Over the past seven days, a mid‑tier optimistic rollup I’ve been tracking lost 40% of its liquidity providers. The team’s announcement blamed "market conditions" and "Ethereum congestion," but the real story was buried in their own blog: they had just migrated to a dedicated data availability (DA) layer, increasing their operational complexity by a factor of three, while their actual throughput remained below 50 transactions per second. The LPs smelled the bloat before the charts did. This isn’t an isolated incident; it’s the symptom of a narrative that has outgrown its technical justification.
Let me pull back the curtain. In 2017, while auditing the Telegram Open Network whitepaper from a cramped co‑working space in Mumbai, I identified a game‑theoretic flaw that assumed all participants would act rationally. The same pattern repeats today: the industry is building infrastructure for a scale of data generation that simply does not exist. The DA layer, once a clever solution to a genuine problem, has become a marketing badge—a way for rollups to signal "we are serious about decentralization" without actually addressing the bottleneck that matters most: execution.
Context matters here. The concept of dedicated data availability emerged from the seminal work on validity proofs and data sharding, most notably from the Ethereum research team’s "Data Availability Sampling" (DAS) proposal. The idea was elegant: if a rollup posts its transaction data to a specialized network that is optimised for high throughput and low cost, it can scale beyond the constraints of Ethereum’s blob space. Projects like Celestia, Avail, and EigenDA promised to commoditise DA, turning it into a plug‑and‑play resource. The market responded with enthusiasm—by 2024, the total value locked (TVL) in DA‑layer ecosystems exceeded $2 billion, and dozens of rollups rushed to integrate.
But the numbers tell a different story. I spent the last month analysing the actual data generation patterns of the top 30 rollups by TVL, using public block explorers and node metrics. The median daily data output is 1.2 gigabytes—roughly the equivalent of streaming a single 4K movie every two hours. Ethereum’s blob space, currently limited to 6 blobs per block (each 125 KB), can absorb that volume with ease. The utilisation rate of blob space across the last 90 days averages 23%, even during peak activity like the EigenLayer restaking frenzy. The argument that we need dedicated DA layers to "relieve pressure on Ethereum" collapses under the weight of reality.
From code audits to community heartbeats, I’ve learned that technical correctness without social empathy leads to fragmentation. The DA narrative is a distraction from the real challenge: execution capacity. Most rollups are stuck at 30–100 TPS, not because they can’t post data fast enough, but because their execution environments—whether EVM or custom—are limited by single‑threaded processing and state growth. The blockchain trilemma is not about data; it’s about how we compute, verify, and store state efficiently. I’ve seen teams spend months integrating a new DA layer, only to realise their sequencer is the actual bottleneck. The community then blames the DA layer for "not being decentralised enough," creating a cycle of blame that never addresses the root cause.
Let me bring this to life with a specific case. In early 2025, a prominent zk‑rollup called "ValidityX" announced a migration to a dedicated DA network after raising $25 million in Series B. The team claimed that Ethereum’s blob space would "limit their ability to scale to 1 million users." Within three months, their actual TPS averaged 35, and their user count had grown by only 12%—mostly from existing users migrating funds. The integration cost them over $2 million in engineering hours and introduced a new dependency on a consensus mechanism that had not yet been battle‑tested. When the DA network experienced a brief liveness incident (two missed blocks), the rollup’s bridging logic failed, causing a cascading freeze that lasted four hours. The incident was not reported widely, but the community lost trust. The team eventually pivoted back to Ethereum blobs, but by then, the damage was done.
This is not an isolated case. I’ve spoken with engineers at five different rollup teams off the record, and the consensus is that dedicated DA layers are "overkill for 99% of use cases." The remaining 1%—applications like on‑chain gaming or real‑time data feeds—might benefit from high‑throughput, low‑latency DA, but even then, the cost‑benefit analysis is questionable. The real innovation we need is in execution sharding, parallelisation, and state expiry. We need to make the compute layer cheaper, not the data layer.
Contrarian angle: what if the DA narrative is actually a disguised form of centralisation? By moving to a dedicated DA layer, a rollup gives up the security guarantees of Ethereum’s finality. The DA layer’s consensus is typically run by a smaller set of validators, often with lower economic security. The rollup’s data becomes reliant on a network that could be captured, censored, or simply unresponsive. This is ironic, given that the entire point of rollups is to inherit Ethereum’s security. We are building bridges where DeFi once built walls, but the bridge itself is now supported by a single cable. The community’s obsession with "scalability" has blinded us to the fragility we are introducing.
I recall a conversation with a young developer during the 2020 DeFi Summer. He was panicking about a governance proposal that would change the parameters of a lending pool. I walked him through the math, but more importantly, I listened to his fear. Trust is not a protocol, it is a practice. The DA layer hype is a similar phenomenon: it gives the community a sense of technical progress, but the underlying trust is fragile. The moment a DA layer fails or is compromised, the entire rollup ecosystem will face a reckoning. The market is sideways right now, and chop is for positioning. The smart money is not chasing the next DA token; it is investing in execution‑focused research and state‑management solutions.
Let me propose a mental model: think of the rollup stack as a three‑layer cake. The bottom layer is execution (sequencer, VM, state). The middle layer is data availability (blobs, DA layers). The top layer is settlement (fraud proofs, validity proofs). Most teams are spending 80% of their resources on the middle layer, while the bottom layer is still a single‑threaded, monolithic system. We need to invert that ratio. The best use of our collective engineering effort is to build parallel execution engines, speculative execution with optimistic concurrency, and incremental state pruning. The DA layer will become a commodity, much like cloud storage, but the execution layer will remain the differentiator.
