Gas is now data. The primary cost for rollups like Arbitrum and Optimism is posting state data to Ethereum for verification, not executing transactions. This makes data availability (DA) the new bottleneck and pricing mechanism.
developer-ecosystem-tools-languages-and-grants
Blog
The Future of Gas: A Shift from Computation to Data Availability
A technical analysis of how Danksharding will invert the L2 cost model, making data availability the primary driver of gas fees and reshaping developer priorities and infrastructure.
introduction
THE PARADIGM SHIFT
Introduction
The fundamental cost of blockchain execution is shifting from computation to data availability, redefining the role of L1s and L2s.
L1s become settlement layers. Ethereum's role evolves from a monolithic computer to a secure data ledger, where its high cost validates the integrity of cheaper L2 execution. This separation is the core thesis of modular blockchain design.
Evidence: Over 90% of Arbitrum's transaction fees are spent on Ethereum calldata. This economic reality forces L2s to adopt EIP-4844 blobs and explore alternative DA layers like Celestia or EigenDA to reduce costs.
thesis-statement
THE DATA LAYER
The Core Inversion
The primary cost and bottleneck for blockchains is shifting from raw computation to data availability and transport.
Gas is now data. The cost of executing a transaction is trivial compared to the cost of publishing its data. Rollups like Arbitrum and Optimism prove this; their execution is cheap, but posting data to Ethereum L1 dominates fees.
The bottleneck moved off-chain. The scaling fight is no longer about virtual machines. It is about data availability layers like Celestia, EigenDA, and Avail, which compete to provide the cheapest, most reliable data posting for execution environments.
This inverts the stack. The base layer's role is becoming a secure data ledger. Execution becomes a commoditized service atop this data, enabling hyper-specialized rollups for gaming, DeFi, or social apps.
Evidence: Arbitrum Nitro's cost is ~80% data posting. Validiums and zk-rollups like StarkNet rely entirely on external DA solutions to cut costs by orders of magnitude versus full settlement on Ethereum.
market-context
THE DATA
The Current State of Play
The fundamental bottleneck for blockchain scaling is shifting from raw computation to data availability and verification.
Gas is now data. The primary cost for L2s like Arbitrum and Optimism is publishing transaction data to Ethereum for security, not executing the computation itself. This creates a direct link between L2 scalability and the cost of Ethereum's data availability.
Rollups are data consumers. The performance of Optimistic and ZK rollups is gated by the throughput and cost of the underlying data layer. This dependency birthed blob transactions (EIP-4844), a dedicated data channel that separates data pricing from execution gas.
The market is fragmenting. Dedicated data availability layers like Celestia and EigenDA now compete with Ethereum's blobs. This modularization creates a new cost/security trade-off for rollup architects, decoupling execution from consensus and data.
Evidence: Post-EIP-4844, L2 transaction fees dropped by over 90% as data costs decoupled from execution gas. The competition is now about data bandwidth, measured in MB/s, not compute in TPS.
key-trends
THE FUTURE OF GAS
Three Data-Backed Trends Proving the Shift
The cost of on-chain execution is being eclipsed by the cost of data availability. Here's the evidence.
01
The Problem: Blob Gas is the New Block Gas
EIP-4844 introduced blobs, making data posting the primary cost for L2s. Execution is now a rounding error.
- Blob gas now consumes ~80% of an L2's mainnet settlement cost.
- Rollups like Arbitrum and Optimism see finality times gated by blob availability windows, not their own execution speed.
80%
Cost Driver
~10 min
Blob Window
02
The Solution: EigenDA & Celestia Redefine the Base Layer
Purpose-built DA layers decouple data availability from expensive consensus, creating a new cost curve.
- EigenDA offers ~$0.10 per MB, a 10-100x reduction vs. Ethereum blobs at peak demand.
- Celestia's modular design enables ~$0.001 per KB for high-throughput chains, making micro-transactions viable.
100x
Cheaper DA
$0.001/KB
New Floor
03
The Proof: L2 Economics Now Favor Data Compression Wars
The most effective scaling is no longer faster VMs, but better data compression to minimize DA costs.
- zkRollups like zkSync and StarkNet leverage STARK proofs for extreme compression, making state diffs tiny.
- Projects like Fuel and Aztec treat the DA layer as the scarcest resource, optimizing every byte to slash user fees.
90%+
Data Reduction
< $0.01
Target Tx Cost
GAS ECONOMICS
The Cost Breakdown: Execution vs. Data Today
A comparative breakdown of cost drivers for a standard Uniswap V2 swap across different environments, highlighting the shift from execution to data availability as the primary expense.
| Cost Component | Ethereum L1 | Optimistic Rollup (e.g., Arbitrum) | ZK Rollup (e.g., zkSync Era) | Validium (e.g., StarkEx) |
|---|---|---|---|---|
Execution Cost (Gas) | $5.20 | $0.03 | $0.02 | $0.01 |
State Storage (Calldata) Cost | $0.80 | $0.25 | $0.18 | $0.00 |
Data Availability Cost | N/A (on-chain) | $0.25 (on L1) | $0.18 (on L1) | $0.05 (off-chain w/ DAS) |
Total User-Paid Cost (Est.) | $6.00 | $0.53 | $0.20 | $0.06 |
Primary Cost Driver | Execution & Congestion | Data Availability | Data Availability | Proof Generation & DAS |
Trust Assumption for Data | None (on-chain) | 1-of-N Fraud Proofs | Cryptographic Validity Proofs | Data Availability Committee (DAC) or DAS |
Supports General-Purpose Smart Contracts |
deep-dive
THE DATA SHIFT
Why Computation Becomes a Commodity
The primary cost and bottleneck for blockchains is shifting from raw execution to data availability, commoditizing compute.
