Celestia and Ethereum are both racing to solve the same bottleneck: making transaction data cheap and accessible for rollups. Their approaches could hardly be more different. Celestia built a standalone network purpose-designed for data availability from the ground up. Ethereum is retrofitting its existing infrastructure through a phased upgrade path called danksharding, adding dedicated blob space without abandoning its monolithic foundation.

Which approach wins matters for rollup developers choosing where to post data, for traders evaluating TIA and ETH, and for anyone trying to understand where blockchain infrastructure is actually heading.

How Danksharding Expands Ethereum's Data Layer

Danksharding is Ethereum's long-term plan to dramatically increase how much data rollups can attach to blocks. According to Ethereum's official roadmap documentation, the full version will expand blob capacity from the current 6 blobs per block to 64, delivering approximately 8.2 MB of data per block and enabling the network to support hundreds of rollups processing millions of transactions per second.

The upgrade rolls out in stages. Proto-danksharding (EIP-4844) shipped with the Dencun upgrade in March 2024, introducing blob-carrying transactions for the first time. Each blob holds 128 KB of data that consensus nodes verify and then automatically prune after roughly 18 days. This temporary storage model was revolutionary for Ethereum because rollup data no longer needed to live on-chain permanently, cutting Layer 2 fees by 90-95% almost overnight.

The Pectra upgrade in May 2025 increased blob targets further, and the Fusaka upgrade that shipped in December 2025 introduced PeerDAS (Peer Data Availability Sampling). PeerDAS lets validators verify data availability by downloading only about 12.5% of blob data rather than the entire dataset. This mechanism is the critical stepping stone toward full danksharding, which remains on the roadmap for 2026 or later.

KZG Commitments and Verification

Ethereum's blob verification relies on KZG polynomial commitments, a cryptographic scheme that lets validators confirm data integrity without downloading every byte. As a16z crypto's research explains, the current design requires 75% of data shares to reconstruct a block, though proposed modifications could reduce that threshold to 25%. Each validator receives a small fragment of the extended data matrix and uses KZG proofs to verify consistency with the block's commitment, keeping hardware requirements manageable even as blob counts scale up.

Celestia's Purpose-Built DA Architecture

Where Ethereum adds data availability capacity onto an existing execution platform, Celestia strips everything else away. The network handles ordering and data availability exclusively, with no smart contract execution, no settlement logic, and no EVM. Rollups post their transaction data as blobs to Celestia, then handle execution and settlement on separate layers.

This specialization enables architectural decisions that a general-purpose chain cannot make. Celestia uses data availability sampling (DAS) with light nodes that verify block availability by checking random data shares rather than downloading entire blocks. More light nodes joining the network actually increases the safe block size, creating a scaling dynamic where participation improves throughput.

Celestia's current mainnet processes blocks between 2 MB and 8 MB with 6-second block times. Testnet benchmarks have pushed far beyond that: the mamo-1 devnet achieved 128 MB blocks at 21.33 MB/s throughput in April 2025, over 16 times the live mainnet capacity. The long-term roadmap targets 1 GB blocks through innovations like the Vacuum data-propagation algorithm and sharded validation. Celestia also recently announced Fibre Blockspace, a parallel DA protocol designed to sustain 1 terabit per second of throughput for rollups with extreme data needs.

The verification model differs from Ethereum's too. Celestia uses Namespaced Merkle Trees combined with fraud proofs rather than KZG validity proofs. This means DA finality includes a challenge period of roughly 10 minutes after block finalization, during which fraud proofs can be submitted. Ethereum's KZG approach provides immediate mathematical certainty at the cost of more computational overhead.

Head-to-Head Comparison

Cost data from Conduit's August 2025 analysis showed Celestia delivering roughly 55x cost savings per megabyte compared to Ethereum blobs for real-world rollup usage. That gap may narrow as Ethereum scales blob capacity through future upgrades, but Celestia's dedicated architecture gives it a structural cost advantage that full danksharding alone is unlikely to eliminate entirely.

What Rollup Developers Actually Weigh

The choice between these DA layers involves tradeoffs that go beyond raw throughput numbers.

Security inheritance remains Ethereum's strongest argument. Rollups posting data to Ethereum blobs inherit security from the network's approximately 500,000 validators and hundreds of billions in economic stake. Celestia's validator set, while growing, is orders of magnitude smaller. For rollups handling significant value, this security differential matters. Vitalik Buterin himself has argued that rollups using external DA layers should be classified as "validiums" rather than true rollups, reflecting reduced security guarantees.

Cost pressure pushes in the opposite direction. DA represents roughly 95% of what rollups pay to operate, according to Celestia's documentation. For data-intensive applications like gaming, social media, or high-frequency DeFi, the 55x cost difference between Celestia and Ethereum blobs can determine whether an application is economically viable. Arbitrum and other major rollups that fill their blob space efficiently pay less per megabyte, but smaller chains frequently waste capacity by posting partially filled blobs, amplifying the cost gap.

Sovereignty and flexibility attract a specific developer profile. Celestia enables sovereign rollups that maintain their own governance and can hard fork independently. Ethereum-settled rollups must work within Ethereum's upgrade cycles and governance decisions. Projects like Eclipse have chosen Celestia specifically because the high-throughput, low-cost DA removes bottlenecks for compute-intensive workloads spanning AI, gaming, and DePIN.

Ecosystem lock-in is the subtler consideration. Choosing Ethereum DA keeps rollups fully integrated with Ethereum's liquidity, tooling, and user base. Choosing Celestia introduces bridging complexity and dependency on a separate network's liveness and security. As Solana and other high-throughput Layer 1s compete for developer attention, the DA layer choice increasingly shapes which ecosystem a rollup aligns with long-term.

The Convergence Timeline

These approaches are converging from opposite directions. Ethereum is becoming more modular by separating data availability from execution through danksharding. Celestia is building bridges into Ethereum's settlement layer through Blobstream, which streams Celestia's data attestations to Ethereum smart contracts.

Full danksharding remains Ethereum's most ambitious scaling milestone. When implemented, it would increase Ethereum's native DA capacity by roughly 10x from current post-Fusaka levels. Whether that closes the throughput gap with Celestia depends on how aggressively Celestia's own roadmap executes, particularly the 1 GB block target and Fibre Blockspace initiative.

The market may not require a single winner. Rollups processing high-value financial transactions may prefer Ethereum's security guarantees despite higher costs. Data-heavy consumer applications may choose Celestia or use it alongside Ethereum in hybrid configurations. The EigenDA comparison adds another dimension, as committee-based DA layers offer yet different tradeoff profiles for rollup teams evaluating their options.

Competing Visions for Blockchain's Data Layer

Danksharding and Celestia represent fundamentally different bets on how blockchain infrastructure should evolve. Ethereum's approach preserves its position as a unified settlement and DA layer, accepting slower iteration in exchange for security consolidation. Celestia's approach assumes that specialization will always outperform general-purpose infrastructure on cost and throughput, betting that the security gap can narrow as the network matures.

For TIA holders, the thesis depends on whether enough rollup developers prioritize cost and sovereignty over Ethereum's security umbrella. Price forecasts reflect this uncertainty, with wide ranges based on adoption trajectory assumptions. The DA market is still early enough that both approaches can grow substantially before competing directly for the same rollup deployments.

Trade TIA on LeveX spot markets or access leveraged positions through TIA perpetual futures. Explore our Crypto in a Minute guides for coverage of other blockchain infrastructure projects shaping the modular ecosystem.