Proto-Danksharding Explained: EIP-4844 Blobs and the Dencun Upgrade

If your Arbitrum or Base transactions suddenly got dirt cheap in 2024, you have proto-danksharding to thank. That's the real story here, not some press-release upgrade name. Proto-danksharding (shipped as EIP-4844) added a new kind of cheap, temporary storage to Ethereum called blobs, and rollups have been dumping their data into blobs ever since. Fees on Layer 2 dropped hard. This post explains what blobs actually are, why they made L2 cheaper, and where the dreaded "full danksharding" fits in.

Last updated: June 2026.

I'm going to keep the marketing brain off and the engineer brain on. No "significant milestone," no roadmap worship. Just what changed and why your wallet noticed.

So what is proto-danksharding, in plain English

Ethereum mainnet is expensive because every byte you post there gets stored by every node, basically forever. Rollups like Arbitrum, Optimism, and Base do all their heavy lifting off-chain, then post a compressed batch of their transaction data back to Ethereum so anyone can verify it. That data-posting step was the single biggest cost a rollup paid, and it got passed straight to you.

Proto-danksharding's whole trick is this: give rollups a cheaper place to dump that data that nobody needs to keep forever.

Before EIP-4844, rollups posted their data as calldata. Calldata is permanent and pricey, 16 gas per non-zero byte, sitting on-chain for eternity even though nobody reads month-old rollup batches. Insane waste. Blobs fixed that by being temporary by design.

"Proto" is the giveaway. This is the prototype, a stepping stone toward full danksharding. It ships the data structure (blobs) and the separate fee market without the heavy scaling machinery that comes later. Small slice now, big version later.

What are blobs in Ethereum (EIP-4844)?

A blob is a chunk of data attached to a transaction that lives on the consensus layer for a short window and then gets deleted. Each blob is about 128 KB. The point is that the data is available long enough for anyone running a node to download it and verify a rollup's claims, but it doesn't squat on the chain forever.

Here's the side-by-side that actually matters:

	Calldata (old way)	Blobs (EIP-4844)
Cost	~16 gas per byte	Roughly an order of magnitude cheaper
Lifespan	Permanent	Deleted after ~18 days (4096 epochs)
Readable by smart contracts	Yes	No, only a commitment is accessible
Fee market	Shared with everything else	Separate blob gas market
Best for	Anything that needs permanent on-chain access	Rollup batch data

That "not readable by smart contracts" line trips people up. Contracts can't see the raw blob contents, only a cryptographic commitment (a KZG commitment) that proves the data existed and matched. For rollups that's perfectly fine. They just need the data to have been available so anyone could reconstruct and challenge the state. They don't need a contract to read it byte by byte.

The 18-day deletion is the part that makes blobs cheap. Permanent storage is the most expensive thing Ethereum sells. Rollup data goes stale fast. Once the data's been available long enough for someone to catch fraud or rebuild state, keeping it is pure dead weight. So blobs get pruned and the cost drops.

How do blobs reduce Layer 2 gas fees?

Roughly 80 to 90 percent of an L2 transaction's cost used to be the L1 data-posting bill. That's the chunk blobs attacked.

Blobs have their own independent fee market. There's a separate base fee for blob gas that adjusts up or down based on how many blobs people are using, totally decoupled from the regular gas market. So when Ethereum mainnet gets congested with a memecoin frenzy, blob fees don't necessarily spike with it. Rollups posting data aren't bidding against NFT mints for the same gas anymore.

When EIP-4844 went live, the effect was immediate and obvious. L2 fees on the major chains fell dramatically, often into fractions of a cent for simple transfers. The exact numbers move around constantly because they depend on blob demand, ETH price, and how busy the network is, so I'm not going to hand you a fake "exactly 92% cheaper" table. The honest version: data costs dropped by roughly an order of magnitude, and that's why the L2s you use got cheap.

One catch worth knowing. There's a target and a max number of blobs per block. When blob demand exceeds the target, the blob base fee climbs, and L2 fees can creep back up during heavy periods. Blobs made things cheaper, not free. Full danksharding is supposed to widen that ceiling.

If you want the bigger picture on where Ethereum's scaling story is heading, I went deep on it in why Ethereum's quiet price action is actually the bullish read.

What is the difference between proto-danksharding and full danksharding?

This is the comparison everyone botches, so let's be precise. Same family, very different scope.

	Proto-danksharding (EIP-4844)	Full danksharding
Status	Live since March 2024	Future, multi-year build
Blobs per block	A handful (target/max set low)	Up to 64-ish, far more space
Who stores the data	Every node downloads every blob	Nodes only sample pieces (data availability sampling)
Scaling goal	Cheaper L2 data, modest throughput	Massive blob space, huge throughput
Main new tech	Blob transactions + KZG commitments	Data availability sampling, PBS-style separation

Proto-danksharding makes every node download every blob. That works fine when there are only a few blobs per block. It does not work if you want dozens of megabytes of blob space, because you'd crush home node operators.

