Hook:

Blob count on Ethereum has dropped 60% from its peak in March 2024. Average blob fee: 0.001 gwei. Near zero. L2s are paying pennies for data availability. The network they were supposed to scale is now burning less ETH than before the upgrade. This is not the script.

I spent four months auditing the Hard Hat Protocol in 2017. I learned then that network congestion hides real demand. When congestion disappears, you see the skeleton. Blobs are that skeleton now. The Dencun upgrade’s promise was to make Ethereum scalable, cheap, and deflationary. One out of three? Not good enough.

Context:

Dencun went live on March 13, 2024. It introduced EIP-4844: proto-danksharding. Blobs — temporary, cheap data chunks attached to blocks — were designed to replace calldata for Layer-2 rollups. Before Dencun, L2s posted transaction data to Ethereum’s calldata, consuming expensive block space. Blobs were supposed to reduce that cost by 90-99%.

And they did. Optimism’s cost per transaction dropped from $0.15 to $0.001. Arbitrum followed. Base, zkSync, Starknet — all slashed fees. The market cheered. TVL on L2s surged past $40 billion. But something odd happened: blob usage plateaued, then declined.

The peak was in April 2024, when daily blob utilization hit 20,000+ blobs per day. By October 2024, that number hovered around 8,000-10,000. The hype faded. Demand didn’t follow.

Core: What the Data Says — and Why It Matters

Let me be direct. The blob market is a pricing failure. When supply outstrips demand structurally, price goes to zero. That’s what we’re seeing. Ethereum can support approximately 28,000 blobs per day (target 3 per block, 7,200 blocks, with 1.5x burst). Current usage is at 30-40% of capacity. That means the network is over-provisioned for the current level of L2 activity.

Key metrics from my dashboards:

• Blob fee trend: March 2024 average: 15 gwei. October 2024: 0.001 gwei. Collapse.
• Blob count 7-day moving average: 8,500. Down 58% from peak of 20,000.
• Ethereum base fee: affected by blob usage indirectly? No. Blobs don’t affect execution gas base fee. But total ETH burned from blob fees: negligible (< 1 ETH per day).
• L2 transaction growth: Base is at 10 million daily txs. Arbitrum at 2 million. Optimism at 1 million. Yet blob demand is flat. Why?

The answer is compression and batching. L2s have optimized their rollup batches to maximize the number of transactions per blob. Each blob is 128 KB. A single blob can contain hundreds of L2 transactions. So transaction volume can increase without taking more blob space. But the real question: is L2 user demand real? If transaction count rises but blob demand stays flat, it means each blob is more full. That's fine. But we also see L2 fees have hit a floor. Base’s average fee: $0.001. At that price, is there latent demand? No.

From my work reverse-engineering Uniswap V2 logic during DeFi Summer, I learned that artificial fee suppression often hides lack of organic demand. When fees are that low, bots and wash trading thrive. The real retail user? They’re not coming in droves.

Blob economics in a bear market:

In a bull market, demand for block space rises. L2s compete for blobs, fees rise, ETH becomes deflationary. In a bear market, the opposite happens. Blobs are empty. ETH inflation returns. Post-merge, Ethereum’s supply was deflationary by ~0.1% per year. With Dencun’s blob fee burn negligible, and execution base fee burning less due to lower activity, Ethereum is now slightly inflationary again (~0.3% annual).

This matters for ETH as a store of value. The narrative of "ultrasound money" is broken. The market hasn’t repriced it yet, but the data is clear.

Contrarian: The Unreported Angle — L2 Centralization Kills Blob Demand

Everyone talks about blob demand being low because L2 usage is low. That’s the surface. The deeper truth: L2 sequencers are centralized. They control batch frequency. They can choose to delay posting blobs to save on costs or to arbitrage. This behavior suppresses real-time blob demand.

Consider: Each L2 has a sequencer that orders transactions and then posts a batch to Ethereum. In a decentralized setup, multiple sequencers would compete to post batches quickly, driving blob demand. Today, Arbitrum One uses a single sequencer run by Offchain Labs. Optimism uses a single sequencer. Base uses Coinbase. zkSync uses Matter Labs. These sequencers have no incentive to maximize blob usage. They batch only as often as needed to keep costs low.

Data point: Arbitrum posts a blob approximately every 10-15 minutes. That’s 96-144 blobs per day. For 2 million daily transactions, that’s efficient. But what if Arbitrum were to post every minute? Blob demand would skyrocket 15x. They don’t because they can. There is no market force forcing them to batch faster.

The result: Blob demand is artificially low because the current L2 architecture is cartelized. Sequencers control the tap. Until decentralized sequencing becomes real, blob demand will remain structurally depressed regardless of user activity.

I’ve written before: Layer2 sequencers are basically single centralized nodes. "Decentralized sequencing" has been a PowerPoint for two years. The blob market is the evidence. Without competition, supply (blob space) will always outstrip demand.

Takeaway: What to Watch

Speed is the only metric that survives the crash. In this market, the L2s that will win are those that develop decentralized sequencer networks first. Not because they care about decentralization, but because it will force efficient blob usage and create real fee markets. Floors are illusions until the bot sees the spread.

Three signals to track:

1. Sequencer decentralization proposals — Espresso, Astria, Radius. If one of these gains real adoption, blob demand jumps.
2. Blob fee floor — If blob fees rise above 1 gwei consistently, it signals organic demand from genuine L2 activity, not just bots.
3. ETH supply growth — Watch the net issuance monthly. If inflation exceeds 0.5% annualized, the "ultrasound money" narrative will crack, leading to a repricing of ETH relative to BTC.

For now, Dencun is a technical success but an economic failure. The code works. The incentives don’t. That’s the story the data tells.

This analysis is based on my own monitoring dashboards and experience auditing smart contracts since 2017. Data sourced from Dune Analytics, Etherscan, and L2beat. I hold no material positions in any tokens mentioned.