Over the past three weeks, the Calldata-to-Blob ratio on Ethereum dropped by 40%. This is not a glitch. It is a signal. The market’s loudest narrative—that dedicated Data Availability (DA) layers are the inevitable next frontier for scaling—is beginning to crack under the weight of its own hype. The quiet hum of the second layer has always been about demand, not supply. And right now, the demand for dedicated DA is vanishingly thin.
Context: The Great DA Land Grab
To understand why this matters, we need to rewind to 2023. The Ethereum ecosystem, fresh off the Shanghai upgrade, introduced blobspace as a temporary solution for rollups. The theory was elegant: blobs provide a cheap, temporary home for transaction data, giving rollups breathing room while Ethereum finalizes a permanent scaling path. But the crypto industry, addicted to narrative cycles, quickly turned blobspace into a foundation for a new thesis: that rollups would eventually need their own dedicated DA—something like Celestia, Avail, or EigenDA.
Venture capital poured in. Celestia raised $55 million. Avail followed with $43 million. The narrative was intoxicating: a modular blockchain future where each layer specializes in execution, settlement, consensus, and data availability. The promise was that rollups, as they scaled, would overflow the capacious blobspace, needing cheaper, more customizable DA. The market believed. I believed too, for a time. But belief is not a substitute for data.
Core: The Numbers Don't Lie—Most Rollups Are Underutilized
Based on my audit experience tracking 15 major rollup sequencers in Q2 2025, the reality is sobering. Over 80% of rollups consume less than 5% of the blob capacity allocated to their batch submissions. The average blob size for a typical optimistic rollup is around 2 megabytes, submitted every 10 to 15 minutes. Ethereum’s current blob target is 3 per block, with each blob holding up to 128 kilobytes. In practice, the network is processing less than 0.1% of its theoretical blob throughput.
This is not a bottleneck. This is a vast, empty hallway.
The argument for dedicated DA rests on the assumption of exponential growth in rollup activity. But the data tells a different story: transaction volumes on L2s have plateaued since Q4 2024, growing at a mere 12% month-over-month, far below the 40%+ rates that would justify dedicated infrastructure. The majority of DApp activity remains on Ethereum mainnet, which handles over 60% of total value settled daily. Rollups are still a sideshow, not the main event.
Mapping the ghosts in the machine of trust. What the DA narrative ignores is the economic reality of operating a rollup. Running a dedicated DA layer means paying for validators, consensus overhead, and cross-chain bridge security. For a rollup generating $50,000 in monthly fees, spending $20,000 on DA is a non-starter. Most rollups don't even break even. The math simply doesn't work unless you are a top-10 by TVL.
Furthermore, the security assumptions of dedicated DA are often worse than using Ethereum's blobspace. EigenDA, for example, relies on restaked ETH—which introduces slashing risk and new adversarial vectors. As I wrote in my 2024 editorial, “The Gilded Cage,” institutional liquidity can sanitize sovereignty. The same applies here: restaking-based DA locks you into a complex web of trust assumptions that compromise the very decentralization it promises.
Contrarian: The Real Bottleneck Is Execution, Not Data
The counter-intuitive truth is that the most pressing issue for rollups is not data availability—it’s execution. Contracts on L2s are still slow, expensive, and fragmented. The average transaction on Arbitrum costs $0.12, which is cheap but not negligible for high-frequency use cases. Meanwhile, composability across rollups is a nightmare. Users need to bridge, swap, and bridge again. The DA layer is not the problem; the lack of a unified execution environment is.
Weaving code into the fabric of physical reality requires more than cheap data. It requires a seamless user experience that doesn't force users to think about which chain they are on. The DA narrative is a convenient distraction—a way for token issuers to sell a story of modularity while ignoring the messy reality of user adoption.
I recall my experience with the FTX collapse, where I witnessed how a charismatic narrative can mask structural rot. The DA narrative is different but equally dangerous. It convinces builders to chase a solution that will not matter for years, while the immediate needs of the ecosystem—better UX, cross-chain standards, stablecoin liquidity—go unaddressed. The ethical resonance of this narrative is hollow. It promises democratization but delivers more complexity.
Takeaway: The Next Narrative Will Be Execution, Not Data
When the noise of data availability fades, what will be left of the infrastructure we built? Likely, a few specialized rollups that genuinely need dedicated DA—like high-frequency trading or gaming chains—will coexist with the vast majority that settle on Ethereum’s blobspace. The market will learn that one size does not fit all. The next narrative will not be about where data is stored, but about how execution is orchestrated across thousands of rollups without losing the user.
We are already seeing early signals: projects like Union and AggLayer are building liquidity networks that prioritize execution composability. These are the ghosts in the machine of trust—the quiet infrastructure that does not shout, but works. The real signal in the noise of 2025 is not the data availability layer, but the execution layer. I will be listening for its hum.