Every week, a new blockchain announces a dedicated data availability (DA) layer—a separate network promised to store the firehose of transactions that rollups will one day produce. In 2025 alone, we have seen Celestia, Avail, EigenDA, and NEAR DA battle for dominance, each claiming to herald the “modular future.” But as I sit in my Mumbai apartment, staring at the node metrics from the past six months, a quieter story emerges: the DA layer is overhyped. 99% of rollups do not generate enough data to justify a dedicated network. The fear of a data shortage is a phantom, and our collective obsession with it is pulling energy away from the real challenge—building trust.
This realization did not come from a Bloomberg terminal or a Messari report. It came from a cold, hard look at the on-chain numbers that I live with every day as the founder of a Web3 community. During the 2020 DeFi Summer, I founded the Mumbai Chain Guardians, a volunteer network of 200 moderators who monitored Aave and Compound for vulnerabilities. We translated 50 technical upgrade proposals into Hindi and English WhatsApp guides. That experience taught me that what users fear is not insufficient block space—it is losing their savings to a bug they cannot understand. The DA layer frenzy is a solution in search of a problem, and it is time we admit it.
The Hook: A Numbers Wake-Up Call
Over the last 90 days, the cumulative data published by all Ethereum rollups averaged roughly 0.3 megabytes per second (MBps) to the L1. That is less than the bandwidth required to stream a single 480p video on YouTube. The most active rollup—Arbitrum—peaked at 0.8 MBps during the GMX trading frenzy in March. Yet, dedicated DA layers are engineered to handle hundreds of MBps, if not gigabytes, per second. The mismatch is so stark that it borders on absurd. I remember attending a conference in April 2025 where a founder pitched his DA network as “the TCP/IP of decentralized storage.” I asked him how much data his current clients actually submit. He laughed nervously and changed the subject.
From code audits to community heartbeats, I have learned to measure a protocol’s health not by its theoretical capacity but by its real usage. The DA layer narrative is built on a premise that has not materialized: that dApps will generate order-of-magnitude more data than they do today. Even the most optimistic forecasts for on-chain gaming—the sector often cited as the killer use case—predict maybe 10 MBps by 2028. That is still within reach of Ethereum’s own EIP-4844 blobs, which already provide 0.75 MBps per slot and are upgradeable.
Context: The Modular Stack and Its Promises
To understand why this matters, we must revisit the modular thesis. The Ethereum ecosystem has embraced rollups as its scaling future. To keep fees low, rollups need to post transaction data somewhere cheap and abundant. The original plan was to use Ethereum’s calldata, but that is expensive because every node must download it. Enter DA layers: specialized chains built for one job—storing data cheaply and making it available for verification. Projects like Celestia launched with the promise of “scalable blockspace” at a fraction of L1 cost. Investors poured billions into the narrative. EigenLayer’s EigenDA, which leverages validator restaking, raised $100 million in 2024.
The logic is seductive: if you separate execution, settlement, consensus, and data availability, each layer becomes simpler and more efficient. But in practice, the separation has created a fragmented security model. Building bridges where DeFi once built walls, I have seen what happens when modularity meets human behavior. Users and developers do not think in terms of layers; they think in terms of safety and continuity. A rollup that posts data to a DA layer introduces a new trust assumption: that the DA layer’s light nodes will honestly attest to the availability of the data. This is a complex trust leap few users understand.
Core: The Data Mismatch Myth
I want to walk through the math that convinced me. Take a high-throughput rollup operating at 100 transactions per second (TPS), with each transaction roughly 200 bytes. That is 20 KBps, or 0.02 MBps. Even a medium-sized rollup like Arbitrum One, which averages around 15 TPS, produces about 3 KBps. To saturate Celestia’s current mainnet capacity of 2 MBps, you would need roughly 100 such rollups running at full tilt. Today, there are maybe three rollups with meaningful activity. The rest are testnets or ghost chains.
