In the second week of April 2025, the average blob inclusion cost on Ethereum Mainnet crossed 0.02 ETH—a 300% increase over the post-Dencun baseline. This is not a temporary spike caused by a single NFT mint or a memecoin frenzy. It is the first structural signal that the blob space, once hailed as the infinite highway for Layer2 scaling, is approaching its capacity ceiling far faster than the optimistic projections of March 2024.
I have been tracking blob usage since the Dencun upgrade went live. At first, the numbers were almost too good to be true. Blob gas was cheap, averaging under 1 gwei per blob for the first three months. Rollups were posting data at a fraction of the cost of calldata, and the narrative of a unified, scalable Ethereum seemed vindicated. But the data from Dune Analytics tells a different story when you look at the trend lines. The average number of blobs per slot has risen from 2.1 in April 2024 to 6.8 in March 2025. The target is 3 per slot, and the maximum is 6. We are already exceeding the target regularly, and the mechanism is designed to increase the blob fee to penalize congestion. The elasticity is not infinite. In fact, the supply curve of blob space is nearly vertical once we hit the maximum number of blobs per slot. The only way to accommodate more rollups is to raise the price.
This is where the economics become uncomfortable. The EIP-4844 specification introduced a separate blob gas market with a 15-minute half-life fee update rule. That means if demand stays high, the fee can rise exponentially within a few hours. We saw a 300% increase in two weeks, and that was with no major protocol launches. Imagine what happens when Base, Arbitrum, Optimism, zkSync, Linea, Scroll, and the next generation of app-specific rollups all compete for the same 6 blob slots. The cost will not just double—it will decouple from the underlying L1 gas price and become a premium for exclusive access to Ethereum’s data availability layer.
From my experience auditing DeFi protocols during the 2017 ICO boom, I learned that the crowd always underestimates the cost of shared infrastructure. Back then, it was the Ethereum network itself that became congested, pushing gas fees to $50 per transaction. Today, the same pattern is repeating at the blob level, but the consequences are even more pernicious because the promise of Layer2 was specifically that it would make transactions cheap forever. That promise is now being tested.
Let’s look at the numbers concretely. The total blob capacity is 6 blobs per slot (12 seconds) under ideal conditions, with a target of 3. Each blob is 128 KB, so the maximum throughput is 384 KB per 12 seconds, or 2.76 MB per minute. That sounds like a lot, but consider that a single rollup transaction can be as small as 12 bytes, but the batch overhead is large. In practice, the number of transactions per blob varies dramatically. Even if we assume 500 transactions per blob (a generous average), the global throughput of all rollups combined is about 3,000 transactions per slot, or 250 transactions per second. That is not nothing, but it is far from the millions per second that some proponents claim. And when the blob fee rises, the cost per transaction on the rollup will increase proportionally. A rollup that currently pays 0.001 ETH per blob might see that cost go to 0.01 ETH, and then 0.1 ETH. The end user will feel it.
Governance is not a vote, it is a vigil. The current governance of the blob market is entirely algorithmic. There is no human committee to decide which rollup gets priority. That is by design, but it also means there is no way to prioritize public goods over private profit. A rollup that pays the highest fee will outcompete a rollup that is building a sovereign identity for a developing nation. The market does not discriminate. In the chaos of summer, we found our winter soul—but if we do not prepare for the winter of blob costs, the thaw will be bitter.
The contrarian angle is this: The dominant narrative today is that Layer2 scaling is solved, and that we are entering a post-scaling era where fees are negligible forever. But that narrative ignores the hard constraint of shared data availability. The Ethereum roadmap assumes that we will eventually move to Dank Sharding, which would increase the number of blobs per slot dramatically. But that is years away, and the timeline is uncertain. Meanwhile, the number of active rollups is growing exponentially. According to L2Beat, over 50 rollups are now in production, and at least 30 more are in development. Each one wants its share of blob space. The result is a classic tragedy of the commons: each individual rollup has an incentive to post as much data as possible, but the collective result is congestion and rising costs.
Silence in the bear market is where truth compiles. During the bear market of 2022-2023, when I was isolated in a cabin in County Wicklow, I wrote about the quiet strength of on-chain truths. One of those truths is that Ethereum’s data availability layer is not a free good; it is a shared resource that must be stewarded. The Dencun upgrade was a brilliant engineering feat, but it did not change the fundamental economics of scarcity. It only shifted the bottleneck from the execution layer to the data layer. We are now seeing the first real-time signal of that shift.
What does this mean for the average user? It means that the cheap fees on Arbitrum and Optimism today are not guaranteed for the long term. If you are building a dApp that relies on low-cost transactions, you should model a scenario where the L2 fee increases by 5x or 10x within the next 18 months. That is not a FUD; it is a mathematical consequence of the blob supply curve. I have seen this pattern before—in 2017, when I audited the EtherSwap protocol and discovered that its governance mechanism allowed whales to bypass consensus. The community ignored the warning signs until the system collapsed under its own weight. The blob market will not collapse, but it will become uncomfortable. And when it does, the projects that have prepared for higher costs will survive; the ones that assumed infinite cheapness will fail.
We do not build walls, we weave nets of trust. The solution is not to abandon blob space, but to design for scarcity. That means encouraging rollups to use compression, or to move to dedicated data availability layers like Celestia or EigenDA. It also means that Ethereum itself must prioritize the Dank Sharding upgrade, and the community must hold the core developers accountable for delivering it. The blob market is a mirror of our collective reliance on shared infrastructure. If we do not design for scarcity, the cost of trust will become a luxury only the few can afford.
Let me be clear: I am not advocating for a return to monolithic chains. Layer2 is the right path. But we must be honest about the constraints. The post-Dencun euphoria has blinded us to the reality that scalability is not free—it is a trade-off between cost, speed, and decentralization. The blob fee data is the first signal that the trade-off is tilting. We have about 18 months before the fee curve becomes exponential. The question is not whether we will see higher fees, but whether we will act before they become a crisis.
Code is law, but conscience is the compiler. In the world of rollups, the compiler is the blob fee market. It is impartial, but it is not just. It rewards the highest bidder, not the most innovative project. As an architect of DAO governance, I have learned that the best systems are those that anticipate failure and build in resilience. The blob market needs a human-in-the-loop mechanism—a governance layer that can adjust the target blob count or introduce priority queues for public goods. Without it, we are simply automating a tragedy.
I will end with a thought that has guided my work since the ethical audit of The DAO clone: The best technology is not the one that scales the fastest, but the one that scales with integrity. The blob saturation signal is a warning. Heed it, or watch the cost of trust become unbounded.