Full danksharding's headline feature is data availability sampling. Instead of every node grabbing every blob, each node randomly samples small pieces and uses math (erasure coding) to be statistically certain the whole thing is available without downloading it all. That's what unlocks the giant scaling numbers. It's genuinely hard to build, which is exactly why proto came first.

So: proto-danksharding is the version that's live and already cut your fees. Full danksharding is the much bigger, much harder upgrade still in the pipeline. Anyone telling you danksharding is "done" is confusing the prototype for the finished thing.

What the Dencun upgrade actually changed

Dencun is the hard fork that delivered EIP-4844. The name is a mashup of Deneb (the consensus-layer side) and Cancun (the execution-layer side), because Ethereum upgrades hit both layers at once now.

EIP-4844 was the headliner, but Dencun bundled several other EIPs. The ones worth knowing:

• EIP-1153 (transient storage): cheap temporary storage that exists only inside a single transaction. Useful for things like reentrancy guards and more efficient DeFi flows.
• EIP-4788 (beacon block root in the EVM): lets execution-layer contracts see consensus-layer state, which helps staking and bridging protocols trust validator data without sketchy oracles.
• EIP-6780 (SELFDESTRUCT changes): mostly neuters the old SELFDESTRUCT opcode, a cleanup step toward future state changes like Verkle trees.

Everything else in Dencun is plumbing. Blobs are the part that changed your fees and the part that's permanent. That's why I'd rather you remember "proto-danksharding" than "Dencun." The event was a Tuesday in 2024. The mechanism is forever.

Quick note on naming, because it confuses people: this is sometimes lumped in with "Ethereum 2.0" talk, which is a dead branding exercise at this point. I wrote about why the whole "Ethereum 2.0" label got retired if you're curious where that phrase went.

Why this matters for normal users

You don't need to run a node or write a rollup to care. The practical upshot:

Use Layer 2s for everyday stuff. Swapping, minting, bridging, gaming, all of it is far cheaper on Arbitrum, Base, Optimism, and the rest than on Ethereum mainnet, and proto-danksharding is a big reason why. If you're still doing small transactions on L1 out of habit, stop. You're lighting money on fire.

Don't expect fees to stay flat forever. Blob space is limited. When a hyped launch floods the L2s, blob fees rise and your transaction costs tick up. That's normal. It's the system working, not breaking.

This is educational, not financial advice. Crypto is volatile and you can lose money. Understanding how blobs cut fees doesn't tell you what to buy. It just tells you which rails are cheaper to ride.

If lower L2 fees made staking ETH more appealing to you, that's a separate rabbit hole. I compared the main liquid staking options in my breakdown of Lido versus Rocket Pool for staking ETH.

Frequently Asked Questions

What is proto-danksharding?

Proto-danksharding is the Ethereum upgrade (EIP-4844) that introduced blobs, a cheap temporary form of data storage. It's a prototype step toward full danksharding, shipping the blob structure and a separate fee market so Layer 2 rollups can post data far more cheaply than before.

What are blobs in Ethereum (EIP-4844)?

Blobs are roughly 128 KB chunks of data attached to transactions that live on Ethereum's consensus layer for about 18 days, then get deleted. Smart contracts can't read their raw contents, only a cryptographic commitment. They exist mainly to give rollups cheap, short-lived data availability.

How do blobs reduce Layer 2 gas fees?

Posting data to Ethereum was 80 to 90 percent of an L2 transaction's cost. Blobs give rollups a separate, much cheaper fee market with temporary storage instead of permanent calldata. Data costs fell by roughly an order of magnitude, which is why L2 fees dropped sharply.

What is the difference between proto-danksharding and full danksharding?

Proto-danksharding (live since 2024) adds a few blobs per block and every node downloads all of them. Full danksharding (future) scales to far more blob space using data availability sampling, where nodes only check random pieces. Proto is the small prototype; full is the big, harder version.

What did the Dencun upgrade change?

Dencun was the March 2024 hard fork that delivered EIP-4844 and several smaller EIPs. Its biggest change was introducing blobs to cut Layer 2 fees. It also added transient storage (EIP-1153), beacon block root access (EIP-4788), and SELFDESTRUCT cleanup (EIP-6780).

What are blobs mainly used for in EIP-4844?

Blobs are mainly used by Layer 2 rollups to post their transaction batch data to Ethereum cheaply. Rollups need that data to be temporarily available so anyone can verify or challenge their state, but they don't need it stored permanently, which is exactly what blobs provide.

When did EIP-4844 go live?

EIP-4844 went live with the Dencun hard fork on Ethereum mainnet on March 13, 2024. Blobs have been active and used by major Layer 2 rollups continuously since then, and they're now a permanent part of how Ethereum handles rollup data.