L2 Gas Tracker
2026-09-11 · By L2 Gas Tracker Research

EIP-4844 Explained: Why L2 Fees Dropped 95% — and When They Could Spike Again

EIP-4844 blob transactions reducing Layer 2 gas fees by 90 percent explained with a fee drop chart

The upgrade nobody noticed

In March 2024, Ethereum activated the Dencun upgrade. Overnight, gas fees on Base, Arbitrum, and Optimism dropped by 80% to 95%. A swap that cost $0.10 the day before cost $0.01 the day after. Most users didn't even know it happened — they just noticed their transactions were suddenly cheaper.

The mechanism behind this is EIP-4844, also called proto-danksharding. It's the single most impactful upgrade to Ethereum L2 economics since the chains launched, and understanding it explains why your fees are what they are today — and when they might spike again.

How blobs work (without the jargon)

Before EIP-4844, every rollup had to post its transaction data to Ethereum mainnet as regular calldata. That data competed with every other transaction for block space, and when mainnet was busy, the cost of posting data went up — which meant your L2 fees went up too, even if the L2 itself was quiet.

EIP-4844 created a separate lane for this data: blobs. Blobs are attached to blocks but priced separately from regular gas. They have their own market, their own fee structure, and their own supply limit. Rollups post their batch data as blobs instead of calldata, and because the blob market is separate from the main transaction market, it stays cheap even when mainnet gas spikes.

Each block can hold up to 6 blobs (roughly 750 KB of data). That's enough for all the major rollups combined on a normal day. The fee for a blob is a fraction of what the equivalent calldata would cost — which is why your L2 fees dropped so dramatically.

The 95% drop, in numbers

I have receipts from before and after the upgrade:

That's a 90% to 95% reduction in a single upgrade. The L2 execution fee didn't change — it was already tiny. What changed was the L1 data fee, which dropped because blobs are so much cheaper than calldata.

The drop held through 2025 and into 2026. Base and Arbitrum fees have stayed in the sub-cent to few-cent range consistently, with occasional spikes during extreme congestion that I'll get to in a moment.

When blob fees spike — and why it matters for your wallet

Blobs have a separate fee market, but it's not immune to congestion. When blob demand exceeds the 6-per-block limit, the blob fee rises sharply — and your L2 fees rise with it.

I've seen this happen twice in 2025. The first was during a major Base airdrop claim window: thousands of users rushing to claim and swap simultaneously, pushing rollup data demand past the blob limit. Swaps on Base briefly hit $0.05 before settling back to $0.005 within an hour.

The second was during a period of sustained high activity across multiple chains at once — Base, Arbitrum, Optimism, and Linea all posting heavy batches simultaneously. Blob fees rose for about six hours, and L2 fees across all chains roughly doubled. Still cheap in absolute terms ($0.01 instead of $0.005), but the direction of travel was noticeable.

The pattern: blob spikes are short, sharp, and event-driven. If you see L2 fees suddenly elevated, check whether there's an airdrop, mint, or major market event happening. Wait an hour and it usually clears.

What comes next: full Danksharding

EIP-4844 is called proto-danksharding for a reason — it's the prototype. Full Danksharding, which doesn't have a firm date yet, will increase blob capacity from 6 per block to 64 or more. That would make blob congestion essentially impossible at current L2 activity levels, and push fees even lower.

For now, the 6-blob limit is the ceiling on how cheap L2 fees can get during peak demand. On a normal day, there's plenty of blob capacity. On a busy day, the blob market tightens and fees rise — that's the mechanism you're seeing when your swap suddenly costs double.

If you want to see whether blobs are currently congested, check the live tracker — if Base and Arbitrum fees are elevated simultaneously, blob congestion is the likely cause. If only one chain is expensive, it's local activity on that chain.