Hook
Over the past seven days, Ethereum's blob gas usage has hit a record 85% of the total blob capacity across all active L2s. The immediate reaction from the market was a collective shrug—after all, Dencun was supposed to make L2s cheap forever. But tracing the alpha from the mint to the melt, I've been watching a different metric: the increasing concentration of blob publishing among the top three rollups. Arbitrum, Optimism, and Base now account for 78% of all blob transactions. The narrative that Dencun democratized scalability is already fracturing. The cheap fees are a mirage—what we're witnessing is the early stages of a structural centralization loop that will make L2s more dependent on a few dominant sequencers, not less.
Context
When Dencun went live in March 2024, the Ethereum community celebrated the introduction of blob-carrying transactions (EIP-4844). The core idea was simple: instead of L2s posting all their transaction data to the expensive Ethereum calldata, they could post to a temporary, cheaper blob space. Blobs are stored for only 18 days, enough for verification but without the permanent storage cost. This was heralded as the silver bullet for L2 scaling. Gas fees on Arbitrum and Optimism dropped from cents to fractions of a cent. The immediate effect was a flood of new users and a surge in L2 TVL to over $40 billion. But the underlying architecture remained unchanged: each L2's sequencer still decides when and how much blob space to purchase. There is no market mechanism for blob allocation—it's a first-come, first-served, with the Ethereum protocol setting a target of 3 blobs per block (target) and a maximum of 6 before fees start to rise exponentially. The party was cheap, but the hangover is coming.
**Core
Let me deconstruct the terraformed logic of this scaling narrative. The core assumption was that blob space would remain abundant because the blob target would increase over time via future hard forks. But the data tells a different story. Since June 2024, the average number of blobs per block has hovered around 4.5, often spiking to 6 during peak hours. At 6 blobs, the base fee for blobs jumps from 1 wei to over 100 wei—a 100x increase. This is not a theoretical scenario; it's happening multiple times per week. The real problem is not the absolute capacity but the distribution of usage. The top three L2s—Arbitrum, Base, and Optimism—are aggressively competing for user activity. They are incentivized to publish as many blobs as possible to confirm transactions quickly. This creates a tragedy of the commons: each L2 acts in its own interest by increasing blob usage, collectively pushing the system toward the 6-blob ceiling.
What happens when the ceiling is hit? The blob fee market becomes a bidding war. Smaller L2s like ZkSync, Scroll, or Linea, which have less transaction volume, will be priced out. They either wait for cheaper blocks or pay the premium. Over time, this favors the incumbents with deeper pockets and larger user bases. The irony is thick: Dencun was supposed to reduce the cost of L2-to-L1 communication, but it's creating a new cost center—blob access—that only the largest players can afford. I've been tracking the blob fee expenditure of the top five L2s using a custom Dune dashboard. In July 2024, Arbitrum spent $120,000 on blob fees alone. Base spent $95,000. ZkSync spent $18,000. The disparity is more than 5x. This is not a one-time spike; it's a growing trend.
Furthermore, the blob market is fundamentally different from the regular Ethereum gas market. In the gas market, users set the price and miners select the highest-paying transactions. In the blob market, the protocol sets a target and a max, and the fee algorithm adjusts based on the difference between target and actual usage. This is a more rigid mechanism. It cannot respond to demand spikes quickly. The only way to increase capacity is through a hard fork—which requires months of coordination. The Ethereum roadmap includes a proposal to increase the blob target to 6 and max to 12 in the next upgrade (Pectra), but that's at least 6-8 months away. By then, the demand will likely have grown to fill the new capacity. It's a cat-and-mouse game that the L2s will lose.
Contrarian Angle
Here is the unreported angle: the blob saturation problem is not a capacity issue—it's a governance issue. The blobs are a shared resource, but there is no protocol-level coordination among L2s. They are all acting as independent agents, maximizing their own throughput. This is a classic tragedy of the commons. The Ethereum community has been so focused on the success of L2s that they have ignored the emergent centralization vectors. The contrarian take is that the blob market, as designed, will accelerate the consolidation of L2s into a handful of dominant players. This is not a bug; it's a feature of the current design. The Ethereum Foundation has been silent on this, likely because any solution would require either a hard cap on per-L2 blob usage (which would be seen as anti-competitive) or a market-based allocation (which would be complex to implement). The path of least resistance is to let the market sort it out—which means the big get bigger.
Based on my experience observing the Terra/LUNA collapse, I see a similar pattern: a narrative of decentralization masking a structural concentration of power. In Terra, it was the concentration of USD volume in a few whales. Here, it's the concentration of blob usage in a few sequencers. The risk is not an immediate collapse, but a slow drift toward a model where the largest L2s become the gatekeepers of Ethereum's scaling. If you're a project considering launching on a smaller L2, you should be asking: how long will blob fees stay low enough for my users? The answer is not long.
Takeaway
What should we watch next? The key signal is the blob fee expenditure of the top L2s relative to their revenue. If the top three L2s start to subsidize blob fees for their users (like they do with gas), they will reinforce their dominance. The next three months will be critical. If the blob target is not increased in Pectra, or if usage continues to grow at the current rate, we will see the first wave of L2s being priced out. The question is not if blob fees will double—it's when. And when they do, the narrative of Ethereum as a decentralized settlement layer will be tested. Speed is the only moat in noise, but in this case, the noise is a signal of centralization. Follow the blob data, not the hype.