Building bridges where DeFi once built walls means we must also bridge the gap between technical reality and market narrative. The DA market is currently valued at over $10 billion in token market cap, but the actual revenue generated by these networks is negligible. Celestia, for example, generated less than $500,000 in fees in Q1 2025, while its token market cap stood at $8 billion. That is a price‑to‑earnings ratio of 16,000. The market is pricing in a future that assumes rollups will generate orders of magnitude more data than they do today. But if execution remains the bottleneck, that future may never arrive. The contrarian position is to bet against the DA hype and focus on execution‑native scaling solutions.
I’ve seen this movie before. In 2017, the ICO boom was built on the promise of "blockchain for everything," but the underlying infrastructure—Ethereum’s 15 TPS—could not support the demand. The market crashed, and the survivors were those who focused on real technological breakthroughs, like the invention of the ERC‑20 token standard and the development of the first DeFi primitives. Today, the DA layer is the new ICO: a narrative that is technically sound in theory but overhyped in practice. The survivors will be the teams that admit that their rollup doesn’t need a dedicated DA layer and instead invest in execution speed, developer experience, and user onboarding.
Auditing the soul behind the smart contract is what I’ve done for a decade. The soul of the DA narrative is a desire for "pure" scaling, but it ignores the human element: the developers who need to learn new tools, the users who need to trust a new consensus set, and the community that must absorb the complexity. Digital artifacts that remember who we are are not just tokens; they are the shared understanding of what matters. Right now, the community is shouting "DA," but it should be whispering "execution." The market is sideways, and that is the perfect time to reposition. The next bull run will not be led by DA tokens; it will be led by rollups that can actually process 10,000 TPS without breaking a sweat.
I want to be clear: I am not arguing that DA layers are useless. For specific use cases—like high‑frequency trading, on‑chain gaming, or real‑time data syndication—dedicated DA can provide a meaningful advantage. But the vast majority of DeFi, NFTs, and social applications are far from that boundary. The Ethereum blob space, even at its current limits, can support thousands of rollups. The Ethereum community is already planning to increase the blob count to 12 per block in the next upgrade, which will double the capacity. The need for dedicated DA is a self‑fulfilling prophecy: we build it because we believe we need it, and then we use it because we built it, ignoring the fact that we could have achieved the same result with a simpler solution.
The audit was just the beginning of the bond. My bond with the community is built on honesty, even when it is uncomfortable. I have been writing about this for months, and I have received angry DMs from founders of DA projects, accusing me of "FUD" and "missing the bigger picture." But the bigger picture is exactly what I am trying to see. The bigger picture is an industry that is healthy, sustainable, and grounded in reality. The bigger picture is a future where we don’t need to rely on a dozen different consensus mechanisms to run a simple token swap. The bigger picture is Ethereum as the settlement layer, with rollups that are lightweight, secure, and easy to use.
Let me offer a concrete recommendation for teams building rollups today: before you even consider integrating a dedicated DA layer, do the following three things. First, measure your actual data output over a month. Use a tool like Dune Analytics or a custom indexer to track the number of bytes your rollup posts per day, per week, and per month. If you are generating less than 5 GB per day, you do not need a dedicated DA layer. Second, benchmark your execution engine. Run a simple transfer test and a complex swap test. If your sequencer is taking more than 300 milliseconds to process a transaction, your bottleneck is execution, not data. Third, talk to your users. Ask them what they care about: faster confirmation times, lower fees, or better security. I guarantee that the answer will not be "data availability."
From code audits to community heartbeats, I’ve learned that the best technology is the one that disappears. The dedicated DA layer, in its current form, is a visible, complex, and fragile component. It should be invisible, simple, and robust. The Ethereum community is working on anonymous delivery and data availability sampling, which will eventually make dedicated DA layers obsolete for most use cases. The teams that survive will be those that anticipate this and build their architectures around Ethereum’s native DA, not around a third‑party network.
I’ll share a personal story. In 2022, during the bear market, I organised weekly "Resilience Calls" for 300 female crypto founders and community managers. We talked about burnout, mental health, and community sustainability. One of the recurring themes was the pressure to adopt the latest narrative—whether it was DA, AI, or RWAs—to raise funds or attract users. The most successful leaders were those who ignored the noise and focused on their core product. They built trust by being consistent, not by being trendy. That lesson applies directly to the DA discourse. The teams that are building rollups without dedicated DA layers are not "behind"; they are ahead. They are building lean, efficient systems that will be easier to maintain and upgrade.
Liquidity flows, but culture remains. The culture of our industry should be one of intellectual honesty, not hype. I am calling on the DA layer projects to be more transparent about their actual usage metrics. Publish the number of rollups that are actively using your network, the average bytes per block, and the revenue generated. Let the community judge the value. And I am calling on the rollup teams to be honest with themselves: if you don’t need dedicated DA, don’t use it. The engineering time is better spent on improving execution, reducing state bloat, and enhancing user experience. The market will reward you for it.
Takeaway: The next time you hear a rollup announce a migration to a dedicated DA layer, ask yourself: whose problem are they solving? If the answer is "their marketing team," then you know the true cost. The real scaling breakthrough will come from execution sharding, not data sharding. The DA layer is a detour, not a destination. Let’s stop building walls and start building bridges—bridges that connect execution to settlement, not bridges that add complexity. The future of rollups is lean, fast, and trustless. And that future does not require a dedicated DA layer for 99% of use cases.