The Great Unbundling: How Modular Data Availability Layers Are Slashing Rollup Costs and Rewriting the Security Playbook

Something quietly radical is happening to the economics of Ethereum scaling. The cost to launch and operate a Layer 2 rollup has dropped by roughly 90% in the past eighteen months, and the technology enabling this plunge is not some marginal optimization. It is a fundamental architectural shift that separates data availability from execution and settlement, a change that could reshape how hundreds of billions in crypto value moves and settles.

The numbers tell part of the story. In early 2023, posting calldata to Ethereum mainnet consumed the lion’s share of rollup operating costs, often running between $100,000 and $500,000 monthly for active chains. Today, a rollup leveraging Celestia, EigenDA, or Avail for data availability can reduce that line item by an order of magnitude or more. Base, the Coinbase-incubated Layer 2, reportedly saved approximately 90% on data availability costs after migrating to blobs and alternative DA strategies. Starknet and others have charted similar trajectories.

Yet this is not merely a story about cheaper infrastructure. The migration to modular data availability introduces new trust assumptions, fresh attack surfaces, and verification challenges that security auditors, bridge architects, and sophisticated users must grapple with now, not later. The savings are real, but so are the trade-offs. Understanding where the guarantees change, and by how much, has become an urgent competency for anyone building or deploying capital in the rollup ecosystem.

What We Mean by Modular Data Availability

To grasp what is changing, it helps to rewind slightly. Traditional monolithic blockchains like Ethereum in its original conception, or Bitcoin, handle three core functions in one place: execution (running transactions), settlement (finalizing state transitions), and data availability (ensuring that transaction data can be accessed and verified). Rollups emerged as a scaling solution by moving execution off the main chain while inheriting Ethereum’s security for settlement. But they still needed to post their compressed transaction data somewhere accessible and verifiable. For years, that somewhere was Ethereum itself, specifically through calldata or, more recently, EIP-4844 blobs.

This arrangement worked but was expensive. Ethereum block space is scarce and valuable. Every byte of rollup data posted to mainnet competed with DeFi transactions, NFT mints, and everything else. Rollups were essentially paying prime real estate prices to store data that only needed to be available for a limited challenge period, not preserved immutably for decades.

The modular thesis, articulated most clearly by the Celestia team and increasingly embraced across the ecosystem, proposes unbundling these functions entirely. A specialized data availability layer can optimize solely for making data available cheaply and efficiently, while Ethereum or another chain handles settlement and finality. Celestia, EigenDA (built on EigenLayer restaking), and Avail represent three distinct approaches to this specialization, each with different technical trade-offs and cryptoeconomic security models.

The Three Contenders: How They Work

Celestia: The Purpose-Built Chain

Celestia launched its mainnet in late 2023 as the first blockchain designed exclusively for data availability. Its core innovation is data availability sampling (DAS), enabled by two-dimensional Reed-Solomon erasure coding. Here is what that means in practice: block data is arranged in a matrix, extended with redundancy, and light clients can verify availability by sampling tiny random chunks rather than downloading full blocks. As more light clients sample, the statistical guarantee that the full data is available approaches certainty.

Celestia uses a native token (TIA) for consensus and data availability attestation. The security model is straightforward in principle: validators stake TIA, produce blocks, and can be slashed for withholding data. The chain does not execute transactions or maintain state in the traditional sense. It is pure data infrastructure.

The economic proposition has proven compelling. Celestia’s data availability cost has ranged between roughly $0.001 and $0.01 per kilobyte depending on network conditions, compared to Ethereum blob space that has fluctuated from $0.01 to $0.10 per kilobyte and higher during congestion. For a rollup processing millions of transactions, this differential compounds rapidly.

EigenDA: Restaked Ethereum Security

EigenDA, developed by EigenLabs and launched in 2024, takes a different path. Rather than creating a standalone chain, it operates as an actively validated service (AVS) on EigenLayer, which allows Ethereum validators to restake ETH to secure additional protocols. EigenDA operators are Ethereum validators who opt in, stake additional collateral, and commit to storing and serving rollup data.

The security model here is arguably more tightly coupled to Ethereum itself. Restaked ETH provides the economic backing, and the validator set overlaps substantially with Ethereum’s own. EigenDA uses a disperser architecture where data is encoded, distributed to operators, and certificates are generated for rollup contracts to verify on Ethereum.

