How L2 Fees Have Changed Since EIP-4844

EIP-4844, also known as "proto-danksharding," is the Ethereum upgrade activated in March 2024 that introduced blob transactions — a cheaper, temporary way for rollups to post transaction data to the base layer — and it cut typical Layer 2 transaction fees by roughly 90–99% almost overnight.

Before that upgrade, a swap or transfer on an optimistic or zk-rollup could still cost more than most users expected for a network marketed as "cheap." After it, the same actions routinely settled for a small fraction of a cent to a few cents on many rollups. Understanding why requires looking at how L2s pay for security in the first place.

Why L2 fees were high before EIP-4844

Rollups execute transactions off-chain but still need to publish enough data back to Ethereum so anyone can verify what happened and, for optimistic rollups, challenge it if it's wrong. Before EIP-4844, that data was posted as calldata — a type of transaction input that Ethereum treats as first-class, permanently stored information.

Permanent storage is expensive. Every byte of calldata competed for space in the same fee market as regular Ethereum transactions, and validators had to store it forever even though rollups only needed it available for a short dispute or verification window. Data availability, not computation, was the dominant cost driver for most L2 transactions — often 80–90% of the total fee a user paid.

What EIP-4844 actually changed

EIP-4844 introduced a new transaction type carrying "blobs" — large chunks of data (about 128 KB each) that Ethereum consensus nodes store for roughly 18 days and then discard, rather than keeping forever. Rollups still get the same core guarantee — data is public and available long enough for validation and fraud-proof or validity-proof checks — but the network no longer bears the cost of storing it indefinitely.

Two changes made this cheap:

• A separate fee market. Blob gas has its own base fee that adjusts based on blob demand, independent of regular execution gas. Rollups posting data no longer compete directly with a busy NFT mint or DeFi rush for the same gas.
• Bulk, ephemeral storage. Because blobs expire, the protocol can afford to make each one large and inexpensive relative to calldata's per-byte cost.

Rollups fetch the necessary commitment data (via KZG polynomial commitments, which let anyone verify a blob's contents without downloading the whole thing) and post it using this new transaction type instead of calldata.

The immediate effect on fees

The result was structural, not temporary. Data-posting costs — the single biggest line item in most rollup fee formulas — dropped sharply the day the upgrade activated, and L2 operators generally passed the savings straight to users, since data cost is typically the dominant variable in their fee calculation.

Cost driver	Before EIP-4844	After EIP-4844
Data publishing method	Calldata (permanent storage)	Blobs (temporary, ~18-day retention)
Fee market	Shared with L1 execution gas	Separate blob gas market
Typical share of total L2 fee from data	80–90%	Much smaller, execution-dominated
Fee volatility driver	L1 gas price spikes	Blob demand across all rollups

Why fees still fluctuate

L2 fees didn't drop to zero and stay flat. Several things still move them:

• Blob demand competition. All rollups share the same blob space. When many L2s post data simultaneously — during high network activity — blob base fees can rise, and that cost flows through to users.
• Sequencer and prover costs. Optimistic rollups pay less for a sequencer to post transactions; zk-rollups also pay for proof generation, which can be a real, sometimes underappreciated cost, especially during periods of high proving demand.
• L1 execution gas. Data posting is cheap now, but the rollup's underlying settlement transactions still touch Ethereum's regular gas market for some operations.
• L2-side congestion. Individual rollups can still get busy enough that their own sequencer fee (separate from the L1 data cost) rises.

None of this is unique to any one chain — you can compare current gas conditions and activity levels across networks on a chains dashboard rather than relying on any single snapshot.

What comes next

Ethereum's roadmap calls for increasing the number of blobs each block can carry over time (often discussed under the "full danksharding" and PeerDAS umbrella), which should expand data capacity further and reduce the odds of blob fee spikes during busy periods. This doesn't guarantee fees only go one direction — new use cases and higher rollup adoption could increase blob demand faster than capacity grows — but the trend since EIP-4844 has clearly been toward L2 transactions becoming a much smaller cost relative to L1 activity, not a return to pre-4844 pricing.

For anyone comparing where to transact, it's worth understanding the difference between optimistic and zk-rollups, since proving costs and dispute mechanics like the challenge period on optimistic rollups both interact with the final fee you pay, not just the data-posting cost EIP-4844 addressed.

Bottom line

EIP-4844 didn't make Ethereum's base layer cheaper — it made data availability for rollups dramatically cheaper by giving them temporary, purpose-built storage instead of permanent calldata. That single change removed the largest cost component from most L2 transactions. Fees can still move with network demand and individual rollup congestion, so check a live gas reading before a large transaction rather than assuming last month's fee still applies.