category.data_availability.label

category.data_availability.promise

Starting path

The cost claim this whole category rests on is out of date.

Data availability layers evaluated on published cost per MB, what actually secures the data, sampling and retention, and blast radius if the layer halts.

category.data_availability.promise Reviewed on published cost per MB, what actually secures the data, sampling and retention, and blast radius if the layer halts.

Start with the smart category assistant or jump straight into quick questions. Both paths narrow to the same recommendation standard.

How we review

Recommendation first, comparison only if needed.

Reviewed and rechecked on the current cycle.

Commercial relationships disclosed before clickout.

Audit

Infrastructure

Methodology

2026.04.0

published 2026-04-27

Providers tracked

580+

across 63 categories

Last verified

2026-08-05

editorial freshness

Editorial board

4

audit-log live

Decision guide

What matters most before choosing in this category

Measure the saving before you buy it. The claim that anchors this category — that Ethereum blobs cost many multiples of an alternative — was written when blob base fees were at a different level. Measured on the same day, blobs came to $0.0177/MiB and Celestia, the cheapest verified layer here, to $0.0121/MiB: a 1.5x gap. EigenDA's own published price is higher than blobs at $0.0274/MiB. Avail and 0G publish no per-MB price at all, and Avail's dominant fee term is multiplied by a governance-mutable parameter its docs tell you to read out of chainstate. If your rollup posts modest volume, the DA line may not be where your costs are.

Key question

Can you read a real price per MB off a public page or on-chain parameter today, or does it require a sales conversation?

Key question

How many days is your data guaranteed retrievable before you depend on archival nodes or your own snapshots?

Key question

If the layer halts, does sequencing stop and can the L1 bridge still verify the attestation?

Current editor lead

Ethereum Blobs (EIP-4844 / PeerDAS)

Native L1 blobspace, sampled since Fusaka, pruned after about 18 days

Data checked Aug 2026

Native L1 blobspace, sampled since Fusaka, pruned after about 18 days. Strongest on published cost per mb (verifiable, not negotiated) (9/10): The only layer whose price is a public on-chain parameter with no vendor in the loop: blob base fee derived at 9,053,288 wei/blob-gas = $0.0177/MiB on 2026-08-05, independently reproducible from block 25,688,703's excessBlobGas of 189,706,571. No account, no deposit, no negotiation. Weakest on sampling live + how long data stays retrievable (8/10): PeerDAS went live with Fusaka on 3 December 2025, with 'each full node holding only 1/8th of the data'; blob capacity now target 14 / max 21 per block after BPO2 on 7 January 2026. Retention is capped: MIN_EPOCHS_FOR_BLOB_SIDECARS_REQUESTS = 4096, 'which is around 18 days'.

Best for

Published cost per MB (verifiable, not negotiated) — 9/10

Main tradeoff

'As safe as Ethereum' is true for the attestation and false for the archive: consensus clients are only required to serve blob sidecars for 4096 epochs, about 18 days. After that the data is gone from the protocol and you need your own archive or a third-party blob indexer. Also, the price you see is a spot price on a volatile fee market — the $0.0177/MiB measured here reflects blob base fee at 0.009 gwei and ETH at $1,867.

Verify before signup

Blobscan front page, fetched 2026-08-05, three consecutive blob transactions: 3-blob tx 'B. Base Fee 3,559.897694208 Gwei'; 1-blob tx 'B. Base Fee 1,186.632564736 Gwei'; 6-blob tx 'B. Base Fee 7,119.795388416 Gwei' — exactly linear at 1,186.632564736 Gwei per blob. EIP-4844: GAS_PER_BLOB = 2**17 (131,072); MIN_BASE_FEE_PER_BLOB_GAS = 1. DERIVED: 1,186.63 Gwei / 131,072 = 9,053,288 wei per blob gas = 0.009053 gwei; $0.002216 per 128 KiB blob at ETH $1,867.16 (CoinGecko 2026-08-05) = $0.0177/MiB raw, $0.0183/MiB on the 126,976 usable bytes. Cross-checked against live block 25,688,703 excessBlobGas 189,706,571.

Weighted criteria

Published cost per MB (verifiable, not negotiated)30%

What actually secures the data30%

Sampling live + how long data stays retrievable25%

Blast radius if it halts + documented uptime15%

Leading options

Shared shortlist for this category

These providers are pulled from the same category comparison catalog used in validation, so the category page, comparison page, and provider reviews stay aligned.

Browse this network

Methodology

How this category is reviewed

Reviewed on published cost per MB, what actually secures the data, sampling and retention, and blast radius if the layer halts.

Reviewed on: Published cost per MB (verifiable, not negotiated), What actually secures the data, Sampling live + how long data stays retrievable, Blast radius if it halts + documented uptime.

This page is a maintained category surface, not a static marketing block. Review freshness, provider positioning, and recommendation logic should stay consistent with quiz and provider pages.

Frequently asked

Questions people ask before choosing da layers

Is Ethereum blobspace really much more expensive than alternative DA layers?

Not by the margin the category claims. Measured on 2026-08-05 with every price taken from the parameter or page that sets it and converted the same day, Ethereum blobs came to $0.0177/MiB and Celestia — the cheapest verified layer in this comparison — to $0.0121/MiB. That is roughly a 1.5x gap, not the 25x or more that alternative-DA marketing typically cites, because those figures were written when blob base fees were at a different level. EigenDA is the sharpest example: its own payments documentation quotes 0.015 ETH/GB, which works out to $0.0274/MiB, making it more expensive than the Ethereum blobs it is sold as an alternative to.

How long does my rollup data actually stay available?

Every layer in this category prunes, including Ethereum, and the guaranteed windows differ by more than an order of magnitude. Ethereum consensus clients are only required to serve blob sidecars for 4096 epochs — around 18 days. Celestia light nodes sample a 7-day window as of celestia-app v6 under CIP-36, and its docs warn that archival nodes serving historical data for free is not guaranteed. NEAR DA prunes from consensus after roughly 36 to 60 hours. Only Ethereum L1 calldata is permanent, and it measured 334x the cost of blobs. If a rollup node can go offline for a long weekend, that retention window is a real operational constraint, not a footnote.

Which DA layers actually run data availability sampling today?

Sampling lets a light client establish that data was published without downloading all of it. Celestia runs it in production, with light nodes sampling a share matrix and gossiping verified shares. Avail runs it light-client-side, sampling cells against per-row KZG commitments to compute a confidence score. Ethereum began sampling with PeerDAS at the Fusaka upgrade, with each full node holding a fraction of the data. EigenDA has none — validators hold disjoint erasure-coded shards sized to their stake — and what 0G describes as sampling is a node-side workload optimisation rather than the trust-minimising kind a rollup user runs.

REVIEWEDApr 2026METHOD4 criteriaCATEGORYdata_availability

Not financial advice · For informational purposes only · Always do your own research

Octopus · The AI CFO that pays for itself

Audit your entire crypto stack — free.

Score concentration, fees, security, and tax complexity across exchanges, wallets, staking, and DeFi.

Not financial advice · For informational purposes only · Always do your own research