What is danksharding?

Danksharding is Ethereum's long-term plan for scaling through rollups. Instead of splitting Ethereum into many separate chains, it adds large amounts of cheap data space to each block, in the form of blobs, and lets nodes confirm that data was published without each downloading all of it. ethereum.org says the goal is a network that supports more than 100,000 transactions per second across layer 2s.

What is danksharding and why is it called that?

The name comes from Ethereum researcher Dankrad Feist, who proposed the design, according to Alchemy. The first stage, proto-danksharding, is named after both of its proposers, Protolambda and Dankrad Feist, ethereum.org notes.

Vitalik Buterin's FAQ sums up the core idea. Ethereum would provide space for blobs of data that the protocol does not try to interpret. Rollups execute their own transactions and simply need somewhere to post the data, so Ethereum only has to prove that data is available.

How does danksharding work, explained simply?

Three ideas fit together.

• One builder per block. Buterin explains that a single proposer chooses all transactions and data for a slot, with transactions and data sharing one fee market. Specialized builders compete to assemble each block, an arrangement called proposer-builder separation, because creating proofs for large amounts of data is heavy work.
• Commitments. Each blob comes with a small KZG commitment, a cryptographic fingerprint that lets anyone check pieces of the data against it. The secret values behind it came from a public ceremony with more than 140,000 contributions, ethereum.org says.
• Sampling. Validators check a few random pieces of the data rather than the whole thing. If enough samples succeed, the full data is almost certainly available. Our guide to data availability explains the maths.

What is the difference between proto-danksharding and danksharding?

Proto-danksharding, EIP-4844, went live in March 2024. Buterin's FAQ says it delivered the blob transaction type and its fee logic but left out the hardest parts, including real data sampling and proposer-builder separation.

PeerDAS, part of the Fusaka upgrade scheduled by the Ethereum Foundation for December 3, 2025, added the first sampling. ethereum.org describes it as a step toward full danksharding. It extends each blob on its own, in one dimension, and spreads the pieces across nodes. Full danksharding would encode the whole set of blobs in two dimensions for stronger redundancy and raise the limit to 64 blobs per block. ethereum.org says it still needs substantial research, so this reflects the roadmap as of November 2026.

How is danksharding different from sharding?

Earlier sharding plans gave each shard its own blocks and proposers. Danksharding has one proposer and does not split transaction processing at all. It only spreads out the job of storing and checking data. The wider history of the idea, including networks that do shard execution, is in our guide to sharding in blockchain.