The Blob Clock Is Ticking: Why Post-Dencun Euphoria Masks a Looming Fee Spike for Rollups
Hook
In the ashes of Terra, we didn't learn to fear collapse—we learned to see the cracks before they break. Today, those cracks run through the heart of Ethereum's blob space. Since the Dencun upgrade went live on March 13, 2024, the daily blob count has surged from an average of 2,300 to over 18,000 by late April, according to on-chain data aggregated across Dune dashboards. At the current compound growth rate of roughly 12% week-over-week, we are on track to hit Ethereum's theoretical blob capacity limit—six blobs per block, or roughly 28,800 blobs per day—within 18 months. By Q3 2025, every single blob slot could be occupied, and when demand exceeds supply, the blob basefee will spike. Rollup gas fees, which have enjoyed a 90%+ reduction since Dencun, will double, triple, and then climb further. I've run the numbers, and the market is not priced for this.
Context
Dencun introduced proto-danksharding via EIP-4844, creating a separate data layer—blobs—for rollups to post transaction data. Before Dencun, rollups posted calldata to Ethereum's execution layer, competing with regular transactions for block space and driving L1 gas to absurd levels. Blobs are cheaper because they are stored only temporarily (about 18 days) and don't execute on the EVM. Each Ethereum block can include up to six blobs, each roughly 128 kB, giving a total daily capacity of about 1.1 GB. The design was intentionally scarce: a small buffer to start, with plans to increase to 8, 16, or 32 blobs in future upgrades. But the assumption among many analysts is that existing capacity is sufficient for years. That assumption is wrong.
Here's why: the growth of rollups is exponential. Base alone went from <1% of L2 transactions pre-Dencun to >40% by late April, and its blob usage is climbing. Arbitrum and Optimism are following. New rollups like Blast (which relies on blobs for its L3 strategy) and emerging zkEVMs are adding demand. Meanwhile, the blob target is only three per block (the equilibrium point for the basefee), and the limit is six. Once the target is consistently exceeded, the blob basefee starts to rise. Currently, we hover around 4.2 blobs per block on average—already above the target. The basefee has been erratic, spiking from a low of 1 wei to occasional peaks of 50 wei per blob gas unit. That's still dirt cheap—a few cents per posting—but the vector is clear.
Core
Let me walk through the arithmetic. Ethereum produces roughly 7,200 blocks per day (12-second slots). With a maximum of 6 blobs per block, that's 43,200 blobs per day absolute ceiling. Realistically, block space is never 100% efficient; let's assume 90% utilization yields ~38,880 blobs per day. Current usage is ~18,000 blobs/day. At a conservative 8% weekly growth (halving the observed rate), we hit 38,880 in about 11 months. If growth continues at 10% weekly—which is plausible given the launch of thousands of new rollups—we hit the ceiling in 7 months. That means by early 2025, we could see sustained periods where every block's blob slots are full, triggering the basefee adjustment mechanism.
The blob basefee is similar to EIP-1559: it increases when there is more than target usage (3 blobs per block), decreases when below. Currently, we are above target, so the basefee is slowly drifting upward. But the real spike will come when we hit the cap, and congestion becomes persistent. At that point, rollups will compete for scarce blob space, and the basefee will rise exponentially until either demand drops (rollups batching less frequently) or the network increases the cap via another hard fork. The last time Ethereum faced persistent L1 congestion, in 2021, gas fees reached thousands of gwei. Blobs will follow a similar pattern, albeit with a higher ceiling before demand destruction.
I've been tracking the blob gas market since the day of Dencun. Based on my experience analyzing fee markets during the 2017 ICO craze and the 2020 DeFi summer, the pattern is indistinguishable. First, cheap capacity encourages overbuilding. Then demand saturates supply. Then fees rise, catching late adopters off guard. The Ethereum community is now in stage two: capacity is tightening, but no one wants to talk about it because cheap blobs are the current bull case for L2s.
From my 2017 Bitcoin.com token sale work, I learned to look where others aren't looking. In 2017, it was the multisig wallet risk. Today, it's the blob utilization ratio. I pulled the daily blob count from a Dune dashboard (publicly available) and ran a simple linear regression. Even assuming a plateau in growth (which has never happened with new Ethereum features), we hit capacity within two years. The Dencun whitepaper itself acknowledges that the initial 6-blob limit is a starting point, but increasing it requires another hard fork, which takes at least 6-12 months of coordination. If we hit the ceiling in mid-2025, we might not get a fix until late 2026. That gap will be painful for rollups and their users.
Contrarian Angle
The prevailing narrative from VCs and L2 marketing teams is that blobs are effectively infinite because we can always add more. They point to the Ethereum roadmap's "full danksharding" phase, which will expand capacity to 16 or 32 blobs per block. But that upgrade (EIP-7623 and beyond) is still in the research stage. The core developers are focused on other priorities like Verkle trees and stateless clients. Increasing blob capacity is technically complex—it requires balancing L1 bandwidth and node hardware requirements. There is no guarantee it happens before the next bull run exhaustion.
Worse, the current cheap blob environment is actually creating a perverse incentive: rollups are encouraged to post data as often as possible, bloating their own operational costs in the future. Every L2 that receives VC funding to subsidize gas today is training its users to expect near-zero fees. When blob fees spike, those users will flee to cheaper L2s—creating a race to the bottom that benefits no one. The "liquidity fragmentation" problem that VCs love to sell solutions for is actually a symptom of this same structural imbalance: too many rollups competing for too little cheap data space. The fragmentation isn't a bug; it's a feature of an artificially under-priced resource. When blobs become expensive, the surviving rollups will be the ones that can batch efficiently or use alternative data availability solutions (like Celestia or EigenDA). But that introduces trust assumptions and defeats the purpose of settling to Ethereum.
Takeaway
Watch the blob utilization metric. It's a leading indicator for the next major gas crisis in the Ethereum ecosystem. When the daily average blob count surpasses 5.5 per block, start hedging: move liquidity to L2s with proven compression and batching strategies, or consider direct L1 usage for high-value transactions. The clock is ticking, and the market is not listening. In the chaos of the next fee spike, the survivors will be those who saw the blob ceiling before it hit.