To No Avail

Avail anchors the case, and standalone data availability is the category under study. Avail repositories and audits, Ethereum’s EIP-4844 and EIP-7594, ethereum.org’s PeerDAS materials, Blobscan and L2BEAT public utilization surfaces, and public materials from adjacent providers anchor the comparison where category structure matters.

Public architecture and utilization evidence support a supply-side argument. They can show when the base layer narrows the architectural and cost rationale for an external DA chain. They do not settle the demand-side calculus rollup teams face when choosing DA providers, because switching costs, SDK integration effort, validator trust assumptions, and internal roadmap constraints sit outside the present materials.

Standalone DA combines three ingredients. Erasure coding allows data reconstruction from partial fragments. Commitment schemes bind published data to compact proofs. Data availability sampling enables light clients to verify publication without downloading the full dataset. These techniques are well understood, implemented across multiple providers, and increasingly embedded in the modular stack’s shared vocabulary. Reed-Solomon-style recovery logic, KZG commitment schemes, and sampling protocols are all public; no single provider holds exclusive claim to any of them.

The public availproject/avail, avail-core, and avail-light repositories (availproject/avail n.d.; availproject/avail-core n.d.; availproject/avail-light n.d.) reveal a Substrate-based chain with custom DA-specific logic, a light client, an SRS strategy tied to Filecoin’s Powers of Tau, and a public audit repository. Throughput and configuration are legible, and the implementation meets the functional requirements for standalone data availability as documented. Engineering quality varies across providers, but the category’s intellectual moat is thinner than its market language typically suggests, because the underlying cryptographic primitives are freely available and widely replicated. Celestia, the largest standalone DA provider by market share, implements a comparable primitive set on a Tendermint-based architecture; EigenDA takes a restaking-based approach. Avail anchors this case study because its Substrate-based design makes the settlement-absorption dynamic most visible, but the category argument applies to standalone DA providers as a class.

The dispute here is category durability.

Ethereum narrowed the standalone DA category by building comparable functionality into its own protocol.

EIP-4844 (EIP-4844 2022) created blob space in March 2024. By early 2026, EIP-7594 (EIP-7594 2024) and ethereum.org’s PeerDAS materials (Ethereum Foundation 2025) had moved native sampling and staged blob increases through BPO forks from roadmap aspiration into concrete protocol architecture. The structural fact is more consequential than the exact capacity numbers: Ethereum shifted from treating DA as a separate modular relief valve to treating it as an internal scaling surface. External DA providers, which had been supplying a missing primitive, now face competition from that primitive’s native implementation at the settlement layer itself.

Cost arithmetic reinforces this trajectory. Before EIP-4844 (EIP-4844 2022), rollups published transaction data as L1 calldata at roughly 16 gas per non-zero byte (EIP-2028 2019), producing DA costs high enough that external providers could offer order-of-magnitude savings while sustaining margins. Blob transactions (EIP-4844 2022) compressed that differential by approximately the same order of magnitude, pulling L1 DA pricing into the range external providers had previously occupied. PeerDAS (EIP-7594 2024) extends the compression further by scaling blob capacity without proportionate cost increases, since sampling distributes verification load across the validator set. The cost advantage that originally justified a separate DA chain has contracted from orders of magnitude to a contested margin, and a contested margin represents a fundamentally different strategic position from structural necessity.

Expanded capacity is underused, so cost convergence has not yet fully squeezed standalone providers. Post-Fusaka blob utilization data shows median 4-7 blobs per block against a target of 10, indicating that Ethereum’s DA capacity currently exceeds demand. Under these conditions, blob fees stay low and standalone DA providers face competitive pressure from cheap native capacity. If rollup demand grows to saturate expanded blob space, fees would rise and the cost-differential case for external DA providers would temporarily strengthen until the next capacity increase. The absorption thesis therefore describes a trajectory in which each successive capacity expansion narrows the cost advantage rather than a completed event. The timeline depends on the relative pace of rollup demand growth and Ethereum capacity expansion.

