DA layer panoramic comparison 2026 · Celestia · EigenDA · Ethereum blobs · Avail · NEAR DA

TL;DR

5 mainstream DA (data availability) layers as of 2026-Q2 , compared across 8 dimensions: architecture · security model · cost per GB-day · throughput · finality · adopting rollups · governance · regulatory exposure
Ethereum blobs (EIP-4844 · see EIP-4844 blob economics and 2026 market impact · full view from Dencun to Pectra and Fusaka) are now the default for ETH-aligned rollups — highest security (directly backed by Ethereum L1 ) · medium cost · target=6/max=9 blob/block after Pectra (see Pectra EIP-7691)
Celestia is the pioneer of modular DA — independent PoS chain · Data Availability Sampling (DAS) + Nakamoto-style block production · lowest cost · but security ≠ Ethereum level
EigenDA is EigenLayer restaking-secured DA (see EigenLayer overview) — shares ETH staking economic security · costs slightly less than blobs · throughput is meaningfully higher
Avail is a Polkadot-line DA layer (formerly Polygon Avail · Avail Project after independence) — KZG + DAS · preparing hybrid mode (EigenLayer connection) · early market share
NEAR DA is NEAR Protocol’s DA service — uses NEAR consensus + extremely low cost · connected to RaaS providers such as Polygon CDK / Caldera · mainly serves non-ETH-aligned rollups
**Estimated cost per GB-day in 2026-Q2 **: Ethereum blob ~$0.10-0.30 (blob fee varies) · Celestia ~$0.02-0.05 · EigenDA ~$0.06-0.15 · Avail ~$0.04-0.08 · NEAR DA ~$0.01-0.03
adoption split: Base/Arbitrum/Optimism/zkSync/Linea/Scroll/Polygon zkEVM = L1 blobs · Mantle/Movement/Cyber/Rivalz = EigenDA · Manta Pacific/Eclipse (partly)/some Polygon CDK = Celestia · some Polygon CDK/Sophon = Avail · some Caldera RaaS = NEAR DA
Path: compare with systems index · ZK-EVM rollup maturity matrix 2026

Wiki route

This entry sits under systems index. Read it against EIP-4844 blob economics and 2026 market impact · full view from Dencun to Pectra and Fusaka for the Ethereum-native DA path that competes with all alt-DA, Pectra EIP-7691 · blob doubling and the L2 economic chain for the blob capacity expansion that affects all DA price competition, and EigenLayer overview for the restaking economics underpinning EigenDA. For the rollup ecosystem that consumes these DA layers, see ZK-EVM rollup maturity matrix 2026 and rollup market share 2026. For the meta-strategy of L1/L2 separation that motivates DA modularity, see Vitalik L1/L2 strategy anchor. For consensus models that DA layers use, see DAG-BFT vs Chain-BFT architecture and threshold BFT consensus Rust implementations. For cross-chain bridge security that interacts with DA choice, see cross-chain bridge security insurance matrix 2026. For restaking AVS landscape including EigenDA as the largest AVS, see restaking AVS landscape matrix.

Why this matrix matters

As of 2023-2024 , the default for DA layers was “rollups automatically use Ethereum L1 calldata.” After Dencun (2024-03) introduced EIP-4844 blobs, L1 blobs became the default choice for ETH-aligned rollups, while also opening the debate over “whether alt-DA still has a market.” In 2025-2026 , 5 DA competition patterns formed:

Core question of the modular thesis: do rollups need Ethereum-level DA security?

ETH-security maximalists (Vitalik’s public stance · see Vitalik L1/L2 strategy anchor): the core commitment of a rollup is that “L1 DA guarantees state reconstructability” · using alt-DA is validium, not rollup · it should not be called a rollup
Modular advocates (Celestia / Avail / EigenDA stance): rollup security ≠ DA must always be on L1 · different use cases can choose different security levels · “modular blockchain” is a new paradigm
Pragmatic middle path (Mantle / Manta / Eclipse, etc.): most dApps do not need L1 DA level · saved costs can be redirected to user subsidies

For developers choosing a rollup stack: DA choice determines sequencer cost, user gas cost, and the dApp economic model. For institutional clients (tokenized fund / DvP): L1 DA is institutional grade, while alt-DA is commercial grade (contrast with CCIP institutional, where institutions clearly demand L1 DA).

