The market is wrong. Cheap rollups are a temporary subsidy, not a permanent feature.
Post-Dencun, the narrative was clear: L2s are now ultra-scalable, fees will stay low forever, and Ethereum’s data availability moat is unbreakable. One month of data tells a different story. Blob usage is accelerating, not stabilizing. At current adoption curves, Ethereum’s blob space will hit 100% utilization within 18 months. When that happens, rollup gas fees will double, triple, and then some. The liquidity mirage of 2020 is repeating, except this time the bottleneck is data, not blockspace.
Let me show you the numbers. Since the Dencun upgrade, blob usage has grown at a compound weekly rate of 14%. At this pace, the target blob count of 6 per block (the soft cap) will be exceeded by Q2 2025. Once the hard cap of 48 blobs is consistently hit, rollups will face fee spikes reminiscent of the 2021 NFT mania. The cost of posting a batch to L1 will rise sharply, and those costs will be passed directly to end users.
This is not a prediction. It is a mathematical observation. The emission schedule of blobs is fixed. The demand function is growing exponentially. Basic supply-demand mechanics dictate the outcome.
The Core Mechanism: Why Blobs Are Not Infinite
Dencun introduced blobs as ephemeral data storage for rollups. Each blob is 128KB, and the protocol targets 6 blobs per 12-second slot, with a maximum of 48. This is a 8x increase over the previous calldata capacity, and it’s a one-time boost. Post-Dencun, the growth in L2 activity—driven by Base, Arbitrum, Optimism, and zkSync—has eaten into that slack rapidly.
According to Dune Analytics, total blobs posted per day increased from 2,000 immediately after Dencun to over 12,000 within six weeks. That’s a 6x increase in two months. If this pace holds, the daily average will reach the equivalent of 48 blobs per slot by mid-2025.
Why does it matter? Because blobs are not competing for blockspace like regular transactions. They have a separate pricing mechanism with a target and a base fee. Once the target is exceeded, the base fee for blob gas increases exponentially. This is the same mechanism that caused Ethereum L1 fees to spike in 2021. Except now, the demand is not from NFTs—it’s from every single L2 transaction settling on Ethereum.
The Hidden Variable: L2 Data Compression
Proponents argue that L2s will improve compression, reducing the need for blobs per transaction. This is true in theory, but in practice, the improvement is marginal. The current batch compression ratios are already close to theoretical limits for most rollups. Further gains require breaking backward compatibility or introducing new transaction formats—neither of which is happening soon.
Based on my audit of L2 fee models for a São Paulo-based fund, I found that even with 50% compression improvement, the growth in transaction volume will outpace compression savings within six months. The reason is simple: L2s are subsidizing user activity with low fees, which encourages elastic demand. As fees stay low, usage grows. As usage grows, blob demand increases. As blob demand increases, fees rise, killing the low-fee environment that attracted users. It is a cycle that ends only when blob supply is exhausted.
The Contrarian Angle: Decoupling Is a Myth
The macro watcher in me sees a parallel to the 2017 ICO bubble. Back then, the narrative was that Ethereum would decouple from Bitcoin because of unique utility. The decoupling lasted four months. Then liquidity dried up, and ETH crashed harder than BTC.
Today, the narrative is that L2s decouple Ethereum from L1 congestion. The truth is that L2s are directly coupled to Ethereum’s data availability capacity. If Ethereum gas fees spike, L2 fees spike. There is no escape. The only decoupling that matters is between user demand and blob supply, and that decoupling is impossible without a protocol change.
I shorted NFT ETFs in 2021 when I saw the same pattern: demand outpacing supply, with no scalable fix. I called the top within a week. This time, I am not calling a top, but I am calling the end of the cheap rollup honeymoon.
The Practical Taking: What You Do Now
If you are building a dApp on an L2, plan for fee volatility. Lock in gas as a business expense, like an airline hedges fuel costs. If you are an investor, look at protocols that minimize blob usage per transaction—those will have a competitive advantage when fee spikes hit. If you are a speculator, do not buy the narrative that L2 fees stay low forever. Utility is dead. Long live speculation.
Yields are taxes on risk you don’t see. The risk here is that cheap blobs mask the real cost of L2 settlement. When the cost resets, the yields on L2-based strategies (liquidity mining, leveraged yields) will evaporate. The question is not if, but when.