EIP-4844 introduced blob-carrying transactions to Ethereum and is the first installment of full Danksharding, hence the name proto-danksharding. The proposal activated on March 13, 2024 at slot 8,626,176 with the Dencun hard fork, and gave rollups a new transaction type that carries large data blobs at substantially lower cost than calldata. The result was a near-immediate ~90% drop in Layer 2 data posting fees, observable on dashboards from the first day of activation.
​

Blob data does not live in execution payloads. Each blob is roughly 128 KB, and a block can carry up to six blobs (with a target of three). Blobs are visible to consensus clients for 18 days, after which they are pruned. This ephemeral storage is what makes blobs cheap; rollups only need data availability long enough for fraud or validity proofs to settle.
​

EIP-4844 is a Core Standards Track EIP authored by Vitalik Buterin, Dankrad Feist, Diederik Loerakker, George Kadianakis, Matt Garnett, Mofi Taiwo, and Ansgar Dietrichs. The full EIP-4844 specification defines the new transaction type, the blob fee market, and the cryptographic commitments that link blobs to transactions.
​

The change is a precursor to Danksharding, the longer-term scaling design that will expand Ethereum's data availability throughput by orders of magnitude. Proto-danksharding ships the transaction format and fee market without yet implementing data availability sampling (DAS), which is the cryptographic technique that lets light clients verify availability without downloading every blob.
​

A Type 3 transaction (introduced by EIP-4844) carries blob commitments alongside its standard execution payload. Each commitment is a KZG polynomial commitment to a 4,096-element field, derived from the trusted setup ceremony that over 140,000 contributors participated in during 2023. The commitment is what gets stored in the EVM-visible execution payload; the actual blob data sits in the consensus layer.
​

Smart contracts cannot directly access blob data. They can read the commitment and verify a KZG proof against it using the new BLOBHASH opcode and a precompile at address 0x0A. This separation is intentional. If contracts could load blob bytes, blobs would have to live in execution storage forever, defeating the cost reduction.
​

The flow for a rollup like Arbitrum or Optimism:

EIP-4844 introduced a separate fee market for blob gas, parallel to the EIP-1559 fee market for execution gas. Each blob costs blob gas (initially priced at 1 wei per blob gas), and the price adjusts per block based on a target of three blobs per block.
​

The blob base fee uses the same exponential adjustment as EIP-1559's base fee, with a different decay factor. If blocks consistently include more than three blobs, the blob fee rises; if fewer, it falls. The minimum is 1 wei, which is where the blob fee sat for most of 2024 because demand was below the 3-blob target.
​

The result is asymmetric pricing. During quiet periods, blob inclusion is essentially free for rollups, which translates directly into lower L2 fees for users. During congestion — typically driven by airdrop seasons or memecoin trading on rollups — blob fees can rise into double-digit gwei territory. The Hildobby Dune dashboard tracks blob count, fee, and per-rollup usage in real time.
​

The cost reduction was immediate and large. Before Dencun, rollups posted batches as calldata, paying full execution gas for every byte. Calldata gas cost 16 gas per nonzero byte. A typical Arbitrum batch consumed several million gas at the prevailing base fee.
​

After Dencun, the same batch posts as a blob at near-zero cost. Median L2 transaction fees on Arbitrum, Optimism, Base, and Linea dropped from $0.20-$1.00 ranges to $0.01-$0.10 within the first week. L2fees.info tracks the moving averages.
​

The cost savings are not evenly distributed. Rollups with high transaction throughput — Base and Arbitrum, which together account for roughly 70% of L2 daily transactions in mid-2024 according to L2BEAT — captured the largest absolute cost reduction. Smaller rollups benefit from cheaper data posting but the per-user impact depends on their batching cadence.
​

EIP-4844 ships the format and fee market for blobs. Full Danksharding will scale blobs by an order of magnitude through data availability sampling. Under DAS, the network distributes blob shards across nodes such that any individual node only stores a fraction. Light clients can probabilistically verify availability by sampling random shards.
​

