH prices + congested L1: $0.01–$0.50 per L2 transaction
- Problem: L2 only needs this data to be available for ~7 days (during fraud proof window for optimistic rollups or for ZK proof verification) — but paying for permanent storage
The inefficiency: L2s were paying for permanent storage when they only needed temporary data availability.
After Dencun: Blob Transactions
EIP-4844 introduced a new transaction type: blob-carrying transactions.
- Each blob is 128 KB of data
- Initially: 3 blobs per block target, 6 blobs max per block
- Blobs are available from Ethereum full nodes for ~18 days (4096 epochs) then pruned
- Blob data is NOT accessible to the EVM (smart contracts can’t read blob contents) — only the KZG commitment hash is stored on-chain permanently
- Gas cost: Separate “blob gas” market, initially near 0 compared to calldata; eventually ~0.0001 ETH per blob
Why this works for L2s: Optimistic rollups need data available during the ~7-day fraud window. ZK rollups need data available until the ZK proof is verified (~hours). 18-day blob availability covers both — L2s don’t need Ethereum to store the data forever.
KZG Commitments: The Technical Foundation
The key cryptographic primitive enabling blobs is KZG polynomial commitments (Kate-Zaverucha-Goldberg):
Why KZG?
- Ethereum validators must verify that blob data is available without downloading all blob data
- KZG commitments allow a prover to commit to a polynomial (encoding the blob data) and later prove specific evaluations
- The commitment is compact (48 bytes) and can be verified quickly
- Validators check the KZG commitment on-chain; the full blob data is distributed via P2P network
The Trusted Setup (KZG Ceremony):
To use KZG, Ethereum needed a one-time trusted setup ceremony:
- 140,000+ participants contributed randomness in Jan–Feb 2023 (the “KZG Ceremony”)
- Largest trusted setup in cryptographic history
- Security: The ceremony is secure as long as at least one participant properly discarded their toxic waste (random contribution)
- 140,000 participants makes this near-certain to be secure
Limitation: KZG requires this trusted setup, which is why many ZK systems are moving to FRI-based commitments (no trusted setup needed). For Ethereum’s purposes, the massive ceremony provides sufficient security.
Impact on L2 Fees
The following sections cover this in detail.
Real-World Fee Reduction (Post-Dencun)
Fees across major L2s after Dencun activation:
| L2 | Pre-Dencun Avg Fee | Post-Dencun Avg Fee | Reduction |
|---|---|---|---|
| Base | $0.10–0.30 | $0.001–0.005 | ~50–100× |
| Arbitrum | $0.10–0.40 | $0.001–0.010 | ~30–100× |
| Optimism | $0.10–0.30 | $0.001–0.005 | ~50–100× |
| zkSync Era | $0.20–0.50 | $0.005–0.020 | ~20–50× |
| Starknet | $0.50–2.00 | $0.01–0.05 | ~30–100× |
Consequence: Base went from ~500K daily transactions to 3M+ daily transactions post-Dencun, becoming one of the highest-throughput EVM chains. The fee reduction unlocked previously uneconomical applications: micro-transactions, gaming moves, social media interactions.
The Trade-Off: Revenue Impact on Ethereum
Before Dencun: L2s paid L1 gas fees for calldata → significant ETH burned (boosting Ethereum’s deflationary economics)
After Dencun: L2s pay blob gas (separate, much cheaper market) → much less ETH burned from L2 data posting
The Ethereum community noted that Dencun made ETH issuance inflationary in some periods (blobs cost so little that fee burn < new issuance). Some argue this is a short-term transition issue; others argue it creates a long-term tension between L2 scaling and ETH value accrual.
Dencun’s Path: From Proto to Full Danksharding
The following sections cover this in detail.
Proto-Danksharding (EIP-4844, Now Live)
“Proto-danksharding” is the partial implementation:
- 3–6 blobs per block
- KZG commitment scheme established
- Blob fee market established
- Throughput: ~375 KB/block for rollup data
Full Danksharding (Future)
Full danksharding targets ~16 MB/block of blob data (vs. current ~768 KB):
- Requires Data Availability Sampling (DAS): Validators only need to download a random sample of blob data (not all of it) to verify availability using KZG proofs
- Requires danksharding validator participation: Different validators are responsible for different shards
- Enables: ~100,000 TPS equivalent across all L2s combined
- Timeline: 2026–2027 at earliest; technically complex
Why “Danksharding?” Named after Ethereum researcher Dankrad Feist, who proposed the design. The original full-sharding proposal by Ethereum was replaced by “danksharding” because the new design was simpler and more elegant — full sharding created 64 separate chains (complex coordination), while danksharding focuses on data availability (validators certify blob availability) rather than execution sharding.
Relationship to the Ethereum Roadmap
Dencun fits into Ethereum’s “Rollup-centric Roadmap” (introduced by Vitalik in 2020):
- Current state: Ethereum L1 handles settlement + consensus; L2s handle execution
- Dencun’s role: Reduces cost of L2 data posting to L1 (proto-danksharding)
- After Dencun: L2 fees are low but blob space is still limited
- With full danksharding: L2 fees approach zero with ~100K theoretical TPS
The roadmap stages:
- The Merge (2022): PoW → PoS (reduced energy, enabling future upgrades)
- Dencun (2024): Blobs (reduced L2 fees)
- Pectra (2025): EIP-7702 (EOA account abstraction), increased validator staking limits, other improvements
- Purge + Verge (2025–2027): Remove historical state, Verkle trees, making light clients viable
- Full Danksharding (2026+): Full blob throughput for L2 data
Other EIPs in Dencun (Beyond 4844)
EIP-4844 was the headline, but Dencun included several other improvements:
- EIP-1153 (Transient Storage — TSTORE/TLOAD): EVM opcodes for storage that persists only for one transaction; reduces gas cost for multi-step DeFi operations (e.g., flash loans) by removing need to store/clear-after-use in expensive persistent storage
- EIP-4788 (Beacon Block Root in EVM): Exposes consensus layer state to execution layer; enables smart contracts to verify validator set and consensus information without oracles
- EIP-5656 (MCOPY): New EVM opcode for efficient in-memory copying (performance improvement for smart contracts doing large data copies)
- EIP-6780 (SELFDESTRUCT limitation): Restricts SELFDESTRUCT to only clear ETH balance unless contract destructs in same transaction as creation; partial step toward eventually removing SELFDESTRUCT (needed for Verkle trees)
- EIP-7044/7045 (Voluntary exit improvements): Better staking UX improvements for validator management
Related Terms
Sources
Feist, D., Buterin, V., et al. (2022). EIP-4844: Shard Blob Transactions. Ethereum Improvement Proposals, EIP-4844.
Buterin, V. (2020). A Rollup-Centric Ethereum Roadmap. Ethereum Research Forum (ethresear.ch).
Kate, A., Zaverucha, G.M., & Goldberg, I. (2010). Constant-Size Commitments to Polynomials and Their Applications. ASIACRYPT 2010.
Buterin, V. et al. (2022). Ethereum Proof-of-Stake Design Rationale. Ethereum Foundation Research.
Thibault, L.T., Sarry, T., & Hafid, A.S. (2022). Blockchain Scaling Using Rollups: A Comprehensive Survey. IEEE Access, 10, pp. 93039–93054.