This matrix compares 5 publicly operating DA layers across 8 dimensions using a 2026-Q2 data snapshot.

Ethereum blobs (EIP-4844)

Architecture: Ethereum L1 native · per 1 block there are 4 blobs (target=3 · max=6 in Dencun · target=6 · max=9 after Pectra) · each blob 128 KiB · total max ~1.1 MiB/block post-Pectra. Blob data is retained for ~18 days · after that, only the commitment (KZG) remains on-chain. Blob pricing is independent from calldata and uses an EIP-1559 -style fee market.

Security model: Ethereum L1 PoS validators (~100 万+) + DAS (EIP-7594 PeerDAS introduced in 2026-Q4+). Slashing is enforced directly by the Ethereum L1 protocol. Current security level = Ethereum L1 itself (Nakamoto coefficient ~5 · total ETH staked ~$120B · attack cost ~$40B+).

Cost per GB-day (2026-Q2 estimate): ~$0.10-0.30 (blob fee varies · peak $1+ · trough $0.05). ~100x cheaper than calldata.

Throughput: after Pectra ~1.1 MiB/block (12s blocks) ≈ ~6 GB/day. Theoretically can reach ~60 GB/day after Fusaka PeerDAS.

Finality: Ethereum L1 finality (~12-15 minutes · 2 epoch).

Adopting rollups: Arbitrum · Optimism · Base · zkSync Era · Linea · Scroll · Taiko · Polygon zkEVM · Blast · Mode · Mantle (partly) · and 95%+ ETH-aligned rollups by default.

Governance: Ethereum protocol governance (EIP process · ACD) · on-chain social consensus. After Pectra comes Fusaka PeerDAS (expected 2026-Q4 ).

Regulatory exposure: Ethereum L1 itself has long been under SEC attention, but is widely recognized as “sufficiently decentralized” and not a security. Blob data itself is only byte storage · regulatory surface is minimal.

Celestia

Architecture: independent PoS L1 · Cosmos SDK + Tendermint BFT · DA-specialized (no EVM / smart-contract execution). Data Availability Sampling (DAS) lets light clients verify availability of the entire block through small samples, without downloading the whole block. 2D Reed-Solomon erasure coding + KZG commitment.

Security model: Celestia validators (~100 active validators · TIA token staking) · Tendermint BFT (2/3+ honest) · DAS provides a stronger guarantee that light clients can detect even majority-malicious validators. However, staking economic scale of ~$1-2B is far below Ethereum L1 .

Cost per GB-day (2026-Q2 estimate): ~$0.02-0.05 (depends on blob congestion). Cheapest · about 1/5 of Ethereum blob.

Throughput: 2 MB/block (2026 upgrade · block ~6s) ≈ ~28 GB/day. In 2027 , planned upgrade to 8 MB/block ≈ 120 GB/day.

Finality: Tendermint instant finality (~6s · single block).

Adopting rollups: Manta Pacific (major L2 customer) · Eclipse (some modules) · some Polygon CDK configurations · some Caldera RaaS · Astria sequencer · Movement Labs · Lyra · and ~30+ chains (2026).

Governance: Celestia Foundation + TIA token DAO · upgrades through on-chain governance vote.

Regulatory exposure: no clear SEC statement after TIA token launched in 2024 · Celestia Labs actively geo-fences US retail. Whether TIA qualifies as a utility token under EU MiCA is undecided.

EigenDA

Architecture: EigenLayer restaking secured (see EigenLayer overview and EigenLayer AVS mechanism) · EigenDA operators borrow ETH stake from EigenLayer for economic security · provide a high-throughput DA service. 3 roles: Dispatcher / Encoder / Validator · KZG commitment · DAS.

