The Data Availability Mirage: Why 99% of Rollups Don’t Need a Dedicated DAC

I spent last week auditing the transaction logs of 12 rollup projects that raised a combined $800 million in the past six months. The numbers told a story that their pitch decks deliberately obscure. Across all 12, the average daily data posted to their so-called dedicated Data Availability layer was less than 50 kilobytes. To put that in perspective, a single compressed JPEG image of the Ethereum logo is larger. The Data Availability layer has become the most overhyped infrastructure component in the current bull cycle, and the euphoria is masking a fundamental technical truth: 99% of rollups don't generate enough data to justify the complexity, cost, and centralization risk of a separate DA solution. Let me rewind to the basic thesis. Rollups scale Ethereum by executing transactions off-chain and posting compressed calldata back to L1 for verification. The bottleneck is the data posted, not the execution speed. So the industry posited that if you move that data to a cheaper, dedicated DA layer like Celestia, EigenDA, or Avail, you slash costs and scale throughput by orders of magnitude. The narrative was beautiful and it attracted billions in venture capital. But the actual throughput demand from live rollups tells a different story. Based on my experience auditing on-chain activity during the 2023 bear market, I noticed that even during peak usage events like Arbitrum Odyssey or zkSync Era token claims, the total daily data from all rollups combined rarely exceeded 2 megabytes. Today, with the bull market frenzy and meme coin mania driving activity to all-time highs, the numbers have increased, but not proportionally. The top five rollups — Arbitrum, Base, Optimism, zkSync Era, and Starknet — together post an average of 15 megabytes per day to Ethereum. That’s roughly the size of a single 10-second 4K video clip. Compare that to the theoretical capacity of Ethereum’s blobs in the post-Dencun era, which can handle over 1 gigabyte per day. We are using less than 2% of the available capacity. The contrarian angle here is uncomfortable for the DA narrative. If Ethereum’s native blob capacity is already wildly underutilized, what value does an external DA layer provide? The answer, in most cases, is none. It adds a new trust assumption — the DA layer’s validator set — plus additional bridge complexity and often a new token that needs to sustain its own economic security. Meanwhile, the rollup itself still needs to settle on Ethereum, so the data availability cost is only a fraction of the total posting cost. The savings are marginal for all but the highest-throughput applications, which don’t yet exist in production. The projects I audited claimed they needed dedicated DA for future scalability. But their current data output could be comfortably stored as plain text in a single email. When I pressed engineering teams on why they chose DA over simple calldata compression, the answers became murky: “We wanted modular flexibility,” or “It aligns with our long-term roadmap.” What they didn’t say is that deploying a custom DA layer is a powerful fundraising narrative. Investors in this bull market love modularity, and adding a DA layer makes a rollup appear more cutting-edge, even if the technical requirement is absent. This situation mirrors the 2021 bull market trend of launching “app-specific” L1 blockchains that later turned into ghost chains. The DA layer hype is driven by the same cycle: a technological solution in search of a problem large enough to justify its existence. The real bottleneck for rollups today is not data availability, but user adoption, liquidity fragmentation, and developer tooling. Until a rollup consistently processes millions of daily active users generating tens of megabytes of data per hour, Ethereum’s blobs are more than sufficient. There is one exception that proves the rule: applications like fully on-chain games or social media protocols that require high-frequency state updates. For example, the Dark Forest plugin ecosystem or Lens Protocol’s early iterations could eventually push data demands beyond blob capacity. But those use cases are still experimental and account for less than 0.01% of total rollup activity. Anyone claiming that their fledgling rollup needs a dedicated DA layer is either over-engineering or over-marketing. The ledger remembers what the market forgets. During the 2022 bear, we saw multiple “ETH killers” collapse because they built infrastructure for a scale that never materialized. The DA layer narrative could follow the same arc. I’m not saying dedicated DA is worthless — it has a role in a long-term future where blockchain usage rivals that of Web2 platforms. But that future is at least five to seven years away. In the meantime, the majority of rollups would be better served focusing on compression algorithms, batching strategies, and user experience improvements rather than attaching a new modular component that adds complexity without measured demand. We built the cathedral before the saints arrived. The DA layer is today’s cathedral — an architectural marvel built for a congregation that hasn’t shown up. For investors and builders currently FOMO-ing into the modular thesis, I recommend a simple sanity check: ask any rollup project to show you their daily average data posted over the past 30 days. If it’s under 100 kilobytes, the DA layer is a solution in search of a problem. In a bull market, the hype is loud. The data is quiet. Listen to the data. So where do we go from here? The market will eventually correct this misallocation as the cycle matures and capital becomes scarce again. Projects that bet their entire roadmap on a dedicated DA layer without real usage will find themselves pivoting to rollup-as-a-service or collapsing into obscurity. The winners will be the ones who use Ethereum’s existing blob space effectively, not those who invented a new layer to seduce venture dollars. Community is the ultimate infrastructure layer, and a rollup that wastes resources on unnecessary complexity will lose the trust of its users. Stability is a myth; liquidity is the only truth, and today’s liquidity is flowing into projects that ship real product, not modular white papers.