It started with a quiet data point—something that should have made headlines but didn’t. Over the past week, the average blob utilization on Ethereum’s mainnet crossed 85% for the first time since the Dencun upgrade went live. A few chains like Arbitrum and Base are already competing for the same scarce resource: those precious 0.375 MB per block of blob space. What looked like a permanent fee reduction for Layer 2s is revealing itself as a temporary subsidy. The honeymoon phase is ending, and the math is unforgiving.
Context: The Promise of Proto-Danksharding
Let’s rewind. When EIP-4844 shipped with Dencun in March 2024, the narrative was simple: give rollups a dedicated data lane called “blobs” so they don’t clog the Ethereum base layer. Before Dencun, L2s posted transaction data to calldata, which was expensive because it competed with regular transactions for block space. Blobs were supposed to be cheap—like a budget airline seat for data. And for a few months, they were. Blob fees dropped to fractions of a cent, and rollup transaction fees fell from dollars to pennies. The Crypto Twitter celebrated. “Ethereum scales,” they said. But they forgot one thing: scarcity.
Core: The Technical Reality of Blob Saturation
Here’s the part most people skip. Each Ethereum block can hold a maximum of 6 blobs, totaling about 0.375 MB. That’s it. No sharding, no expansion—just a fixed pipe. The entire L2 ecosystem—Arbitrum, Optimism, Base, ZkSync, Scroll, Linea, and more—must share that single pipe. During the first three months after Dencun, blob usage hovered around 40–50% because not all L2s were actively posting blobs. But as more chains went live and user activity grew, the competition tightened.
Based on my experience auditing smart contracts and analyzing on-chain data for projects like OpenLedger Academy, I’ve seen this pattern before. When a resource is artificially cheap, demand explodes until it hits the ceiling. The current blob target is 3 per block, with a soft limit of 6. Once the average exceeds 3, the base fee for blobs starts to rise exponentially. We’re already there. On May 15, 2025, the average blob fee per block hit 0.002 ETH—a 50x increase from the post-Dencun lows. The trend is accelerating.
Let me give you a concrete example. I recently analyzed a week of blob data from Dune Analytics. On May 12, a single block had 6 blobs, and the total fee paid for those blobs was 0.01 ETH. That’s $30 at current prices. A month ago, the same blobs would have cost $0.50. The fee multiplier is 60x. Now imagine what happens when all major L2s are operating at full capacity. The blob base fee will spike to multiple ETH per block, and rollups will have to pass that cost to users. The $0.01 transaction will become $0.10, then $0.50, then $1.00. The era of cheap L2 transactions is a finite resource, and we’re burning through it faster than anyone predicted.
Why is this happening? Because the blob market is identical to the calldata market it replaced—just a different storage slot. The only difference is that blobs are temporary (expire after ~18 days) and have a separate fee market. But the core economics are the same: a fixed supply of data space, competing demands from multiple rollups, and a fee mechanism that penalizes congestion. The Dencun upgrade didn’t solve the scaling problem; it just kicked the can down the road by creating a temporary cheap lane. Once that lane fills up, we’re back to square one.
Democracy isn’t a transaction where every voice holds weight. The same applies to data availability. If every rollup has equal access to blob space, the market will price it until only the richest applications can afford it. That’s not decentralization—that’s a plutocracy of protocols.
Contrarian: The Blind Spots of the Optimism Narrative
Here’s where I part ways with the mainstream. Most Ethereum evangelists will tell you that the solution is to upgrade to full Danksharding, which will increase blob capacity to 16 MB per block. That’s coming, they say, in a few years. But I’ve been in this space long enough to know that timeline is optimistic. Full Danksharding requires complex changes like peer-to-peer data availability sampling, which is still in research phase. Even if it ships by 2027, the demand for blob space will grow faster than the supply. The entire history of blockchain scaling is a story of supply always lagging demand.
Consider this: the number of active L2s has grown from 10 to over 40 since Dencun. Each one wants to post blobs. Meanwhile, the number of Ethereum blocks per day is fixed (about 7,200). The blob capacity per day is roughly 2,700 MB (6 blobs × 0.125 MB × 7,200 blocks). That’s 2.7 GB per day for the entire L2 ecosystem. To put that in perspective, a single popular NFT collection on Base can generate 100 MB of transaction data in a day. We’re already consuming 1–2 GB per day on peak days. The math doesn’t add up.
Another blind spot is the assumption that L2s will compete on price. They won’t. They’ll compete on user experience, and that means they’ll pay whatever it takes to ensure their transactions are included. Base, backed by Coinbase, has deep pockets. Arbitrum has a treasury of over $1 billion. They can afford to bid up blob fees. Small L2s will be priced out, leading to a consolidation of blob usage among a few dominant chains. That’s the opposite of the “many rollups” vision that Vitalik articulated.
Decentralization is a verb, not a noun. It requires constant effort to maintain. If we don’t design for scarcity from day one, we end up with a system that centralizes power by default.
Takeaway: The Path Forward
So what’s the answer? More blob capacity is necessary but not sufficient. We need a fundamental rethinking of how L2s interact with Ethereum. One idea: require L2s to use compression or batching techniques that reduce blob footprint. Another: implement a blob futures market where chains can pre-purchase capacity, similar to how airlines sell cargo space. But the most radical solution is to stop pretending that every transaction needs to be finalized on Ethereum. High-value transfers can use blobs; low-value microtransactions should use alternative settlement layers like Lightning or state channels. That’s the hard truth: Ethereum can’t be the settlement layer for all of humanity’s transactions. It has to be selective.
I’m not saying rollups are a failure. They’re a necessary step. But we must stop treating Dencun as a final solution. Every upgrade creates new bottlenecks. The question is whether we’re willing to look beyond the hype and see the structural limits. The next bull run will be brutal for those who ignore blob economics. The winners will be the protocols that optimize for data efficiency, not just marketing.
Perhaps the most honest thing we can say is that crypto is still in its early stages. We’re building the tracks while the train is moving. And sometimes, the tracks lead to a cliff. The choice is ours: fix the design now, or let the market crash teach us the lesson.