Security model: of EigenLayer’s ~$14B TVL, ~$6B opts in to EigenDA (see restaking AVS landscape matrix) · slashing enforced by EigenLayer protocol. Security ≈ ETH staking, but with 1 layers in between (restaking consent) · Vitalik warned that “restaking consensus should not be over-relied upon” (see Vitalik L1/L2 strategy anchor).

Cost per GB-day (2026-Q2 estimate): ~$0.06-0.15. Between blobs and Celestia, because operator + restaker rewards must be paid.

Throughput: ~15 MB/s sustained (~1.3 TB/day theoretical · in practice ~100-500 GB/day depending on load). The highest-throughput DA layer.

Finality: EigenDA dispatcher confirmation (~10-30s) + depends on EigenLayer slashing window.

Adopting rollups: Mantle (major) · Movement · Cyber · Rivalz · multiple ZK Stack hyperchains · multiple OP Stack rollups · some Arbitrum Orbit. EigenDA is the default alt-DA for modular rollup-as-a-service.

Governance: EigenLabs team + EIGEN token holders + Security Council.

Regulatory exposure: EIGEN token launched in 2024 · no clear SEC statement · EigenLayer actively delayed EIGEN airdrop to US users. Restaking economics look closer to securities from a regulatory perspective (see Liquid Staking + Restaking Ecosystem + CEX Exposure).

Avail

Architecture: independent DA chain (formerly Polygon Avail · spun off in 2023 as an independent project). Substrate framework (Polkadot lineage) · Babe+Grandpa consensus · KZG + DAS · similar to Celestia’s design but from the Polkadot academic lineage. Plans EigenLayer integration (“Avail + EigenLayer” hybrid mode) to further borrow ETH stake and strengthen security.

Security model: Avail validators (~150 active) · Babe+Grandpa BFT · staking economic scale ~$200-500M (2026). After EigenLayer hybrid mode goes live, it can borrow ETH stake to strengthen security · but in 2026-Q2 it mainly relies on its own AVAIL token staking.

Cost per GB-day (2026-Q2 estimate): ~$0.04-0.08. Slightly higher than Celestia because staking economic scale is smaller and validator subsidy is needed.

Throughput: ~2 MB/block (20s blocks) ≈ ~8.6 GB/day. Planned upgrade to 4 MB/block ≈ 17 GB/day.

Finality: Grandpa finality (~30s-1min).

Adopting rollups: some Polygon CDK chains · Sophon (zkSync ZK Stack partnership) · QuarkChain · Madara · LumioVM · some RaaS configurations. Customer count is lower than Celestia / EigenDA.

Governance: Avail Foundation + AVAIL token DAO (after 2025 token launch).

Regulatory exposure: AVAIL token is relatively new (2025 launch) · regulatory position unclear. Avail team is mainly Europe-based · prioritizes MiCA compliance.

NEAR DA

Architecture: DA service module of NEAR Protocol · uses NEAR sharded consensus + storage staking. Not an independent chain; a DA service on NEAR L1 · accessible to EVM rollups through cross-chain bridges.

Security model: NEAR validators (~250 active) · NEAR PoS consensus · staking economic scale ~$1-2B (NEAR token market cap × stake ratio). Bridge security is an additional surface (see cross-chain bridge security insurance matrix 2026).

Cost per GB-day (2026-Q2 estimate): ~$0.01-0.03. Cheapest · uses NEAR’s extremely low storage-cost economics.

Throughput: depends on NEAR sharding · theoretical ~100 MB/s sustained · actual rollup-client service ~10-50 GB/day.

Finality: NEAR finality (~2-3s · meaningfully faster than Ethereum).

Adopting rollups: some Caldera RaaS chains · experimental deployments by some OP Stack rollups · some Polygon CDK · mainly serves non-ETH-aligned rollups. Customer count < Celestia / EigenDA / Avail.

Governance: NEAR Foundation + NEAR token DAO · NEAR DA team leads the roadmap.

