We didn't build L2s to become rentiers.
That thought hit me while staring at the latest Dencun upgrade metrics last week. Blob usage on Ethereum has already hit 35% of theoretical capacity just four months after the upgrade. If the current growth curve holds โ and it will, because every new rollup wants a piece of that cheap calldata โ we are looking at full saturation by late 2026. And when blobs are full, the base fee spikes. Rollup gas fees will double overnight. The narrative that "L2s make Ethereum cheap forever" is a lie. Let me show you the numbers.
Context: What Dencun Actually Did
Dencun introduced proto-danksharding, a new data storage mechanism called "blobs." Rollups now post their transaction data to blobs instead of permanent calldata. This reduced L2 gas fees by 90%+ in the first week. Magic, right? But blobs have a fixed per-block target: 3 blobs per block, with a maximum of 6 before the base fee rises exponentially. That's roughly 43,200 blobs per day theoretical max, but practical daily throughput is much lower due to block time variance. Currently, the network processes around 15,000 blobs per day, driven primarily by Arbitrum, Optimism, and Base. At a 10% monthly growth rate (conservative for a bull market ramp-up), we hit target saturation in about 18 months.
Based on my audit experience analyzing rollup economic models since 2020, I've seen this pattern before. Every scaling solution starts with an efficiency gain, then gets flooded by demand until the friction returns. It's the same game as liquidity mining: projects subsidize TVL with high APY, users pile in, and when the incentives stop, they vanish. Blob capacity is the new subsidy. Let me explain why that matters.
Core: The Unsustainable Subsidy of Cheap Blobs
Rollups are competing for blob space because it's the cheapest way to finalize transactions. But the blob market is a congested commons. Unlike Ethereum's calldata, which is priced by gas market dynamics, blobs have a soft cap. When demand exceeds the target, the base fee adjusts. But here's the kicker: rollups don't pay the full economic cost of decentralization. The cost of running blob verification nodes is subsidized by Ethereum's security budget. If every rollup used blob space at scale, the total cost to the network would skyrocket, forcing higher fees or fragmenting consensus.
We didn't build L2s to be dependent on cheap blobs; we built them to be independent execution environments. Yet today, rollup roadmaps are built on the assumption of infinite cheap blob space. That's a cognitive blind spot. Let me quantify: if blob demand grows at 15% per month (the average growth rate of active rollups since March 2024), we hit 3 blobs per block by mid-2026. At that point, the blob base fee starts climbing. By early 2027, it will be 5-10x today's levels, effectively doubling rollup transaction costs for end users. The so-called "L2 scaling revolution" becomes a slow, expensive relic.
We didn't account for the composability tax, either. When blob space is scarce, rollups will start bidding against each other for priority. The largest L2s (Arbitrum, OP Mainnet) will outbid smaller chains like zkSync or Scroll, driving up costs across the board. This creates a winner-take-most dynamic that contradicts the decentralized multi-rollup future we were promised.
Contrarian: Maybe Saturation Is a Feature, Not a Bug
Here's the uncomfortable truth: blob saturation might be exactly what Ethereum needs. It forces rollups to innovate โ using data compression, zk-proofs for data availability, or moving to alternative DA layers like Celestia or EigenDA. If blob space remains cheap forever, rollups have no incentive to optimize. The market will naturally push them toward efficiency, just as Ethereum's high gas fees pushed users to L2s in the first place.
But this argument ignores human behavior. Blockchains are not efficient markets; they are tribal networks. Rollups will choose to stay on Ethereum because of liquidity and brand recognition, even if costs rise. They won't switch to Celestia until it hurts enough. And by then, users will already be conditioned to expect cheap transactions. The emotional toll of gas fees doubling again will alienate the retail base that just returned after the bear market.
We didn't build a system where resilience depends on pain. We built it to withstand pressure. That's where my 2022 bear market support network experience kicks in: communities need realistic expectations, not utopian promises.
Takeaway: Prepare for the Blob Cliff
Here's my forward-looking judgment: by Q4 2026, at least three major L2s will announce fee increases due to blob saturation. The narrative will shift from "L2s are cheap" to "L2s are still better than L1." But that's a weak defense. The real solution is not more blobs (EIP-7742, anyone?) but a fundamental rethinking of rollup economics. Developers should already be experimenting with compressed transactions and zero-knowledge DA. If your rollup hasn't started optimizing for blob scarcity, it will become uncompetitive in two years.
We didn't come this far to become victims of our own success. The blob clock is ticking. Will you build for abundance, or for the inevitable shortage?