The chart didn’t scream; it whispered. Over the past 30 days, average blob inclusion rates on Ethereum have crept from 65% to 89% — but nobody in the Telegram groups is talking about the math. They’re still celebrating how cheap L2 transactions feel. I’ve been staring at the mempool data since the Dencun upgrade went live in March 2024, and the pattern is unmistakable: we’re burning through spare capacity faster than a liquidity miner chasing a farm with a 10,000% APR.
Every rollup—Arbitrum, Optimism, Base, zkSync—is shoving calldata into blobs like there’s an infinite buffet. But blobs aren’t elastic. They’re fixed at six per block (with a target of three), and each blob carries about 128 KB. Ethereum’s block space is a finite resource, and Dencun just gave everyone a temporary discount. The catch? The discount expires when demand saturates the supply. And based on on-chain metrics, that saturation is only 12 to 18 months away at current growth rates.
Let me rewind for context. Before Dencun, rollup gas fees were directly tied to Ethereum L1 basefee—often $0.50 to $5 per transaction. Post-Dencun, the blob gas market operates on a separate, multi-dimensional fee mechanism. Blob basefee adjusts based on the number of blobs in each block relative to the target. When demand is below target, fees are near zero (we’ve seen sub-penny blob fees). But when demand exceeds target, the basefee rises exponentially. Right now, we’re still in the “cheap zone” because total blob usage is hovering around 4-5 per block, under the target of 6. But as more rollups launch and existing ones scale, that headroom will vanish.
Tracing the trail from NFT peaks to DeFi valleys, I’ve learned one hard truth: every temporary efficiency in crypto becomes a permanent congestion point once the user base grows. The same happened with Ethereum base layer in 2020-2021. Dencun is not a solution; it’s a time-buying patch. And the clock is ticking.
The Saturation Scenario
Here’s the crunch I’ve been modeling. There are currently eight major rollups using blobs, plus dozens of smaller ones. Each rollup aims to scale to millions of daily transactions. If we assume average blob usage per rollup is 0.8 blobs per block (some use more, some less), the current system can handle about 7-8 rollups before hitting the target of 3 blobs per block. We’re already at 5-6 rollups consistently using blobs. Add in one more large rollup (say, Coinbase’s Base expanding, or a new zk-rollup for a major DeFi protocol), and we cross the target threshold.
Once that happens, blob basefee will spike, and rollup operators will pass those costs to users. My estimate? Two years from now, average L2 gas fees could be 2-3x current levels—back to $0.10-$0.30 per transaction, which feels cheap compared to L1, but is a 10x increase from today’s sub-cent fees. And for complex operations like swapping through a multi-step DeFi protocol, those costs add up to $1-$3 again.
I remember the 2022 DeFi deflationary crisis when Terra collapsed and every playbook ripped apart. The emotional barometer of users back then was pure panic. Post-Dencun, the vibe is the opposite—euphoric cheapness. But that’s precisely when the herd underestimates the cost of infrastructure. I’ve been in enough bull and bear cycles to spot the pattern: when something feels too cheap, the game theory shifts and people over-consume. That overconsumption will hit a wall.
Contrarian: The Unreported Blind Spot
What nobody is discussing is that rollup teams have zero incentive to optimize blob usage as long as fees are near zero. Why compress calldata when it costs nothing? Why batch transactions more efficiently? The current fee environment creates moral hazard—rollup operators are building for the cheap world, not the saturated one. When blob fees rise, they’ll scramble to implement compression algorithms, but by then, user trust will have eroded.
Another blind spot: EIP-4844 (proto-danksharding) was designed to be a stepping stone to full danksharding in the future. But that future is years away—likely not before 2027-2028. So the window between blob saturation and danksharding deployment is a dangerous gap. Projects that rely on consistently low L2 costs—like fully on-chain gaming, decentralized social, or micropayment streaming—could face a growth ceiling.
I’ve been hosting informal debates in Buenos Aires with developers from StarkNet and Optimism. The consensus among them is that blob scaling is not a priority. They’re focused on proving security and onboarding users. That’s short-sighted. The 2025 regulatory gridlock taught me that the best technical decisions are the ones that anticipate constraints, not ignore them.
Real-World Data Check
Let me share a specific data point from my monitoring. On May 20, 2024, during a wave of meme token activity on Base, blob usage hit 7 per block for three consecutive blocks—far above the target. The blob basefee surged to 800 gwei per blob (compared to the normal 1-10 gwei). That’s an 80x spike for one block. Fortunately, it didn’t last. But as more rollups launch, such spikes will become the norm, not the exception.
Based on my experience auditing Layer2 infrastructure during the 2021 NFT peak, I’ve seen how quickly a feature that’s “free” becomes a bottleneck. In 2026, as AI-agent trading bots flood the network with micro-transactions, blob demand will explode. We’re looking at a 10x increase in daily blob consumption by 2026. The supply is fixed. The result is inevitable: L2 gas fees will double, then triple.
The Takeaway
The sprint to the ETF finish line has distracted everyone from the real story underneath. While we obsess over Bitcoin price and regulatory ETFs, the plumbing of Ethereum’s scaling roadmap is silently tightening. If you’re a DeFi builder, start optimizing your calldata compression now. If you’re a user, don’t mistake current fees for permanent reality. The race isn’t won by the fastest roller–coaster; it’s won by those who understand that every cheap ride comes with a hidden toll. The next 12 months will test whether the ecosystem learned the lessons of 2022 or if we’re doomed to repeat them.