Regulatory exposure: NEAR token was reviewed in multiple jurisdictions in 2024-2025 , but has not been directly designated a security by the SEC. Regulatory exposure of the overall NEAR Protocol is relatively moderate.

Big comparison matrix table

5 DA layers × 8 dimensions (state as of 2026-Q2 ):

DA Layer	Architecture	Security model	Cost/GB-day	Throughput	Finality	Adopting rollups	Governance	Regulatory exposure
Ethereum blobs (EIP-4844)	Ethereum L1 native · KZG · target=6/max=9 (post-Pectra)	Ethereum PoS ~100万 validators · DAS (PeerDAS 2026-Q4) · attack cost $40B+	$0.10-0.30	~6 GB/day (post-Pectra) · ~60 GB/day after Fusaka	~12-15 min (2 epoch)	95%+ ETH-aligned rollups: Arbitrum · Optimism · Base · zkSync · Linea · Scroll · Taiko · Polygon zkEVM	Ethereum protocol governance (EIP/ACD)	Extremely low · Ethereum sufficiently decentralized consensus
Celestia	Independent PoS · Cosmos SDK+Tendermint · DAS · 2D RS+KZG	~100 validators · TIA staking ~$1-2B · strong DAS guarantee	$0.02-0.05 (cheapest modular DA)	2 MB/block 6s ≈ 28 GB/day · 2027 8 MB target	~6s instant (Tendermint)	Manta Pacific · some Eclipse · some Polygon CDK · Astria · Movement · ~30+ chains	Celestia Foundation+TIA DAO	TIA SEC no statement · geo-fence US · MiCA undecided
EigenDA	EigenLayer AVS · Dispatcher/Encoder/Validator · KZG+DAS	EigenLayer ~$6B opt-in (restaking) · slashing via EigenLayer · borrows ETH stake	$0.06-0.15	~15 MB/s sustained (~100-500 GB/day in practice)	~10-30s dispatcher + slashing window	Mantle · Movement · Cyber · Rivalz · ZK Stack hyperchains · some OP Stack · some Arbitrum Orbit	EigenLabs+EIGEN+SC	EIGEN SEC no statement · restaking is close to securities
Avail	Independent DA chain · Substrate (Polkadot lineage) · Babe+Grandpa · KZG+DAS	~150 validators · AVAIL staking ~$200-500M · EigenLayer hybrid planned	$0.04-0.08	2 MB/block 20s ≈ 8.6 GB/day · 4 MB upgrade ≈ 17 GB	~30s-1min (Grandpa)	Some Polygon CDK · Sophon · QuarkChain · Madara · LumioVM	Avail Foundation+AVAIL DAO	AVAIL 2025 new launch · MiCA compliance priority
NEAR DA	NEAR L1 module · sharded consensus · storage staking · EVM through bridge	NEAR ~250 validators · staking ~$1-2B · bridge security is additional surface	$0.01-0.03 (cheapest)	Theoretical 100 MB/s · actual ~10-50 GB/day	~2-3s (NEAR fast finality)	Some Caldera RaaS · OP Stack experiments · some Polygon CDK	NEAR Foundation+NEAR DAO	NEAR reviewed in multiple jurisdictions · moderate regulatory exposure

How to read the matrix:

Cost ranking: NEAR DA < Celestia < Avail < EigenDA < Ethereum blobs (blobs are most expensive but safest · NEAR is cheapest but has fewer customers)
Throughput ranking: EigenDA (highest) > NEAR (high theoretical · medium actual) > Celestia ≈ Avail ≈ Ethereum blob (post-Pectra)
Security-level ranking: Ethereum blobs (highest · L1 native) > EigenDA (borrows ETH restake) > Celestia ≈ NEAR DA (independent PoS · staking ~$1-2B) > Avail (smallest staking scale)
Finality speed ranking: NEAR DA (2-3s) > Celestia (6s) > EigenDA (10-30s + slashing) > Avail (30s-1min) > Ethereum blobs (12-15min)
Adoption scale ranking: Ethereum blobs (95%+ ETH-aligned) > Celestia (~30+ chains · alt-DA leader) > EigenDA (~20 chains · fastest growth) > Avail / NEAR DA (each <10 major chains)