EigenDA has demonstrated throughput capabilities that currently exceed Celestia’s, with published figures suggesting 10 megabytes per second or higher in test conditions, compared to Celestia’s initial throughput in the hundreds of kilobytes per second range. Whether this throughput advantage translates to sustained real-world differentiation depends on demand patterns and how each network scales.

Avail: The Polygon Spin-Out

Avail originated within Polygon Labs before spinning out as an independent entity in 2023. Its approach combines data availability sampling with validity proofs and a nominated proof-of-stake consensus mechanism. Avail emphasizes interoperability, positioning itself as a unification layer that can serve multiple ecosystems beyond Ethereum.

Avail’s technical design incorporates Kate polynomial commitments for efficient data availability verification, an alternative to Celestia’s commitment scheme. The project has focused heavily on developer experience and multi-chain support, recognizing that data availability needs extend across an increasingly fragmented landscape of rollups and app-chains.

The 90% Cost Reduction: Where It Comes From and Where It Goes

The dramatic cost reduction stems from several converging factors rather than any single breakthrough.

First, specialization itself yields efficiency. A chain optimized solely for data availability avoids the computational overhead of general-purpose execution. Celestia nodes do not need to maintain complex state or run the Ethereum Virtual Machine. This simplicity translates to lower operational costs passed through to users.

Second, supply and demand dynamics differ markedly. Ethereum blob space, even after EIP-4844, remains a scarce resource auctioned among all rollups and users. Dedicated DA layers can provision capacity more elastically and price it closer to marginal cost. The token economics of TIA, EIGEN, and AVAIL create different incentive structures than ETH-denominated blob fees.

Third, technological innovations in data availability sampling reduce the cost of verification. When light clients can confirm availability without downloading full blocks, the network can support more participants, increasing decentralization without sacrificing performance.

Real deployments illustrate the impact. Manta Pacific, a zero-knowledge rollup, migrated to Celestia for data availability in late 2023 and reported transaction cost reductions of approximately 99% for end users, with typical transactions falling below $0.01. Aevo, a derivatives exchange operating as an Optimistic rollup, integrated with Celestia and similarly achieved order-of-magnitude fee reductions. These are not isolated cases; by mid-2024, over a dozen rollups had committed to or launched with alternative DA layers.

The savings flow through the system in predictable ways. End users pay lower fees. Rollup operators retain more margin or compete more aggressively on price. Treasury runways extend for early-stage chains. Some of this surplus gets reinvested in improved execution layers, better user experiences, or developer incentives.

New Verification Assumptions: Where the Security Model Shifts

Here is where the story grows more complicated and more consequential. Moving data availability off Ethereum changes the trust assumptions, sometimes in subtle ways that are not immediately obvious to users or even to developers.

The data availability committee problem. Before dedicated DA layers, many rollups used data availability committees (DACs), small groups of trusted parties who attested to data availability. This was widely recognized as a weaker security model than posting to Ethereum. The modular DA layers replace these committees with cryptoeconomic guarantees, but the guarantees are not identical to Ethereum’s.

With Celestia, you trust the Celestia validator set and the correctness of the DAS implementation. The validator set is smaller than Ethereum’s and the token distribution differs. An attacker controlling a significant portion of staked TIA could theoretically withhold data or produce invalid availability attestations. The cost of such an attack is measurable and has fluctuated with TIA’s market capitalization, ranging from hundreds of millions to billions of dollars at various points, generally below Ethereum’s total stake.

EigenDA’s trust model is different but not strictly superior. You trust the EigenDA operator set, the EigenLayer restaking mechanism, and the slashing conditions. The restaking model introduces correlated risk: if EigenLayer experiences a systemic issue, multiple AVSs including EigenDA could be affected simultaneously. The “shared security” vision has benefits, but concentration in EigenLayer itself becomes a concern.

The bridge and settlement gap. Perhaps the most underappreciated risk lies in the interface between DA layer and settlement layer. A rollup using Celestia for data availability still typically settles on Ethereum. The Ethereum contract verifying rollup state transitions must somehow confirm that the underlying data was actually made available on Celestia. This verification path introduces latency and complexity.