A standalone DA provider can solve a genuine engineering problem and find its strategic position eroding when the settlement layer internalizes the function that originally justified the provider’s existence as a separate category.

L2BEAT’s data-availability summaries (L2BEAT n.d.) and public utilization reporting across the external DA market converge on a consistent signal. Usage exists, but demand is concentrated among a small number of rollups, many of which are affiliated with or subsidized by the DA provider. Ethereum’s native blob market has become the default publication surface for rollups already anchored to Ethereum settlement. Pre-Fusaka Blobscan data (Blobscan n.d.) showed utilization near capacity under the original Pectra parameters (target 6, maximum 9). Post-Fusaka, with BPO1 raising the ceiling to target 10, maximum 15, the median blob count per block fell instead of rising to fill the new capacity; Blobscan’s March 2026 data showed median blob counts in the 4-7 range against the new target of 10, indicating that expanded capacity outpaced current demand. Ethereum has expanded native DA headroom beyond what current rollup demand consumes, narrowing the cost-differential case for external alternatives. Precise quantitative assessment of external DA utilization rates, or of the fraction of available capacity consumed by unaffiliated and unsubsidized demand, is unavailable from public sources at the time of writing. The absence of standardized utilization metrics across external DA providers limits comparison to directional observations.

The narrowing creates space for residual cases in which standalone DA might persist. It may stay meaningfully cheaper at relevant scales, achieve throughput or flexibility levels the settlement layer cannot yet offer in practice, or serve a class of applications that accepts weaker settlement-linked security because the cost-performance tradeoff is sufficient. A stronger residual case concerns multi-chain settlement. The absorption thesis assumes Ethereum as the primary settlement layer, yet rollups settling on Cosmos-based chains, alternative L1s, or sovereign settlement environments lack native blob space entirely. Celestia’s go-to-market strategy targets this segment explicitly. If a substantial share of rollup demand settles outside Ethereum, the category retains structural necessity for that segment regardless of Ethereum’s native DA expansion. The absorption trajectory therefore turns on Ethereum’s settlement-layer share, a variable that can be identified without forecasting. PeerDAS narrowed the category without eliminating it, and the remaining burden is durable niche value or sustained superiority within a landscape where native capacity already exists.

Avail’s widening roadmap beyond DA suggests that the project itself does not treat standalone DA as its sole future-facing thesis. If a category is narrowing, adjacent infrastructure gains strategic importance; cross-chain coordination, aggregation, and security composition are all more plausible long-term positions than insisting that DA alone will remain a sufficient value proposition. A project can adapt intelligently and bear the burden of proving each new layer on its own merits. Functional success on the DA layer does not automatically validate the remainder of the roadmap.

Supply-side conclusions lose their force if external DA providers sustain fee revenue from more than five independent rollup clients, each contributing at least 10% of the provider’s total demand, after EIP-4844 capacity increases have been live for twelve months or more. Such a distribution would indicate a diversified and structurally anchored demand base.

The thesis also weakens if Ethereum’s blob space stays persistently saturated at prices high enough that external DA retains a durable cost advantage exceeding an order of magnitude, signaling that the base layer’s internalization of the function is incomplete.

More broadly, revision is warranted if Avail, Celestia, or EigenDA demonstrate client-base diversification away from affiliated or subsidized rollups toward unrelated third-party consumers. At that point the chapter would need to accommodate a more persistent external-DA market whose demand structure resists the absorption trajectory described here.

Supply-side evidence supports several testable projections.

Avail is technically competent, yet the category around it can weaken. Avail shows that publication guarantees, commitments, and sampling can be implemented coherently on an external chain. Ethereum’s blob roadmap shows that the base layer can absorb the same function and narrow the cleanest architectural reason for keeping it separate.

Once the settlement layer no longer lacks the function that created the category in the first place, the rationale for separation erodes even when the standalone implementation remains sound. The category then has to carry itself through durable niche value, sustained cost advantage, or a settlement environment for which Ethereum’s native path is irrelevant.