Before vs after Dencun

Before Dencun (before 2024-03 ):

ETH-aligned rollups used L1 calldata · high cost · main rollup cost (~80%)
Alt-DA (Celestia 2023-10 mainnet · EigenDA 2024-Q1 mainnet · Avail 2024-Q2 · NEAR DA 2023) positioned itself as “90%+ cheaper” · took cost-sensitive customers
Modular thesis narrative: “rollups should be free to choose DA · no need to be bound to Ethereum L1 DA”

After Dencun (after 2024-03 ):

L1 blobs reduced ETH-aligned rollup costs by ~100x · many rollups chose to stay on blobs
Alt-DA price advantage narrowed from ~50x to ~3-5x (blob $0.10-0.30 vs Celestia $0.02-0.05)
Modular thesis narrative shifted: “for small cost differences, security level + ecosystem alignment should be considered · most ETH-aligned rollups choose blobs · non-ETH-aligned rollups choose alt-DA”
Pectra (2025-Q2) doubled blob target · further compressed alt-DA price advantage
Fusaka PeerDAS (planned 2026-Q4 ) adds another 10x blob capacity · alt-DA may be further compressed in ETH-aligned market share

Alt-DA differentiation strategies

Facing blob price pressure, each alt-DA chooses a different differentiation path:

Celestia:

“Modular sovereignty” route · emphasizes Cosmos / Move / non-EVM ecosystems (Manta · Eclipse · Movement · Astria)
Does not directly compete with ETH-aligned rollups · serves sovereign rollups that “want to choose their own settlement layer”
TIA token economic incentives · token incentives for rollup customers

EigenDA:

“ETH-aligned but extremely high throughput” route · serves high-TPS rollups (Mantle · heavy DeFi dApps)
Borrows economic security from EigenLayer $14B+ TVL · close to L1 DA level
Default for modular rollup-as-a-service (RaaS) · Caldera · AltLayer · Conduit · etc. all default to EigenDA

Avail:

“L1 DA alternative + EigenLayer hybrid” route · combines independence + borrowed ETH stake
Polkadot academic lineage · advantage in Polygon CDK / Substrate-line ecosystems
Current customer count is the smallest · whether EigenLayer hybrid attracts customers in the long term is the question

NEAR DA:

“Extremely low cost + NEAR fast finality” route · serves non-DeFi high-frequency use cases (gaming · social)
Customers are mainly RaaS · not major rollups
Long-term value depends on the health of the NEAR Protocol ecosystem

Who chooses which DA · decision dimensions

ETH-aligned + institutional grade (Coinbase Base · institutional RWA · tokenized fund):

Ethereum blobs required · security level = Ethereum L1 · minimum regulatory surface
Cannot choose alt-DA because institutional risk committees do not accept non-L1 DA

ETH-aligned + DeFi-native (major ZK/OP rollups such as Arbitrum · Optimism · zkSync):

Mainly Ethereum blobs · ecosystem alignment + user expectations
Some high-throughput child chains can choose EigenDA (Arbitrum Orbit · OP Stack child rollups)

Modular rollup-as-a-service (Caldera · AltLayer · Conduit customers):

Default EigenDA (ETH-aligned · high throughput)
Some customers choose Celestia (cheap) or Avail / NEAR DA (cheaper)

Non-ETH-aligned (Cosmos · Move · Solana ecosystem extensions):

Mainly Celestia (Cosmos-line alignment) · or NEAR DA (NEAR ecosystem)
Do not choose Ethereum blobs · ecosystem misalignment + unnecessary L1 fee exposure

Gaming / social / high-frequency low-value (Web3 game · social dApp):

Mainly NEAR DA (cheapest) · or Celestia (cheap + medium security)
Do not choose Ethereum blobs · cost still exceeds native gas budget