Execution is a solved problem. Modern L2s like Arbitrum Nitro and Optimism Bedrock achieve near-identical performance; the marginal cost of processing a transaction is negligible compared to posting its data on-chain.
The bottleneck is data, not logic. The cost of a transaction is the cost of its calldata or blobs on the base layer. This makes the L2's execution engine a commodity, with differentiation moving to sequencer design and proving costs.
Evidence: Post-EIP-4844, over 95% of an L2's cost structure is data availability. This is why zkSync, Starknet, and Polygon zkEVM all compete on proof efficiency, not VM speed.
protocol-spotlight
THE FUTURE OF GAS
Infrastructure Adapting to the New Reality
The primary cost of on-chain execution is shifting from raw computation to data availability and verification, forcing a fundamental redesign of infrastructure.
01
The Problem: Ethereum's Blob Gas Market
EIP-4844 introduced a volatile, auction-based market for data blobs. While cheaper than calldata, its cost is unpredictable and scales with demand, creating a new bottleneck for rollups like Arbitrum and Optimism.\n- Blob capacity is capped at ~0.75 MB per block.\n- Price spikes during high demand can still make L2 posting prohibitively expensive.
~0.75 MB
Blob Capacity
10x+
Volatility
02
The Solution: Modular DA Layers (Celestia, EigenDA)
Decouple data availability from execution by using specialized, cost-optimized layers. This creates a competitive market, driving costs toward the marginal price of storage and bandwidth.\n- Celestia uses Data Availability Sampling (DAS) for ~$0.01 per MB.\n- EigenDA, built on Ethereum, offers restaking-secured blobs at lower cost than mainnet.
-99%
vs. Calldata
$0.01/MB
Target Cost
03
The Architecture: Verifiable Databases (Avail, Near DA)
The next evolution: treat the DA layer not as a dumb blob store, but as a verifiable database with light-client proofs. This enables stateless execution and trust-minimized bridging.\n- Avail provides validity proofs for data ordering and availability.\n- Enables sovereign rollups and interop without a shared execution layer.
~2s
Proof Time
KB-sized
Client Footprint
04
The Endgame: Zero-Knowledge Proofs of DA
The final piece: cryptographically prove data was made available, eliminating any trust assumptions about the DA layer itself. Projects like zkPorter and Polygon Avail are pioneering this.\n- Validity proofs for data availability (e.g., using KZG or STARKs).\n- Enables maximum security with minimal cost, the holy grail for modular chains.
Trustless
Security Model
~1KB
Proof Size
counter-argument
THE EXECUTION EDGE
The Steelman: "Execution Will Still Matter"
The shift to data-centric blockchains redefines, but does not eliminate, the competitive moat of execution environments.
Execution becomes a premium service. When data availability is commoditized by EigenDA or Celestia, the primary value shifts to the execution environment that processes that data. This is where sequencers like Arbitrum and Optimism compete on speed, cost, and finality.
Performance is the new moat. The execution layer determines user experience. A rollup using cheap blob storage on Ethereum still needs a high-throughput sequencer to order and process transactions faster than its rivals.
Evidence: Arbitrum Nitro processes over 200,000 TPS internally, but settles only compressed proofs to Ethereum. This gap between internal execution and on-chain settlement is the new performance frontier.
risk-analysis
THE DATA AVAILABILITY FRONTIER
The New Risks and Bottlenecks
As L2s and modular chains proliferate, the primary bottleneck and cost center shifts from raw computation to the secure and efficient publication of transaction data.
01
The DA Cost Spiral
L2s currently pay ~$1M+ daily to post data to Ethereum. This cost scales linearly with usage, creating a hard ceiling for cheap transactions.\n- Bottleneck: Ethereum's ~80 KB/s data bandwidth.\n- Risk: High DA costs get passed to users as volatile L2 fees, undermining the 'cheap L2' narrative.
$1M+
Daily DA Spend
~80 KB/s
Ethereum Bandwidth
02
EigenDA & the Modular Gambit
EigenDA offers a cheaper, high-throughput data availability layer by using Ethereum restaking for security. It's a bet that cryptoeconomic security can replace full consensus for data.\n- Trade-off: Security shifts from consensus (L1 validators) to slashing conditions and restaker vigilance.\n- Capacity: Targets 10-100 MB/s, a 100x+ increase over Ethereum's base layer.
