Ethereum's Dencun Upgrade — What Proto-Danksharding Actually Changed for L2 Costs

The Dencun upgrade went live on Ethereum's mainnet on March 13, 2024. It introduced EIP-4844, also called "proto-danksharding," with a clear headline promise: dramatically cheaper transactions on Layer 2 networks.

Fourteen months later, the data is in. Let's be precise about what actually changed.

## What EIP-4844 Did

Before Dencun, Layer 2 networks like Arbitrum and Optimism posted their transaction data to Ethereum's *calldata* — permanent storage on the main chain priced to reflect its indefinite cost. Sixteen gas per byte of calldata, scaling linearly with data volume. For rollups that need to post compressed transaction batches frequently, this was a significant and growing cost.

EIP-4844 introduced a new data type: *blobs*. Blob data is attached to transactions but not stored permanently on the base layer — Ethereum nodes hold it for approximately 18 days and then discard it, retaining only the cryptographic commitment (a KZG proof). This dramatically reduces the storage burden, and the pricing reflects that. Blob space has its own separate fee market, disconnected from the main execution gas market.

The target capacity at launch was 3 blobs per block, maximum 6. Each blob holds 128 KB. At 12-second block times, this provides roughly 375 KB/second of L2 data throughput at target capacity.

## The Actual Fee Reduction

The numbers were striking. On the day Dencun activated, transaction costs on major L2 networks dropped approximately:

| Network | Pre-Dencun avg | Post-Dencun avg | Reduction |
|---------|---------------|-----------------|-----------|
| Arbitrum | ~$0.50 | ~$0.03 | ~94% |
| Optimism | ~$0.40 | ~$0.03 | ~93% |
| Base | ~$0.30 | ~$0.02 | ~93% |
| zkSync Era | ~$0.35 | ~$0.03 | ~91% |

This is close to the 90%+ reduction that had been projected. The mechanism worked as designed.

The blob fee market remained nearly empty for the first several months after launch. Blobs were essentially free because demand hadn't grown to fill capacity. That situation ended during high-activity periods in May–June 2024 (primarily meme coin cycles) when blob fees spiked briefly before falling again.

## What It Didn't Solve

Here's what's worth noting: the cost reduction happened, but it exposed a different bottleneck.

With L2 transaction costs at fractions of a cent, the user experience bottleneck shifted from fees to *fragmentation*. A user with assets on Base who wants to transact on Arbitrum still needs bridging infrastructure, still faces different security assumptions on different chains, still has to manage multiple token balances. The cost reduction didn't solve the coordination problem — it just made it more visible.

The fee reduction also compressed L2 operator economics. Several rollups had structured their business models around calldata costs that are now largely eliminated. Base, backed by Coinbase's revenue, can absorb this. Smaller rollups face real pressure.

Multi-blob parallel posting wasn't included in Dencun; it requires further protocol development. Full danksharding — which would increase blob capacity by 64x or more — remains years away on current development timelines.

## Where L2 Economics Stand

Rollup transaction volumes increased roughly 3x between March 2024 and March 2025, while fees fell 95%. Net fee revenue to L2 operators declined significantly in absolute terms despite volume growth. The economics are being rebuilt around sequencer revenue, MEV capture, and native token appreciation rather than per-transaction fees.

The numbers suggest something different from the simple narrative that "blobs solved the L2 cost problem." They dramatically reduced user costs while simultaneously compressing operator margins. The next phase of L2 economics will be defined by how rollups sustain themselves at near-zero per-transaction revenue — which is a harder problem than the one EIP-4844 solved.

> **Key Takeaway:** EIP-4844 delivered on its promise — average L2 transactions fell 90%+ from ~$0.30–0.50 to ~$0.02–0.05. The unresolved challenges — ecosystem fragmentation, compressed operator economics, limited blob capacity — are now more visible precisely because the fee problem is largely solved. Proto-danksharding was a necessary step; it wasn't the final one.

Ethereum's Dencun Upgrade — What Proto-Danksharding Actually Changed for L2 Costs

// COMMENTS

ON THIS PAGE