Boundary cases / future trajectory

Impact of PeerDAS (Fusaka 2026-Q4 planned):

Ethereum blob capacity 10x · per 1 day ~60 GB DA
Alt-DA price advantage narrows from 3-5x to ~1.5-2x · further pressure
Celestia / Avail plan to respond with their own capacity upgrades (Celestia 2027 8 MB block)

EigenDA’s restaking economic boundary:

EigenDA revenue in the 2026 estimate is ~$120M ARR (see restaking AVS matrix)
70-90% of revenue is distributed to operators + restakers · EigenLabs extracts 10-30%
If total EigenLayer TVL falls because restaking regulation tightens, EigenDA’s security budget falls in sync
Conversely, if restaking regulatory clarity + TVL growth arrive, EigenDA becomes dominant in alt-DA

Celestia long-term path:

Already the banner for the modular thesis · but ETH-aligned customers will not grow (because blobs are already cheap)
Long-term reliance on Cosmos / Move / sovereign rollup ecosystems · diverges from the ETH ecosystem
TIA token economics require rollup customers to keep paying DA fees, but fee increases are difficult in an alt-DA price war

Can Avail fight back?:

Strong Polkadot academic lineage · but small customer count
“EigenLayer hybrid” is key · if successful, borrowed ETH stake strengthens security and may take some customers from Celestia
Substrate / Polygon CDK integration is another path

NEAR DA niche path:

Clear price advantage · but overall NEAR Protocol ecosystem scale is limited
Long term, it may specialize in gaming / social / high-frequency low-value DA rather than top DeFi rollups
Serving EVM rollups through cross-chain bridges introduces an additional security surface

DA choice for institutional clients:

2026 institutional RWA cases (BlackRock BUIDL · Franklin Templeton FOBXX · etc.) are all deployed on Ethereum L1 or Polygon zkEVM L1-DA · alt-DA is not accepted
Whether institutions accept EigenDA (borrowed ETH stake) long term is EigenLabs’ hopeful path · but at present institutional risk committees still prefer pure L1 DA

Relationship between DA layers and ZK proving:

ZK rollups compress transaction batches into small proofs · proofs go to L1 · large data (state diff) goes to DA
The larger the DA layer capacity, the larger and more economical the batches ZK rollups can process
ZK rollups are more sensitive to DA choice because it is the largest cost component (see ZK-EVM rollup matrix 2026)

Multi-DA / DA aggregation:

Some rollups experiment with a “L1 blob + alt-DA two-layer” mode · store core commitment on L1 blob + full state diff on alt-DA
Provides fallback security (even if alt-DA fails, reconstruction is possible from the L1 commitment)
Hybrid path of the modular thesis · but engineering is complex and only a few deployments exist

Regulatory asymmetry:

Ethereum L1 is already sufficiently decentralized · minimum regulatory surface
Tokens of Celestia / EigenDA / Avail / NEAR DA are all in a regulatory gray zone · long-term uncertainty
Institutional clients prefer L1 DA to avoid regulatory risk · this is the long-term ceiling for alt-DA

Formal verification / engineering quality:

Ethereum blob implementation has multi-client support (Geth / Nethermind / Besu / Erigon / Reth) + long-term audits · most mature
EigenDA is EigenLayer’s main AVS · multiple audits + strict review before slashing launch
Celestia / Avail / NEAR DA have all passed audits · but production experience < Ethereum blob

Sources

Celestia docs · https://docs.celestia.org/
EigenDA docs · https://docs.eigenda.xyz/
Ethereum.org DA · https://ethereum.org/en/developers/docs/data-availability/
Avail docs · https://docs.availproject.org/
NEAR DA blog · https://near.org/blog/near-da
L2Beat DA summary · https://l2beat.com/data-availability/summary
DefiLlama rollups · https://defillama.com/chains/Rollup
Blobscan · https://blobscan.com/
Vitalik blog · “Layer 2s as cultural extensions of Ethereum” (public)
Ethereum Foundation EIP-7594 PeerDAS spec