Buterin says PeerDAS will make L2 faster
Vitalik Buterin argues that the PeerDAS scheme is key to scaling the Ethereum ecosystem, and its launch in the Fusaka upgrade should multiply L2 throughput.
PeerDAS is a protocol‑level mechanism for Ethereum, formalized in an EIP. In short, it introduces a new rule that lets nodes verify blob data availability without downloading the entire dataset.
As Buterin explains, the mechanism checks data availability in fragments: a node requests a limited number of random fragments and, with high confidence, infers that more than half of the block is valid and retrievable. If availability is confirmed, the missing parts are reconstructed via erasure coding – without every node having to store the network’s entire history.
The discussion was sparked by stats highlighted by Hildobby of Dragon XYZ, showing the network consistently hitting its target of six blobs per block for the first time since launch. Blobs (dedicated containers for L2 data) lower costs for rollups.
At peak load, Base used the most blob space (about 42%), followed by World (~25%), Arbitrum (~8%), and OP Mainnet (~4%), Scroll and Soneium each used about 3%..
There are still practical constraints. By Hildobby’s estimates, a validator today needs more than 70 GB for uncompressed blobs, and without pruning the volume can exceed 1.2 TB. About 10% of blobs are posted underfilled (especially by smaller networks), and roughly a third of blocks with blobs are touched by MEV activity.
The Fusaka upgrade, slated for December 3, is meant to address these issues. It will activate PeerDAS, after which maximum blob capacity will rise from the current 6 to 9, and then increase further in two steps: first to 15, then to 21.
Buterin notes there will still be scenarios at first where full block data is needed: during initial propagation and during recovery if the publisher didn’t provide the entire dataset. Even then, a single honest participant is enough for the protocol to succeed, since work is shared across nodes.
For users, that points to cheaper, more predictable fees on L2. For developers, it means a gradual, staged lift of limits without sudden network stress.
