It was a quiet Tuesday morning when I saw a tweet from a prominent L2 dev celebrating the Dencun upgrade: “Blobs are infinite! Gas fees on Arbitrum are now under $0.01 forever.” I nearly choked on my coffee. Not because he was wrong about the current low fees — he wasn’t. But because the word “forever” in crypto is always followed by a rude awakening. Within hours, I pulled up the blob usage data from Etherscan and did a quick projection. The numbers didn’t lie. At the current growth rate of blob-posting transactions, we will hit 100% of the target blob capacity within 18 to 24 months. After that? Every rollup will have to compete in a fee auction for limited blob space. Fees will not just rise — they could double, triple, or worse.
That tweet, and the euphoria it represented, is exactly the kind of bull-market amnesia I’ve learned to distrust. Back in 2017, during my BlockNaija days, I watched people claim that “blockchain scales infinitely” while ignoring the 15-minute confirmation times during the CryptoKitties jam. In 2021, I saw DeFi projects boast about “unlimited TVL” until the Terra collapse proved otherwise. The pattern is always the same: a technological breakthrough triggers a wave of optimism, and the community forgets that every constraint is merely postponed, not eliminated. Dencun’s blob mechanism is a masterpiece of engineering — it reduces L2 costs by 90–95% today. But it is not a magic wand. It is a new resource with a hard ceiling, and we are racing toward that ceiling faster than most people realize.
Let me walk you through the math. Before Dencun, rollups posted their transaction data to Ethereum’s calldata. Each block had a limited gas per block, and calldata was expensive because it is permanent state. The maximum throughput for all L2s combined was roughly 200–300 transactions per second (TPS) when using calldata, but the cost per transaction was often $0.20 or more. Dencun introduced blobs — temporary data storage that lasts about 18 days before being pruned. Blobs are cheap because they are ephemeral and do not require the EVM to execute them. Each blob carries about 125 kilobytes of data, and the target is three blobs per slot (a slot is every 12 seconds). That gives a theoretical maximum L2 throughput in the order of several thousand TPS, but only if all three blob slots are filled. Today, on average, we are using about 1.2 blobs per slot. The growth rate of blob usage has been accelerating as more rollups adopt the mechanism. In March 2024, after the upgrade, blob usage was negligible. By June, it was 0.5 blobs per slot. By September, 0.9. By December 2024, it hit 1.4. The trend is exponential, driven by both existing rollups posting more data and new rollups entering the ecosystem.
Based on my own modeling (I built a simple script using data from Dune Analytics), if the current quarterly growth rate of 40% in blob count continues, we will reach the target of 3 blobs per slot by early 2026. Once we hit the target, the fee mechanism changes. Currently, blobs have a base fee that is adjusted dynamically, similar to EIP-1559, but with a target. When usage exceeds the target, the base fee starts rising exponentially. Right now, we are below target, so fees are near zero. Once we cross the target, every rollup will have to pay a premium to get their blobs included. The consequence is that L2 gas fees will no longer be $0.01 — they will settle at a market-clearing price determined by the highest-value transactions. That price could easily be $0.10 or $0.20 per transaction, effectively erasing the cost advantage of blobs over calldata.
But wait — there is a nuance that most analysts miss. The Ethereum roadmap includes future upgrades like PeerDAS (Peer Data Availability Sampling) which will increase the blob target from 3 to 6 or even 8 per slot. That would push the saturation point further into the future, perhaps by another two years. However, the rate of rollup adoption is also accelerating. We have seen the rise of “blob heavy” applications like fully on-chain games, AI inference verification, and zero-knowledge proof aggregation — all of which produce large amounts of data. In 2023, the total data posted to Ethereum by all L2s was about 1 GB per day. By late 2024, it was 5 GB per day. By 2026, if the trend continues, it could be 50 GB per day. Even with PeerDAS, the blob capacity might only support that for another year before hitting saturation again. The fundamental issue is that blob space is a shared resource, and there is no inherent mechanism to prioritize high-value vs low-value data. Every rollup is free to spam the blob space as long as they pay the fee. The tragedy of the commons is baked into the design.
Here is the contrarian angle: the real bottleneck is not technical bandwidth, it is economic coordination. The Ethereum community assumes that market forces will naturally allocate blob space efficiently. But in practice, the largest rollups — Arbitrum, Optimism, Base — have strong incentives to maintain low fees to attract users. They might engage in a fee war that drives up base fees for everyone. Or they might resort to off-chain aggregation solutions, effectively recreating the centralization they were supposed to avoid. I have seen this before: in the early days of DeFi, everyone assumed that liquidity would flow to the most efficient market, but instead, we got yield farming wars that distorted risk premiums. The same pattern will replay with blobs. The most probable outcome is that the smallest rollups get priced out, and the ecosystem consolidates around a few dominant players who can afford the blob fees or build private infrastructure.
I remember a conversation I had in December 2024 with a founder of a mid-size L2. He told me, “We’re preparing for the blob crunch by pre-negotiating deals with Ethereum validators to get preferred blob inclusion.” That is essentially a bribe — a form of out-of-protocol priority. It is not illegal, but it violates the principle of permissionless access. If we allow that, we are no better than the traditional financial system where large players pay for network access. The beauty of Ethereum is meant to be that every transaction is equal in the eyes of the protocol, but blobs introduce a new hierarchy: those who can pay for data vs those who cannot. We need to start discussing governance mechanisms to prevent blob monopolization before it becomes entrenched.
Trust the process, but verify the code. The Dencun upgrade is a beautiful piece of engineering, but it solved a short-term scaling problem by shifting the bottleneck to a new dimension. We now have a new resource — blob space — that will become scarce within two years. The community must proactively design fee markets, perhaps with different tiers or priority lanes, to ensure that small rollups and innovative applications are not squeezed out. Otherwise, the promise of “L2 scaling for everyone” will become a privilege reserved for the well-funded few.
So here is my takeaway: if you are building a rollup or an app that relies on cheap blob posting, you have a window of opportunity — maybe 12 to 18 months — to capture users and establish network effects before fees rise. Don’t assume the low fees are permanent. Plan ahead, design your compression strategies, and consider bundling transactions to reduce your blob footprint. And most importantly, participate in the governance discussions about blob fee markets. The future of decentralized scaling depends on getting this right. Otherwise, we will repeat the same cycle of centralization that every scaling solution has faced since the dawn of blockchain.
As I wrote in my 2022 article after the Terra collapse, “Bear markets reveal the cracks; bull markets paper them over.” The current bull market is painting a beautiful picture of cheap L2 bliss. But the cracks are already forming. The question is whether we will fix them before they split the foundation.