The roadmap published by the Ethereum Foundation targets full Danksharding for the post-PeerDAS phase. Vitalik's proto-danksharding FAQ walks through the staged plan: from 3 blobs/block today to 16, 32, and eventually 64+ blobs/block as DAS lands incrementally.
​

The Pectra upgrade (Prague + Electra), expected in 2025, includes EIP-7691 to raise the target from three to six blobs per block and the cap from six to nine. This staged increase is calibrated to avoid overwhelming consensus client bandwidth before DAS provides headroom.
​

EIP-4844's cryptographic core is the KZG (Kate-Zaverucha-Goldberg) polynomial commitment scheme. A KZG commitment compresses a polynomial of degree 4,095 into a single point on an elliptic curve, and lets anyone prove that a specific value belongs to that polynomial without revealing the rest.
​

The commitment scheme requires a one-time trusted setup ceremony to generate parameters. The Ethereum KZG ceremony ran throughout 2023, with over 140,000 participants contributing randomness. The ceremony was structured so that as long as a single participant was honest and destroyed their contribution, the resulting parameters are secure. Both Vitalik Buterin and the Ethereum Foundation participated, but no single participant's honesty is required.
​

The participation count was the largest of any cryptographic ceremony at the time. Anyone could contribute via a browser, a CLI, or a hardware device. The full ceremony transcript is publicly verifiable on the ceremony.ethereum.org page.
​

For developers, the practical implication is that blob commitments are deterministic and verifiable. A rollup's smart contract can verify a KZG proof against a commitment using the precompile at address 0x0A. The verification cost is fixed and modest — far cheaper than the equivalent calldata cost.
​

Three claims show up in developer threads that miss how blobs actually work:

Stablecoin transfers across L2s benefit directly from EIP-4844. USDC and USDT activity on Arbitrum, Base, and Optimism saw aggregate transaction counts rise sharply in the months following Dencun, partly attributable to fees that are an order of magnitude lower. Artemis stablecoin metrics tracks the per-chain volume.
​

For cross-chain stablecoin orchestration, lower L2 posting costs reduce the floor on a profitable solver fill. A solver routing USDC from Ethereum mainnet to a destination L2 needs to cover the destination-side gas. Pre-EIP-4844, that floor was high enough that small transfers were uneconomical. Post-EIP-4844, the threshold for a profitable cross-chain fill dropped meaningfully, expanding the addressable market for retail-sized transfers.
​

Eco's cross-chain stablecoin swap infrastructure integrates rollups whose blob-driven fee profile makes routine $50-$500 transfers viable. Without EIP-4844, the same routes would carry destination gas costs that consumed too much of the principal.
​

EIP-4844 activated on March 13, 2024 with the Dencun hard fork, at slot 8,626,176. The activation required every Ethereum execution and consensus client to ship coordinated updates. All major clients — Geth, Nethermind, Erigon, Besu, Reth, and on the consensus side Prysm, Lighthouse, Teku, Lodestar, Nimbus — were ready before the fork epoch.
​

Blobs are visible to consensus clients for 18 days (4,096 epochs), after which they are pruned. The KZG commitment remains in the execution layer indefinitely. Long-term blob archival is handled by external indexers and rollup teams, not by the Ethereum protocol itself.
​

Most consumer wallets do not expose Type 3 transactions because the blob format is targeted at rollups posting batches. Specialized tools and SDKs — including Blossom Labs' Blobby and Foundry's cast command — can construct blob transactions for testing and posting arbitrary data.
​

During typical 2024 conditions the blob base fee sat at 1 wei (the minimum), making each 128 KB blob effectively free. Under congestion, blob fees have spiked to several gwei. Even at peak congestion, posting a blob is far cheaper than the equivalent calldata.
​

Indirectly. Standard Type 0, 1, and 2 transactions still pay the regular EIP-1559 base fee. EIP-4844 added a parallel blob fee market that does not interact with execution gas pricing. Block proposers still earn priority fees from execution transactions. The blob lane is a separate accounting space.
​