Rollups are not permanent. They rely on temporary data availability layers like Ethereum calldata or external DACs, creating a future where transaction history can be lost. This makes L2s perpetual beta software, not settlement layers.
layer-2-wars-arbitrum-optimism-base-and-beyond
Blog
Why 'Pay-for-Permanence' Models Will Dominate L2s
Ethereum L2s face a state bloat crisis. The winning model will decouple ephemeral execution from permanent settlement, forcing users to pay for the data they truly need to keep forever. This is the only scalable path forward.
introduction
THE INCENTIVE MISMATCH
Introduction
The current L2 economic model, built on temporary data availability, is unsustainable and creates a critical security vulnerability for users.
Users subsidize sequencers. The prevailing pay-for-blockspace model lets sequencers profit from transaction ordering while externalizing the long-term cost of data storage. This is a classic tragedy of the commons.
Permanence is a public good. Protocols like Arweave and EigenLayer AVS operators demonstrate that persistent data storage requires dedicated, incentivized networks. L2s must internalize this cost.
Evidence: The EIP-4844 blob fee market proves data is a volatile, scarce resource. Relying on it for permanence makes L2 state hostage to Ethereum's gas auctions.
key-insights
THE END OF TEMPORARY DATA
Executive Summary
The current L2 subsidy model is unsustainable. The future belongs to protocols that treat data as a permanent asset, not a temporary liability.
01
The Problem: The Subsidy Cliff
Current L2s rely on sequencer profits to subsidize data posting to L1. This creates a fundamental misalignment. As transaction fees compress, sequencers face a choice: degrade security or become unprofitable. This model has a hard expiry date.
- Economic Insecurity: Data availability is a cost center, not a revenue stream.
- Centralization Pressure: Only heavily VC-backed sequencers can sustain long-term losses.
- User Risk: Your transaction history is only as permanent as the sequencer's balance sheet.
$0
Inherent Value
T-?
Cliff Date
02
The Solution: Data as a Bonded Asset
Pay-for-Permanence (PFP) models, pioneered by protocols like EigenDA and Celestia, flip the script. Users or rollups pay upfront to cryptographically guarantee data availability for a set duration (e.g., 1 year+).
- Aligned Incentives: Revenue funds verifiers and creates a sustainable service marketplace.
- Provable Guarantees: Permanence is a verifiable on-chain state, not a promise.
- Capital Efficiency: Unlocks restaking and other DeFi primitives on the security asset.
1 Year+
Guaranteed
Market Rate
Pricing
03
The Arbiter: Ethereum's Proto-Danksharding
EIP-4844 (blobs) is the catalyst. By creating a dedicated, low-cost data layer, it makes PFP models economically viable. Blob space is a scarce commodity to be priced by market demand, not hidden in a subsidy.
- Clear Cost Basis: L2s now have a predictable, standalone DA cost to pass through.
- Modular Competition: Enables a true market between EigenDA, Celestia, and Ethereum itself.
- Fee Market Evolution: Users will pay for speed (sequencing) and permanence (DA) separately.
~100x
Cheaper DA
Modular
Architecture
04
The Outcome: L2s as Premium Service Providers
The endgame is L2s competing on execution quality and user experience, not who can hide DA costs the longest. PFP turns data from a liability into a feature.
- Sustainable Stacks: Revenue flows to security providers (stakers/restakers).
- User Choice: Protocols can offer tiered permanence (e.g., 30 days vs. forever).
- Enterprise Adoption: Guaranteed data integrity is a non-negotiable for institutional use cases.
Tiered
Service Model
Provable
Integrity
thesis-statement
THE ECONOMIC IMPERATIVE
The Core Thesis: Permanence is a Premium Product
The market will segment into volatile, ephemeral L2s and expensive, permanent ones, because data availability is the ultimate cost center.
Permanence is the bottleneck. The cost of storing data forever on Ethereum is the only non-negotiable expense for an L2. Everything else—execution, proving—is commoditized and trending to zero.
'Rollup-as-a-Service' commoditizes execution. Platforms like Conduit and Caldera let anyone spin up a chain, but they outsource data to Celestia or EigenDA. This creates a two-tier market: cheap, temporary chains versus expensive, Ethereum-aligned ones.
Users and apps self-select. A memecoin casino uses Avail for sub-cent fees. A derivatives protocol like dYdX pays for Ethereum-grade security. The premium for permanent, verifiable state is a feature, not a bug.
Evidence: Arbitrum, Optimism, and Base collectively pay over $1M daily in Ethereum L1 data fees. This is the premium for permanence. Chains using alternative DA save 90-99% but inherit that system's security and liveness assumptions.
market-context
THE COST MODEL
The L2 Scaling Paradox: Cheap Txs, Expensive State
L2s optimize for cheap execution but externalize the long-term cost of state storage to users, creating a misaligned economic model.
Execution is cheap, state is not. L2s like Arbitrum and Optimism compress transaction data for cheap posting to Ethereum, but the permanent state storage cost remains on the base layer. This creates a subsidy where L2 sequencers pay for temporary data availability, not final settlement.
