Ethereum's rollup ecosystem set a throughput record this week, days after an upgrade raised the number of blobs the network targets per block. To understand why that single parameter matters, you have to understand what a blob actually is — because it is the hinge on which Ethereum's entire scaling strategy turns.
What a blob is, and why rollups live and die by it
Rollups execute transactions off-chain, then post a compressed batch back to Ethereum so anyone can reconstruct and verify the result. The expensive part was never the computation — it was making that batch data available. Before last year's blob upgrade, rollups had to buy permanent, general-purpose storage in the same fee market as every other transaction, so a busy day on Ethereum made every Layer-2 expensive at the same time.
Blobs broke that coupling. They are a separate, temporary data lane — the data is guaranteed available for roughly two weeks, long enough for anyone to challenge a fraudulent batch, then pruned. Rollups don't need it forever; they need it long enough to prove honesty. Crucially, blobs have their own fee market, so rollup costs are now largely insulated from congestion on Ethereum's execution layer.
Why raising the target changes the economics
Blob pricing works like Ethereum's EIP-1559 gas market: there is a target number of blobs per block and a base fee that rises exponentially when demand exceeds it, falls when it doesn't. For most of the past year rollups routinely bumped against the ceiling, which meant the blob base fee kept spiking and the promised savings never fully materialized. Raising the target gives that pressure valve more room — demand now sits comfortably below capacity, so the base fee stays near its floor.
The result shows up in users' wallets as cents rather than dollars. Median fees on the leading rollups fell below a cent during off-peak hours. That is not a marginal improvement; it is a threshold. At sub-cent cost, entire categories that were purely theoretical — micropayments, fully on-chain games, high-frequency social — cross from "demo" to "deployable."
The demand question, and the roadmap it depends on
The honest uncertainty is elasticity. Does cheaper capacity summon genuinely new demand, or does it just make the same transactions cheaper for the same users? History leans toward the former — every prior drop in transaction cost eventually filled the new headroom — but there is no law guaranteeing it repeats at this scale.
There is also a ceiling to how far this particular lever goes. Every full node still downloads every blob today, so the target cannot be raised indefinitely without pricing out ordinary node operators and quietly centralizing the network. The long-term fix — data-availability sampling, where nodes verify that data exists by checking small random pieces rather than downloading all of it — is the upgrade that turns today's incremental gains into a genuine order-of-magnitude one. This week's record is a preview of that future, not the arrival of it.
