We don’t need more users; we need more stewards. That phrase has guided my work since 2022, when I retreated to a cabin in Yilan and watched the Terra collapse take my idealism with it. Back then, the narrative was simple: rollups would save Ethereum from itself. But after three years of auditing governance models, mentoring DAO builders, and watching the post-Dencun data landscape shift, I’ve come to a hard conclusion. The cheap gas we’re enjoying today is a temporary illusion. Within two years, blob data will be saturated, and every rollup’s gas fee will double—or worse.
Let me start with the hook. On March 13, 2024, Ethereum’s Dencun upgrade went live, introducing blob-carrying transactions (EIP-4844) and slashing L2 gas fees by over 90% overnight. Arbitrum, Optimism, Base—all of them dropped from $0.50 per transaction to under $0.01. It was a miracle. But what most articles missed is that the blob space is fixed. Each block can hold up to 6 blobs (target 3), and the total capacity is a hard limit. The demand for blobs, however, is growing exponentially. In just the first six months after Dencun, blob usage rose from near zero to over 60% of the target capacity during peak times. At current growth rates—driven by DeFi activity, AI data provenance chains, and the rise of consumer-facing dApps—we’ll hit blob saturation by late 2025. After that, rollups will have to compete for scarce blob space, and fees will climb back to pre-Dencun levels, or higher.
This is not a prediction pulled from a spreadsheet. In 2017, I watched a similar pattern unfold with Ethereum’s block gas limit. When CryptoKitties clogged the network, everyone said “just build more L2s.” Now we’re building L2s that themselves depend on a shared resource. During my 2024 work with The Alignment Circle, I mentored three teams launching DAOs on different rollups. Every single one assumed that low fees were a permanent feature. They designed their tokenomics to rely on frequent, cheap on-chain votes. When I asked them what happens when blob prices rise, they had no answer. That’s the blind spot: the entire Layer2 ecosystem is built on a cost assumption that has never been stress-tested.
Let’s drill into the numbers. Each blob can carry about 128 KB of data. With a target of 3 blobs per block (12-second slots), Ethereum’s blob bandwidth is roughly 2.5 MB per minute. Today, major rollups publish data to L1 every few minutes, consuming roughly 1 blob per batch. But as more rollups launch—and as existing ones increase throughput to compete with Solana—the demand will spike. Consider that Base alone now processes over 1 million transactions per day. Each transaction requires a data commitment. If every rollup optimizes for maximum throughput, they’ll need to publish more blobs per minute. The math is simple: at 3 blobs per block, the network can support about 15–20 rollups publishing at full capacity. There are already over 40 active rollups, and the number is growing. Saturation isn’t a question of if, but when.
Based on my audit experience with Harmony Bridge in 2025, I saw firsthand how even a well-intentioned protocol can stress a shared resource. Harmony Bridge used a custom data availability solution, but the majority of rollups rely on Ethereum’s blob layer. The moment blob fees rise, the cost of posting data becomes a significant portion of each rollup’s operational budget. For a rollup that charges $0.01 per transaction, if the blob cost per transaction rises to $0.02, the economics flip. Users will either pay more or leave. The irony is that rollups exist to make Ethereum scalable, but they themselves become bottlenecks when the underlying resource is finite.
Now, the contrarian angle. Some argue that we don’t need blobs at all. They point to alternative data availability layers like Celestia, EigenDA, or Avail as solutions that can absorb overflow. I’ve spent time with teams building on these alternatives, and I respect the innovation. But the market has spoken: Ethereum’s blob layer is the most secure and most trusted. As of 2026, over 80% of rollup data is still posted to Ethereum blobs. The alternative layers introduce additional trust assumptions and fragmentation. More importantly, they don’t solve the underlying problem—they shift it. If everyone moves to Celestia, Celestia’s data capacity will become the new bottleneck. The industry is addicted to a single, high-security resource, and that addiction will have consequences.
Trust is the only protocol that cannot be coded. The real risk is not technical—it’s behavioral. We’ve built a culture of “build first, worry later.” In 2022, I burned out because I believed in projects that promised scalability without trade-offs. The post-Dencun euphoria feels eerily similar. Every week, I see a new rollup launch with a marketing campaign that says “unlimited L2 scaling.” They conveniently forget to mention that their data availability is dependent on a shared, finite resource. This is the same pattern I saw in 2017 with ICOs that promised decentralization while centralizing token distribution. The solution is not a technological fix—it’s a governance fix. We need rollups to commit to data efficiency, to batch more aggressively, and to be transparent about their blob consumption. We need stewards, not just users.
We built not for the peak, but for the valley. In a bear market, survival matters more than gains. Look at the protocols that are bleeding liquidity right now. They are the ones that assumed cheap fees would last forever. They designed their tokenomics around high transaction volumes, and now they’re discovering that every extra cent of gas eats into their margins. The teams that will survive are the ones that have planned for blob saturation. They’re the ones that compress calldata, use state diffs, and accept slightly lower throughput in exchange for cost predictability. During my 2024 mentorship program, I advised one DAO to implement a “blob budget” that caps monthly data posting costs. It was a boring change, but it saved them from a 3x fee spike in March 2026.
What happens when blob space is fully saturated? The fee market will kick in. Rollups will bid for blobs, and the price will rise until the least profitable rollup stops posting. That means smaller rollups or those with lower transaction fees will be priced out. The network effects will favor the largest rollups—Arbitrum, Optimism, Base—while smaller chains become ghost towns. This is not decentralization; it’s consolidation disguised as scaling. The very problem Ethereum was designed to solve—centralization of power—will be replicated at the L2 level.
I’m not saying this to spread fear. I’m saying it because I’ve been in the valley, and I’ve seen what happens when idealism ignores reality. The next two years will separate the builders who are serious about long-term sustainability from those who are chasing short-term hype. For the reader who holds assets in a rollup, ask yourself: has the team published a blob cost analysis? Do they have a contingency plan for when fees double? If not, your “safe” L2 might be more fragile than you think.
The vision forward: we need to treat blob space as critical infrastructure, not a free public good. Regulation may eventually force rollups to disclose their data availability costs, but the industry should act first. I’m working with a small group of developers to create a “blob efficiency score” that ranks rollups by their data usage per transaction. It’s a governance tool, not a technical one. It’s the kind of work that matters in a bear market—slow, deliberate, and built on trust. We don’t need more users; we need more stewards. The question is: will you be one?