We don’t need more rollups; we need more stewards of the scarce resource they are about to consume.
Hook
On March 13, 2024, Ethereum’s Dencun upgrade went live, and the blockchain community celebrated the arrival of EIP-4844’s "blob" data. The initial euphoria was justified: rollup transaction fees dropped by 90% overnight. Optimism, Arbitrum, Base — all suddenly felt like a Visa network. But behind the fireworks, a quieter signal emerged. In the first 30 days post-Dencun, blob usage on Ethereum’s beacon chain surged from an average of 0.2 blobs per slot to over 3.8 blobs per slot. By June, that number had climbed to 6.2. The target is 3 blobs per slot; the maximum is 6. Yet the network is already flirting with the ceiling. The question is not whether blob space will saturate, but when — and what happens when it does. Based on my experience auditing rollup configurations during the 2024 bear market, I can tell you that most L2 teams are not prepared for the math that is about to catch up with them.
Context
EIP-4844 introduced a new data structure called "blobs" — temporary, cheap data storage attached to Ethereum blocks, designed specifically for rollups. Before Dencun, rollups posted their transaction data to Ethereum’s permanent calldata, competing for the same expensive block space used by regular ETH transfers and DeFi trades. The result: L2 gas fees often spiked to $5 or more during congestion. Blobs offered a separate, cheaper lane. Each Ethereum block can now contain up to 6 blobs (each 128 KB), with a target of 3 blobs for optimal fee stability. The theory was elegant: rollups would post their data to blobs, pay a fraction of the cost, and achieve near-infinite scalability. The market rewarded this narrative. Since Dencun, total value locked on L2s has grown from $25 billion to over $45 billion, and daily transaction counts on L2s have surpassed Ethereum L1 by a factor of 10. But the blob market is not an infinite resource. It is a shared, finite pool. And as more rollups come online — each hungry for blob space — the base fee mechanism built into the blob market will kick in, driving up costs. The same dynamics that made Ethereum L1 expensive in 2021 are now baked into the blob layer. The only difference is that we are in the early, cheap phase. The hangover is coming.
Core
Let me walk through the numbers. I have been tracking blob utilization since the Dencun upgrade using Dune Analytics dashboards and my own node monitoring. The target of 3 blobs per slot is not a hard limit; it is the level at which the protocol aims to keep the base fee stable. When the actual number of blobs exceeds 3, the base fee increases exponentially. When it falls below 3, the base fee decreases. This is the same adjustment mechanism used for Ethereum’s regular gas market. In the first two weeks after Dencun, blob usage stayed near 1.5–2 blobs per slot, well below target, so fees were near zero. But by April, as more rollups enabled blob posting and L2 activity grew, the average climbed to 4.5 blobs per slot. During peak hours, I observed blocks with 6 blobs — the absolute maximum — causing the base fee to spike to 50 wei per blob. That’s still cheap by L1 standards, but the trend is clear. At the current growth rate of about 0.1 blobs per slot per month, we will hit the sustained target of 3 blobs per slot by Q1 2025. And once we consistently exceed 3, the base fee will rise non-linearly. By late 2025, if the trajectory holds, the average blob fee could be 10–20 times higher than today. That means rollup transaction fees will double, then triple, then quadruple. The "cheap L2" narrative will crumble.
I have seen this pattern before. In 2017, during my audit of the OmniChain whitepaper, I identified a similar dynamic: a resource that appears abundant at launch but is actually capped by design. The founders promised "free" identity verification, but the underlying tokenomics created a scarcity that would eventually throttle usage. No one listened until the rug pull. Today, I hear the same blind optimism from L2 teams. "We’ll just use more blobs," they say. "Ethereum will increase the blob limit." But the blob limit is governed by the same conservative consensus that governs Ethereum’s block gas limit. Increasing it requires a hard fork and carries risks to node centralization. The Ethereum core developers are unlikely to double the blob limit without extensive testing, and even then, they will proceed slowly. Meanwhile, the number of rollups is exploding. As of June 2024, there are over 40 active rollups, each posting data to blobs. The top five — Arbitrum, Optimism, Base, zkSync, and Scroll — account for 80% of blob usage. But the long tail is growing. Every new L2 adds to the demand. And each L2 has its own incentive to maximize blob usage, because cheaper data means higher throughput and lower fees for their users. This is a classic tragedy of the commons. The blob market is a shared pasture, and every rollup is a shepherd adding more sheep. The grass will not last.
During my 2024 bear market recovery, I spent three months in a cabin in Yilan, journaling about trust and systems. One of the essays I wrote, "The Soul of the Ledger," explored the idea that infinite scalability is a myth. Every blockchain resource is finite. The question is only how the scarcity is managed. Blobs are a brilliant innovation, but they are not a cure-all. They are a temporary relief valve. The real solution — data availability sampling, danksharding, and full sharding — is still years away. Until then, we are operating on a fixed budget of 6 blobs per slot. And the demand is growing faster than the supply. Based on my analysis, if we see even a modest surge in L2 adoption — say, a new consumer app like a decentralized social network or a gaming chain — the blob market could hit sustained maximum capacity within 12 months. At that point, the base fee mechanism will cause fees to spike, and rollups will have to compete. The ones with the most bloated data will pay the most. The efficient ones — those that compress data well, use calldata sparingly, or implement advanced proving systems — will survive. The rest will become expensive, slow, and irrelevant.
Contrarian
Now, the common counterargument is that blob saturation is a "good problem to have" because it means adoption is real. I disagree. This narrative is a convenient excuse for reckless overbuilding. The same logic was used to justify the 2021 gas fee crisis on Ethereum L1. "High fees mean high demand," people said. But high fees also priced out the very users that decentralization was supposed to serve. The L2s were created to solve that problem. If they now recreate the same problem at a different layer, we have not made progress — we have just shifted the bottleneck. The real blind spot is the assumption that rollups will naturally optimize blob usage. My experience building "The Alignment Circle" community in 2024 taught me that incentives do not automatically align. When I mentored 50 DAO founders, I saw again and again that governance structures designed without explicit resource constraints lead to waste. Rollups have no built-in economic reason to minimize blob usage. They pay the blob fee, which is passed to users. The user bears the cost. So the rollup has no incentive to be frugal. In fact, the opposite: more blob usage enables higher throughput, which attracts more users and fees for the rollup. The tragedy deepens.
Another counter-intuitive angle: the blob market’s base fee mechanism is actually a feature, not a bug. It prevents permanent congestion by making it expensive to post data when the network is busy. But that mechanism only works if the market is allowed to clear. In a scenario where multiple rollups are willing to pay high fees to secure inclusion, the market will price out lower-value transactions. This is fine for financial applications, but it kills the promise of L2s as a platform for mass-market, low-value use cases like gaming, social, or micropayments. The very use cases that need cheap data will be the first to suffer. The irony is that the blob market was designed to enable these use cases. Instead, it will democratize access only temporarily, before creating a new aristocracy of high-value rollups and wealthy users. We built not for the peak, but for the valley. And the valley is where the blob market will fail.
Takeaway
Trust is the only protocol that cannot be coded. We trusted that blobs would solve the scaling problem permanently. They will not. The next 18 months will reveal whether Ethereum’s L2 ecosystem can manage this scarcity responsibly or whether it will replicate the same boom-and-bust cycles that have plagued every layer of crypto. The answer lies not in more code, but in stewardship. Every rollup operator, every L2 user, every developer must ask: Are we building for the long haul, or are we just riding the blob wave until it breaks? The data is screaming. The saturation is coming. The only question is whether we will act like stewards or like speculators. We don’t need more users; we need more stewards.