Current implementations use data availability certificates attested by Celestia validators, or bridges that relay availability proofs. These bridges are new software with their own bug risks. The time to confirm data availability on Celestia and propagate that confirmation to Ethereum adds to the effective finality time. For applications requiring rapid cross-chain settlement, this matters.

The light client assumption. DAS enables light clients to verify availability with high probability, but “high probability” is not certainty. The security guarantee strengthens with more light clients sampling more chunks. Early in a network’s life, or during periods of low participation, the guarantee is weaker. An adversary might exploit this bootstrap phase.

The data retention question. Data availability is not the same as data storage. These layers guarantee that data was available at publication time, not that it remains retrievable indefinitely. Rollups must still plan for long-term data archiving, and the economics of this remain unresolved. If historical data becomes expensive to store or retrieve, the practical security of fraud proofs and historical verification degrades.

Real-World Tensions: Case Studies in Trade-Off

The theoretical risks have already manifested in practical challenges.

The Celestia launch and initial centralization. Celestia’s mainnet launched with a relatively concentrated validator set, typical of new proof-of-stake chains. While the set has diversified, early reliance on a limited number of operators meant that the theoretical cost of corrupting availability was lower than the market capitalization of all staked TIA might suggest. The network has worked to decentralize, but this illustrates how cryptoeconomic security evolves in practice.

EigenLayer’s restaking controversies. EigenLayer’s rapid growth to over $15 billion in total value locked by mid-2024 raised concerns about systemic risk. If restaked ETH secures dozens of AVSs, a catastrophic failure in one could cascade. EigenDA specifically has faced scrutiny about whether its throughput claims hold under adversarial conditions and whether the slashing conditions are sufficiently tested.

The Optimism-Bedrock-Celestia integration. Optimism’s Bedrock architecture was designed to be modular, allowing alternative DA layers. The actual integration with Celestia required careful engineering around the fault proof system. Optimistic rollups depend on the ability to challenge invalid state transitions, which requires access to the underlying transaction data. If the DA layer’s availability guarantee differs from Ethereum’s, the fault proof game’s assumptions shift. This engineering work is ongoing and not without complexity.

Base’s blob strategy. Base’s approach illustrates hybrid strategies rather than pure migration. By aggressively using EIP-4844 blobs and optimizing their data format, Base achieved substantial cost reductions while remaining on Ethereum for DA. This suggests that the 90% figure varies by implementation and that Ethereum-native optimizations remain competitive for some use cases.

The Auditor’s Dilemma and the Bridge Builder’s Burden

For security professionals, the modular DA shift creates immediate practical challenges.

Smart contract auditors reviewing rollup bridges must now understand not just Ethereum’s data availability guarantees but those of Celestia, EigenDA, or Avail. The verification logic in bridge contracts has grown more complex. An auditor must ask: what exactly is being attested? By whom? Under what slashing conditions? What is the latency? What happens if the attestation is wrong?

Bridge architects face harder design choices. A bridge between two rollups using different DA layers must verify availability on both, or trust relayers, or introduce new assumptions. The composition of security guarantees across modular stacks is not straightforwardly additive. A bridge might inherit the weaker of two guarantees, or introduce its own vulnerabilities in the composition.

The tooling for these assessments remains immature. There is no standard framework for comparing DA layer security equivalent to the maturity of Ethereum smart contract auditing. Teams like L2Beat have made progress in tracking and comparing these risks, but the field evolves faster than documentation.

Practical Guidance for Navigating the Shift

Whether you are deploying capital, building applications, or setting policy, the modular DA landscape demands new habits of verification and risk assessment.

For Users and Traders

• Verify where a rollup’s data actually lives. L2Beat and similar resources track this, but verify directly in documentation and contracts when significant value is at stake.
• Understand that “secured by Ethereum” means different things for execution, settlement, and data availability. A rollup settling on Ethereum but using Celestia for DA has different guarantees than one using Ethereum for all three.
• Consider the effective finality time, not just the headline transaction confirmation. DA layer confirmation plus settlement layer confirmation equals your actual wait for strong guarantees.
• Watch for centralization indicators: small validator sets, high token concentration, or reliance on single bridge implementations.