10-100 MB/s
Target Bandwidth
100x+
Capacity Increase
03
Celestia and the Sovereign Stack
Celestia decouples data availability and consensus into a dedicated layer, enabling sovereign rollups that control their own execution and governance.\n- Risk: Introduces a new liveness dependency outside of Ethereum's ecosystem.\n- Benefit: Creates a modular marketplace for execution, settlement, and DA, driving specialization and cost competition.
Sovereign
Rollup Model
New Liveness
Security Assumption
04
Data Availability Sampling (DAS)
The core tech enabling scalable DA layers. Light nodes can probabilistically verify data availability by sampling small random chunks, eliminating the need to download entire blocks.\n- Requirement: Needs a sufficient number of light nodes performing sampling for the system to be secure.\n- Threshold: Security approaches 100% as more nodes participate, creating a network effects challenge.
Probabilistic
Security Model
Light Nodes
Critical Mass
05
The Interoperability Fragmentation Risk
Multiple DA layers (Ethereum, Celestia, EigenDA, Avail) create data silos. Cross-chain messaging and bridging become more complex and risky when state roots are anchored to different consensus sets.\n- Problem: A bridge proving an asset lock on Celestia must be trusted by a chain using EigenDA.\n- Result: Proliferation of trust-minimized bridges like LayerZero and Axelar becomes critical, but introduces new oracle/relayer risks.
Data Silos
New Challenge
Bridge Complexity
Increases
06
The Verifier's Dilemma
With cheap DA, the cost of fraud shifts from publishing fake data to the computational cost of proving fraud. In optimistic rollups, this creates a verifier's dilemma where no one is incentivized to spend resources to challenge invalid state transitions for small thefts.\n- Mitigation: ZK-rollups (like zkSync, Starknet) provide validity proofs, making DA the only security assumption.\n- Trade-off: ZK-proof generation is computationally expensive, reintroducing a computation bottleneck at the prover level.
Verifier's Dilemma
Optimistic Risk
Prover Cost
ZK Bottleneck
future-outlook
THE DATA
The 24-Month Outlook: Blob Markets and Modular Wars
Gas pricing will decouple from computation and shift to a market for data availability, fundamentally reshaping rollup economics and infrastructure.
Gas is now data. The EIP-4844 blob fee market separates execution costs from data posting costs. Rollups like Arbitrum and Optimism now pay for blobs, not calldata, reducing L2 transaction fees by over 90% in some cases.
Blob markets are inefficient. The current 6-blob target creates a volatile, winner-take-all auction. This inefficiency will spawn specialized DA auctioneers like EigenDA and Celestia, competing on price and latency for rollup contracts.
Rollups become DA shoppers. The core competency shifts from pure scaling to cost-optimized data procurement. Teams will run continuous auctions across Ethereum, Avail, and near-data validiums to minimize this new dominant cost.
Evidence: Post-EIP-4844, Base's average transaction fee dropped from $0.31 to $0.002. This proves the cost structure is now data-led, not compute-led.
takeaways
THE FUTURE OF GAS
TL;DR for Builders and Architects
The blockchain cost model is shifting from paying for raw computation to paying for data availability, fundamentally changing how you design and scale applications.
01
The Problem: Gas is a Bad UX Tax
Users pay for the risk of wasted computation, not just the work done. This creates unpredictable, volatile fees that kill mainstream adoption.
- User Abstraction is impossible when gas is a direct, mandatory payment.
- Micro-transactions are non-viable with $1+ base fees on L1s.
- Batch operations (e.g., NFT mints) become prohibitively expensive.
$1+
L1 Base Fee
1000x
Volatility
02
The Solution: Pay for Data, Not Execution
Separate data publishing (DA) from execution. Users pay a flat, predictable fee to post data; execution becomes a commoditized, competitive market.
- Ethereum's Blobs enable ~$0.01 fixed-cost data posting via EIP-4844.
- Rollups (Arbitrum, Optimism) become pure execution layers.
- Intent-based systems (UniswapX, CowSwap) can now guarantee settlement costs upfront.
~$0.01
Per Blob Cost
100x
Cheaper L2 Txs
03
Architect for Modular DA
Build with a pluggable DA layer. Your stack should be agnostic to whether data settles on Ethereum, Celestia, Avail, or EigenDA.
- Separation of concerns: Let specialized DA layers compete on $ per byte and latency.
- Interoperability: Use shared DA for secure cross-chain messaging (e.g., LayerZero, Across).
- Future-proofing: Your app's economics improve as DA gets cheaper, without a migration.
~$0.001
Future $/byte
Modular
Stack Design
04
The New Abstraction: Account & Session Keys
With predictable data costs, you can abstract gas entirely. Users sign intents; a solver network competes to fulfill them, baking costs into the quote.
- ERC-4337 Account Abstraction becomes viable at scale.
- Sponsorship: Apps can pay for user operations as a customer acquisition cost.
- Session Keys: Enable seamless gaming and social interactions without constant wallet pop-ups.
0
User Gas
ERC-4337
Standard
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