I used to think Ethereum's Dencun upgrade was the ultimate scaling fix. After analyzing on-chain data for the past six months, I've concluded that its data blobs will become the next scarce resource, driving L2 gas fees back up by 2026. Here is what the charts don't tell you.
The Dencun hard fork introduced EIP-4844 (proto-danksharding), creating a temporary data layer for rollups. Early results are euphoric: L1 data costs dropped by over 90% for many L2s, and transaction fees on Arbitrum and Optimism fell below $0.01. The narrative of "blob solution" is now gospel. But scarcity economics hasn't been abolished—only deferred.
How Blobs Work A blob is a temporary data chunk attached to a block, stored by validators for ~18 days. Each block has a target of 3 blobs and a maximum of 6. The blob gas price adjusts via a similar target-demand mechanism as EIP-1559. In the first two months post-Dencun, average blob usage stood at 1.5 per block—well below capacity. But usage is climbing exponentially as more rollups migrate.
The Hidden Saturation Timeline Let me walk through the math. Based on Dune Analytics data, the rolling 7-day average of blob count per block has grown at a compound rate of 8% per month since March 2024. At this rate: - Current usage (~2.2 blobs/block) will hit the target of 3 by late 2024. - By mid-2025, average usage will exceed 5 blobs/block, activating the base fee increase. - By early 2026, we will consistently hit the max of 6 blobs per block, causing persistent congestion.
This isn't a linear forecast. The upcoming EIP-7623 (blob capacity increase to 8–10) may push the timeline by a year, but the underlying structural issue remains: blob space is a fixed supply that must be shared by all L2s. The current euphoria ignores that demand growth outstrips capacity upgrades.
Why This Matters for L2 Economics Let's examine the cost structure. A typical rollup transaction requires 5–10 KB of blob data. At current blob gas prices (near zero), cost per tx is negligible. When usage hits the target, base fee will rise exponentially—just like L1 calldata did in 2021. My back-of-envelope calculation: with 4 blobs per block, a 10 KB tx would cost ~$0.10. At 6 blobs, it jumps to $0.50. That's a 50x increase from today.
For comparison, the consensus among L2 teams is that $0.10 per tx is acceptable for DeFi but unacceptable for gaming or identity. The bull case for L2 mass adoption (payments, social) depends on sustained sub-cent fees. If blob fees rise, those use cases become uneconomical again.
The Contrarian Lens: A Fixed Resource in a Boom The blockchain space loves to promise infinite scalability. But data availability is a real constraint. Dencun's blobs are an elegant temporary solution, not a permanent one. The narrative that "L2s now have free data" is dangerous. It leads to over-reliance on Ethereum's base layer for DA, while alternative DA layers (Celestia, EigenDA, Avail) are still maturing.
I've seen this pattern before. In 2020, when Compound's governance token crashed, everyone said "it's fine, the protocol is sound." But the psychological damage of lost savings changed how users trust DeFi. Similarly, the feeling of "blobs fix everything" will reset when fees spike. The question isn't if it happens—it's when, and how fast L2 players diversify their DA stack.
What This Means for the Rollup Roadmap The Ethereum Foundation's rollup-centric roadmap assumes that L2s will rely on Ethereum for DA indefinitely. But if blob space becomes a friction point, we will see a pivot to validium (off-chain DA with fraud proofs) or sovereign rollups that use alternative DA layers. This will fragment the ecosystem rather than unify it.
Data availability is the new bandwidth. The first L2s to integrate Celestia or EigenDA as a backup will have a competitive advantage when Ethereum blob fees rise. I've already seen several teams quietly testing these integrations. The public narrative hasn't caught up yet.
My Personal Experience During the 2020 DeFi summer, I audited the Solidity code of a yield aggregator that used Compound's oracle. I found a critical logic flaw in the liquidation mechanism that would have caused a bank run. I submitted the fix, but the team ignored it for weeks. When the crash came, they called it "black swan." It wasn't—it was predictable.
The blob saturation is similar. The data is public. The growth rate is clear. Yet most L2 teams are betting on Ethereum to increase blob capacity indefinitely. That's a bet on governance, not engineering. I prefer to bet on engineering.
Takeaway: Follow the Fear, Not the Chart If you can't measure the supply of blob space, you can't manage L2 fee expectations. The next bull run won't be in L2 tokens that rely on Ethereum blobs alone. It will be in DA-layer protocols that offer scalable, decentralized data availability independent of Ethereum's base layer.
The charts today show low fees and euphoria. The fear is hidden in the blob base fee curve. Don't wait until the spike to ask why you didn't see it coming.
"Follow the fear, not the chart." "If you can't measure it, you can't manage it."