Users will pay for permanence. The current model where L2s absorb state costs is unsustainable at scale. Future models like EIP-4844 blob storage and volition architectures (e.g., StarkNet, zkSync) shift this cost to users who require long-term data availability, separating ephemeral interaction from permanent record-keeping.
Proof systems dictate cost. A ZK-rollup's validity proof finalizes state instantly, making its posted data less critical over time. An optimistic rollup's fraud proof window requires data availability for 7+ days, creating a longer and more expensive custody burden for the network.
Evidence: Arbitrum processes over 1 million transactions daily, but its cumulative state growth on Ethereum exceeds 500 GB. The cost to regenerate this state from only calldata, if data were pruned, would be prohibitive, proving the state permanence subsidy.
L2 DATA AVAILABILITY
The Cost of Forever: Data Storage Economics Compared
A comparison of long-term data storage models for Layer 2 blockchains, analyzing the economic and security trade-offs between on-chain, off-chain, and hybrid solutions.
| Feature / Metric | On-Chain (e.g., Arbitrum, zkSync) | Off-Chain DAC (e.g., Celestia, Avail) | Hybrid / Modular (e.g., EigenDA, Near DA) |
|---|---|---|---|
Data Availability Guarantee | Ethereum Consensus | Data Availability Committee | Cryptoeconomic Security + Committee |
Cost per MB (Est.) | $1,200 - $3,000 | $0.01 - $0.10 | $0.10 - $1.00 |
Settlement Finality Time | ~12 minutes (Ethereum) | ~2 seconds | ~2 seconds to ~12 minutes |
Censorship Resistance | |||
Requires Native Token Staking | |||
Proposer-Builder Separation (PBS) Support | |||
Data Prunability Risk | None (Full History) | High (Expiry Policies) | Configurable (Slashing for loss) |
Integration Complexity | Low (Native L1 Bridge) | Medium (Light Client / Bridge) | High (Dual-Trust Model) |
deep-dive
THE SUBSIDY SHIFT
Architectural Blueprint: How Pay-for-Permanence Works
Pay-for-Permanence re-architects L2 economics by shifting the cost of data availability from the user to the protocol, creating a superior user experience and a sustainable business model.
User experience is the bottleneck. Current L2s force users to pay for data publication to Ethereum, creating unpredictable, volatile fees that break UX. Pay-for-Permanence protocols like Arbitrum BOLD and Base's Onchain Summer subsidies absorb this cost, presenting users with a simple, predictable fee.
The subsidy creates a moat. By internalizing the cost of data availability (DA), protocols transform a public good cost into a competitive feature. This model mirrors how AWS or Google Cloud abstract infrastructure costs, allowing builders to focus on product, not gas economics.
Revenue shifts to premium services. The free base layer becomes a loss leader. Monetization moves to value-added services like priority ordering, fast finality proofs, and enterprise-grade data indexing, similar to Celestia's rollup-as-a-service model.
Evidence: Arbitrum sequencers already subsidize ~30% of L1 posting costs. This is not charity; it's a calculated investment in user acquisition and retention that pure utility-token models cannot match.
protocol-spotlight
PAY-FOR-PERMANENCE PIONEERS
Early Movers: Who's Building This Future?
These protocols are moving beyond temporary data availability to architect the permanent, secure, and economically sustainable L2s of tomorrow.
01
Celestia: The Modular DA Layer
The Problem: Monolithic chains force L2s to pay for expensive, bundled execution and consensus.\nThe Solution: Celestia provides a pure, scalable data availability layer, decoupling security from execution.\n- $1B+ market cap for a pure data layer validates the thesis.\n- Enables ~$0.001 per MB data posting, a 100x+ cost reduction vs. Ethereum calldata.
100x+
Cheaper DA
Modular
Architecture
02
EigenLayer & EigenDA: Restaking for Permanence
The Problem: Dedicated DA layers require bootstrapping new trust networks and capital.\nThe Solution: EigenLayer restakes Ethereum's $15B+ staked ETH to cryptoeconomically secure new services like EigenDA.\n- Leverages Ethereum's established trust for rapid adoption.\n- Provides a high-throughput, cost-effective DA alternative directly secured by ETH.
$15B+
Restaked TVL
ETH Secured
Security
03
Arbitrum & the Nova Sequencer
The Problem: Even optimistic rollups need fast, cheap data posting for a viable user experience.\nThe Solution: Arbitrum Nova uses a DAC (Data Availability Committee) with EigenDA as a fallback, creating a hybrid model.\n- ~$0.01 average transaction cost for high-volume apps.\n- Proves that top-tier L2s are actively integrating next-gen DA to stay competitive.
~$0.01
Avg. Tx Cost
Hybrid DA
Model
04
Avail: Polygon's Zero-Knowledge Bet
The Problem: Data availability sampling requires light clients to download large data chunks.\nThe Solution: Avail uses ZK-proofs and validity proofs to allow light clients to verify DA with minimal data.\n- Enables trust-minimized bridging and sovereign rollups.\n- Positions itself as the DA backbone for the broader Polygon ecosystem and beyond.
