Since EIP-4844 shipped in March 2024, Ethereum’s blob fee market has done something the base layer rarely manages cleanly: it has created honest price competition between Layer 2s, and the market is responding in ways that weren’t fully anticipated.
Blobs - the temporary data containers that rollups use to post transaction data to Ethereum - have their own fee market, separate from execution gas. When blob space is scarce, fees rise. When it’s empty, posting data to Ethereum is nearly free. For most of 2024, blobs were cheap enough that L2 transaction costs fell dramatically across Arbitrum, Base, and Optimism. Users noticed.
What’s becoming clearer in 2026 is that blob pricing is now a meaningful differentiator in rollup strategy. Chains with higher throughput and more efficient data compression - posting fewer bytes per transaction - can absorb fee spikes without passing costs to users. Chains that post data less efficiently get squeezed when blob demand rises.
Base Has Become the Stress Test
Coinbase’s Base network has grown into one of the highest-volume blob consumers on Ethereum, driven partly by Aerodrome’s trading activity and a wave of consumer-facing apps. During periods of elevated blob demand - typically coinciding with market volatility or token launches - Base’s transaction fees have spiked noticeably compared to competitors like zkSync Era, which uses ZK proofs to compress data more aggressively.
This is the kind of structural pressure that EIP-4844’s designers anticipated: over time, rollups have an economic incentive to improve their data efficiency, or accept that users will migrate to competitors when fees climb.
The Road to Full Danksharding
Ethereum’s longer-term roadmap calls for full danksharding, which would expand blob capacity significantly by distributing data storage across validators using a technique called data availability sampling. The current implementation - sometimes called proto-danksharding - is a stepping stone. Ethereum researchers have indicated that full danksharding remains several years out, pending progress on PeerDAS, the peer-to-peer layer that underpins the sampling mechanism.
Until then, the blob market stays constrained. That constraint is doing real work: it’s forcing L2 teams to treat data efficiency as an engineering priority rather than an afterthought, and it’s giving users a fee signal they can actually act on. The base layer, for once, is applying pressure in the right direction.