For Builders and Developers

• Architect with modularity in mind but do not treat DA layers as interchangeable commodities. Each has distinct latency, cost, and security characteristics that affect application design.
• Implement robust fallback mechanisms. If your primary DA layer experiences degradation, what happens? Can your rollup pause gracefully, or does it enter an unsafe mode?
• Invest in understanding the verification path. If you use Celestia, understand DAS light client security. If EigenDA, understand restaking slashing conditions.
• Plan for data retention beyond availability guarantees. Your users may need historical data for accounting, compliance, or dispute resolution.

For Investors and Analysts

• Evaluate DA layer tokens on fundamentals beyond speculation: actual rollup adoption, data throughput, fee generation, and network decentralization metrics.
• Assess the competitive dynamics. Ethereum’s blob space will improve with future upgrades. Dedicated DA layers must maintain their cost and performance edge.
• Consider regulatory exposure. Data availability layers may face scrutiny as infrastructure providers, particularly if they become chokepoints for sanctioned transaction flows.

For Policymakers and Regulators

• Recognize that data availability infrastructure is becoming critical financial plumbing. Its concentration or failure could affect consumer protection and market integrity.
• Avoid prescriptive technology mandates that ossify around current architectures. The field is evolving rapidly, and rules based on monolithic assumptions will misalign with reality.
• Focus disclosure and transparency requirements that help users understand where their transaction data lives and what guarantees apply.

Looking Ahead: The Next 12 to 24 Months

The modular data availability story is still in its early chapters. Several developments will shape the landscape through 2025 and beyond.

Ethereum is not standing still. Proto-danksharding (EIP-4844) was a first step; full danksharding, if implemented, would dramatically expand blob capacity and could narrow the cost gap with dedicated layers. The Ethereum roadmap’s pace and success will significantly influence whether alternative DA layers maintain their edge or become niche solutions for specific latency or interoperability needs.

The EigenLayer ecosystem’s evolution merits close watching. If restaking becomes a dominant security model, EigenDA’s integration with Ethereum’s validator set could prove more resilient than standalone chains. Conversely, if restaking introduces systemic risks that materialize, the entire model faces reassessment. The first major slashing events, when they occur, will be instructive.

Cross-DA interoperability standards are likely to emerge. Projects like the Optimism Superchain, zkSync’s Elastic Chain, and similar aggregation layers will need to handle heterogeneous DA assumptions gracefully. Standards for availability proofs, bridge verification, and security parameter comparison would reduce friction and risk.

Regulatory clarity, or its absence, will shape deployment patterns. Jurisdictions that provide clear frameworks for modular infrastructure may attract development; those with ambiguous rules may see builders self-select into simpler, more monolithic architectures to reduce compliance surface area.

Perhaps most fundamentally, the question of whether users actually care about these architectural distinctions remains open. If transaction costs fall sufficiently and experiences improve, most users will not actively choose based on DA layer design. The market may segment into premium offerings with stronger guarantees and commodity offerings with weaker ones, similar to how cloud computing tiers operate today.

What seems certain is that the unbundling of data availability from execution and settlement is irreversible. The economic pressure is too strong, the technical solutions too mature, and the developer demand too clear. The task ahead is not to reverse this trend but to navigate it wisely, understanding where the savings come from and what they cost in shifted assumptions.

The rollups that thrive will be those that communicate their security model transparently, compose their modular stack carefully, and build fallback mechanisms for when components behave unexpectedly. The auditors and architects who add genuine value will be those who understand these systems deeply enough to explain their trade-offs to non-experts without oversimplification or fear-mongering.

The 90% cost reduction is real. The new verification assumptions are equally real. Both demand our sustained attention as this infrastructure becomes the foundation for the next wave of on-chain activity.

---

What to Do Next

• Save this guide and revisit it during your next allocation decision.
• Cross-check key metrics with public dashboards.
• Share with your team and define one execution step this week.

Recommended Next Reads

• Crypto security basics: /category/cybersecurity/
• DeFi risk management: /category/defi/
• Blockchain technology explainers: /category/blockchain-technology/

Sources and Further Reading

• Chainalysis
• DefiLlama
• CoinGecko Research

FAQ

What is the main takeaway?

Focus on practical risk, utility, and execution rather than hype.

Who should care most?

Builders, active users, and investors exposed to the discussed sector.

What should readers do next?

Use the checklist, compare tools, and validate claims with primary sources.

Stay Updated

Subscribe to your site newsletter for weekly market breakdowns, tool comparisons, and risk alerts.