Our earlier article on Base fees explains that a Base transaction pays for two things at once: cheap execution on Base itself, and a smaller fee for anchoring a compressed record of that transaction back on Ethereum. That second piece, the L1 data fee, used to be a much bigger share of the total than it is today. The reason it shrank so much comes down to a single Ethereum upgrade called Dencun, and a new kind of data it introduced called a blob.
The problem blobs were built to solve
Before March 2024, when Base bundled up a batch of transactions and posted them to Ethereum for safekeeping, it stored that data as ordinary calldata. Calldata sits inside a regular Ethereum transaction, and Ethereum treats it the same way it treats everything else on chain: permanently, verified by every node, priced accordingly. That permanence is expensive, because it means the data has to be kept and re-verified forever, even though rollups like Base only really need it to be available for a short window while anyone who wants to can double check the batch.
Paying permanent storage prices for temporary data was the core inefficiency, and it was the single biggest driver of fees on rollups like Base, Optimism, and Arbitrum.
What a blob actually is
EIP-4844, often called proto-danksharding, introduced a new kind of data that rides along with an Ethereum block without going through the same expensive path as calldata. Each blob holds around 128 kilobytes of data, and unlike calldata it is not processed by the Ethereum virtual machine at all. Nodes store it, make it available so anyone can verify a rollup's batch against it, and then discard it after roughly 18 days.
Eighteen days is far longer than anyone actually needs to check a batch in practice, but nowhere close to forever. That difference in lifetime is exactly what makes blobs so much cheaper than calldata: Ethereum is charging for temporary shelf space instead of a permanent archive.
Blobs have their own fee market, separate from the regular gas market used for normal transactions. It works on the same EIP-1559 style logic already covered in our fee guide: a base fee that rises when more than the target number of blobs are included in a block, and falls when demand is lighter. Because that market is walled off from ordinary Ethereum gas, a spike in NFT minting or DeFi activity on Ethereum mainnet does not automatically make blob space, and therefore Base's L1 data fee, more expensive. The two markets can move independently.
What changed for Base users
Ethereum activated the Dencun upgrade, which included EIP-4844, in March 2024. Base supported it from day one. The effect was immediate: rollups that switched from posting calldata to posting blobs saw their L1 data fees drop by roughly an order of magnitude within days, in many cases more than 90 percent lower than before.
Nothing about how you use Base changed. You still connect a wallet, still approve and swap and send the same way, and your wallet still shows you one combined estimate before you confirm. What changed is invisible: the portion of your fee that pays for anchoring your transaction back to Ethereum got dramatically cheaper, because it is now measured against blob prices instead of calldata prices.
This is also part of why the L1 data fee, while still the more variable of the two components in a Base transaction, moves in a narrower range than it used to. Blob space has its own demand curve driven by every rollup competing for the same limited blob slots per block, rather than sharing a fee market with every other kind of Ethereum activity.
Why this matters beyond the fee number
Blobs are also a preview of where Ethereum's scaling roadmap is heading. Proto-danksharding is explicitly described as a stepping stone toward full danksharding, a future design meant to support many more blobs per block and push rollup costs down further as usage grows. Base and other rollups do not need to do anything special to benefit from that future work either, since it plugs into the same blob mechanism that already exists today.
For anyone using Base day to day, the practical takeaway is simple. A meaningful part of what you pay in fees is not really a Base decision at all, it is a reflection of how Ethereum prices a scarce resource, blob space, that Base and its peer rollups all compete for. When blob demand across all rollups combined is low, that slice of your fee is close to nothing. When many rollups are posting large batches at once, it ticks up a little, though nowhere near as sharply as calldata pricing used to move.
The short version
Before March 2024, Base paid Ethereum's permanent storage prices for data it only needed to keep available for a couple of weeks. EIP-4844 introduced blobs, a cheaper, temporary form of data storage with its own separate fee market, and Base adopted it immediately. The result was an immediate, large drop in the L1 portion of Base transaction fees, and a fee that now tracks blob market demand rather than the price of ordinary Ethereum gas. You do not have to do anything differently to benefit from any of this. It already happened in the background, and it is a big part of why swapping on Base costs a fraction of a cent instead of several dollars.