Critics will argue that future use cases like on-chain order books, high-frequency trading, or fully on-chain games will require higher throughput. I agree that these applications will grow, but not at the pace needed to justify the current level of capital and attention directed at DA layers. Based on my audit experience in 2017, when I spent four months dissecting the TON whitepaper, I learned that hype curves often detach from technical reality. The TON team promised a million TPS. They did not deliver. The DA layer hype today echoes that same pattern: a focus on theoretical maximums rather than actual demand.
Moreover, Ethereum’s blob space is already expanding. EIP-4844 introduced a separate fee market for blobs, and future upgrades like danksharding (EIP-7594) will increase blob count from 16 per block to 64 or more. That provides a clear upgrade path without adding a new trust layer. Rollups deploying on Ethereum’s blobs retain the security of the L1. The moment you move to a dedicated DA layer, you introduce a new validator set with its own incentive risks. I have seen what happens when incentive assumptions fail—the Terra collapse in 2022 was a game-theory failure as much as a financial one. Trust is not a protocol, it is a practice.
Contrarian: The Hidden Cost of Modular Security
Here is the contrarian view most DA-layer advocates will not state: their networks are not only overpriced, they are potentially less secure than staying on L1. Consider the data availability problem: a DA layer relies on light nodes to randomly sample data blobs and confirm they are available. If a malicious sequencer withholds a blob, light nodes have a high probability of catching it, but not certainty. The probability increases with the number of samples, but that costs bandwidth. Most DA layers aim for 99.99% confidence—which means a 1-in-10,000 chance of a successful attack per block. Over a year of 10,000 blocks, that is nearly a certain exploit. On Ethereum L1, you have full consensus finality.
I brought this up at a panel in 2025, and the response was: “We will add more light nodes.” But more light nodes mean more work, which introduces centralization pressure. The more efficient you make the DA layer, the more you rely on a small number of full nodes to reconstruct history. It is a trade-off we rarely discuss because it is uncomfortable. Auditing the soul behind the smart contract means examining these hidden assumptions.
Furthermore, the coordination costs of running a rollup on a dedicated DA layer are non-trivial. The rollup operator must manage two asymmetric keys, two sets of endpoints, and two separate governance processes. I saw a team of five engineers in Bengaluru spend two months debugging an integration with Celestia because of a version mismatch in the light node client. Those two months could have been spent on user experience or community growth. The complexity tax is real.
Takeaway: Redefining Value Beyond the DA Hype
So where do we go from here? The DA layer narrative is not entirely wrong—it merely arrived too early and with too much hype. The real bottleneck in Web3 today is not data availability; it is trust availability. Users do not understand rollups, they do not understand DA layers, and they certainly do not understand the difference between a data availability committee and a fraud proof. What they need is a seamless experience where they can trust the system without reading a white paper. That is the bridge we must build.
I have seen the power of trust in action. In 2021, I partnered with the Tata Trusts on “Heritage on Chain,” an NFT initiative preserving Indian textile patterns as ERC-721 tokens. We raised $150,000 in ETH, with 70% going directly to artisan communities. The project succeeded not because we had the fastest DA layer, but because we built a narrative of cultural dignity and transparency. The artisans trusted us because we showed we cared. Liquidity flows, but culture remains.
For the Web3 ecosystem to move forward, we must stop fetishizing modularity and start focusing on human-centric design. The next bear market will wash away the projects that built for speculators and engineers. It will preserve those that built for people. From the 2022 bear market counseling circles I ran for 300 female founders, I learned that resilience comes from community, not from infrastructure. We need to redirect the billions of dollars spent on DA layers toward user education, security audits, and mental health support for builders. That is where the real value lies.
Digital artifacts that remember who we are—that is the promise of blockchain. Not a data layer that can store a petabyte of zeroes, but a system that stores our trust. Let us stop building data highways for cars that have not been invented. Let us build safe streets for the communities that are already here.