ZK-Proofs
Core Tech
Sovereign
Rollups
05
The Economic S-Curve: Subsidies → Sustainability
The Problem: L2 growth is currently fueled by unsustainable token subsidies and sequencer profit capture.\nThe Solution: Permanent data slashes the core operational cost, flipping the unit economics from negative to positive.\n- Transforms L2s from cash-burning startups into profitable data markets.\n- Enables sustainable revenue sharing back to token holders and ecosystem builders.
Positive
Unit Economics
Profit Capture
Flip
06
Near's Nightshade: Sharding for Scale
The Problem: Throughput bottlenecks limit how much cheap data a single DA layer can provide.\nThe Solution: Nightshade implements dynamic sharding, where each block contains a slice (shard) of all transactions.\n- Aims for 100k+ TPS of pure data availability capacity.\n- Provides the foundational throughput required for mass adoption of permanent data storage on-chain.
100k+
Target TPS
Dynamic Sharding
Architecture
counter-argument
THE ECONOMIC REALITY
The Counter-Argument: Isn't This Just Kicking the Can?
Permanent data storage is not a cost to be avoided, but a premium service that will define L2 competitiveness.
Permanence is a product feature. The 'can' is not being kicked; the cost is being productized. Users and applications will pay for guaranteed data availability the same way they pay for faster block times or lower fees. This creates a clear market segmentation between ephemeral, low-cost chains and premium, permanent ones.
L2s compete on security guarantees. A rollup that relies solely on Ethereum calldata or a validium with a Data Availability Committee offers a weaker trust model. Chains like Arbitrum and zkSync already segment their offerings (AnyTrust vs. Rollup) based on this exact trade-off. Permanence is the ultimate security premium.
Evidence: The market votes with its TVL. Arbitrum Nitro, which posts all data to Ethereum, consistently holds over 2x the TVL of its AnyTrust-based Arbitrum Nova. This demonstrates that for high-value applications, users pay for cryptographic certainty, not temporary promises.
takeaways
THE ARCHITECTURE SHIFT
Strategic Implications: What This Means for Builders
The move from ephemeral to permanent data availability is a fundamental change in L2 economics, forcing builders to rethink their stack and business model.
01
The Problem: The Blob Debt Spiral
Rollups currently treat data as a recurring operational cost, creating a fragile dependency on volatile L1 gas prices. This model is unsustainable for protocols requiring long-term state guarantees or high-frequency data writes.\n- Cost unpredictability jeopardizes protocol economics.\n- Data pruning after 18 days breaks composability for long-tail assets.
~$2M/day
Current Blob Spend
18 Days
Pruning Window
02
The Solution: Prepaid State as a Service
Protocols will shift from renting temporary blob space to purchasing permanent state slots. This transforms data from an OpEx to a CapEx, enabling new primitives.\n- Guaranteed perpetual access enables verifiable asset histories (e.g., NFT provenance, RWA ledgers).\n- Predictable, one-time cost allows for precise financial modeling and novel subscription models.
100%
State Uptime
Fixed Cost
Pricing Model
03
The Arbiter: Celestia vs. EigenDA vs. Avail
The competition for permanent data isn't about raw throughput; it's about cryptoeconomic security and integration surface. Builders must choose based on threat models.\n- Celestia: Maximizes sovereignty with light clients, ideal for app-chains.\n- EigenDA: Leverages Ethereum restaking for security, best for ETH-aligned rollups.\n- Avail: Focuses on data availability proofs for scalable validity.
$1B+
Security Budgets
3-5s
Finality Target
04
The New Business Model: Data Annuities
Pay-for-permanence enables L2s to sell lifetime data packages, creating a recurring revenue stream decoupled from transaction volume. This mirrors SaaS models.\n- Monetize state, not just execution via one-time sale of permanent storage slots.\n- Unlocks institutional use-cases (e.g., legal contracts, audit trails) that require immutable, long-term data guarantees.
Recurring Rev
Revenue Shift
Institutional
New Market
05
The Integration Mandate: Rethink Your Stack
Builders can no longer treat the DA layer as a black box. Permanent data requires direct integration with proof systems and indexers. This is a core engineering competency.\n- Must integrate light clients (e.g., Celestia's Blobstream) for trust-minimized bridging.\n- Requires new indexer infra to query perpetual state efficiently, beyond The Graph's current model.
New Primitive
Light Clients
Core Infra
Indexing Layer
06
The Endgame: Sovereign Rollup Dominance
Permanent, cheap data is the final piece for truly sovereign rollups. This reduces L1 dependency to a pure security function, shifting power to L2 communities.\n- Enables forkless upgrades and custom fee markets controlled by the rollup.\n- Accelerates the shift from 'Ehereum L2s' to 'Ethereum-secured L1s', a la Polygon CDK and Arbitrum Orbit chains.
Sovereignty
Key Trait
L1-Secured
Security Model
ENQUIRY
Get In Touch
today.
Our experts will offer a free quote and a 30min call to discuss your project.
NDA Protected
Confidentiality24h Response
Time to ValueDirectly to Engineering Team
No Sales Fluff10+
Protocols Shipped
$20M+
TVL Overall