ethereum layer-2 eip-4844 economics scaling

The upgrade that worked too well: Ethereum's blob fee paradox

bitcoinindex.net · · 5 min read
The upgrade that worked too well: Ethereum's blob fee paradox

The week ending March 30, 2025, Ethereum earned 3.18 ETH from blob fees. At $1,900 per ETH, that’s $6,000. For a network with a $230 billion market cap.

The week before, it was $22,000. Three weeks earlier, $160,000. The collapse wasn’t gradual. It was a cliff.

This is what success looks like. Ethereum’s Dencun upgrade delivered exactly what it promised: cheap Layer 2 transactions. Base fees dropped 99.8%, from $0.50 to $0.001. Arbitrum went from $1.50 to $0.02. Optimism cut 90%. Mission accomplished.

The problem nobody wanted to talk about: Ethereum L1 isn’t making money from the thing it was built to do anymore.

How blobs broke the fee model

Before EIP-4844 launched in March 2024, Layer 2s posted their transaction data to Ethereum using calldata. Permanent, expensive, stored in every block forever. L2s like Base were paying over $300,000 monthly just for data availability.

Dencun introduced blobs: temporary data packets that live for 18 days and cost a fraction of calldata. The target is 3 blobs per block, max 6. They use their own EIP-1559-style fee market: when demand exceeds the target, prices rise. When it falls below, they collapse toward zero.

In practice, blob usage has rarely exceeded the target. Post-Dencun, it hovered around 50-70%. After Pectra doubled capacity to 6 blobs in May 2025, utilization dropped to 67% of the new target. Galaxy Digital calculated the median blob cost at $0.00000000035. Nine zeros.

L2s got their cheap data layer. Ethereum L1 got virtually no revenue from it.

The Pectra problem

Pectra was supposed to be another step forward. Launched May 7, 2025, it doubled blob capacity from 3 to 6 per block. More headroom for L2s to scale. More transactions, more users, more ecosystem growth.

It also meant even less scarcity. With twice the blob supply and roughly the same demand, fees didn’t just stay low. They went lower. The 73% week-over-week drop in late March 2025 happened in a post-Pectra world where blob space was abundant and nearly free.

The roadmap makes this worse. Fusaka, deployed December 8, 2025, introduced PeerDAS (Peer Data Availability Sampling), which will scale blob capacity to 24-48 initially and eventually 128+ per block. Vitalik’s January 2025 roadmap calls for 100,000 transactions per second across Ethereum’s L2 ecosystem.

Every upgrade makes L2s cheaper. Every upgrade makes L1 blob revenue smaller. If blob fees are already near zero at 6 blobs per block, what happens at 128?

Where ETH value comes from now

Vitalik Buterin addressed this directly in his January 23, 2025 post on scaling Ethereum. He called out the need to “think explicitly about economics of ETH” and make sure the token continues to accrue value in an L2-heavy world.

His proposals: cement ETH as the primary asset across L1 and L2 (collateral, fees, DeFi). Encourage L2s to burn fees or stake permanently to Ethereum public goods. Explore based rollups so L1 validators can capture L2 MEV. Consider a minimum blob price and count on induced demand as capacity grows.

The last point is the gamble. If you raise blob capacity to 128 and fees stay at current levels, Ethereum could burn 713,000 ETH per year. But Vitalik added a caveat: “such a favorable demand curve is not guaranteed.”

Translation: we don’t know if L2 usage will grow fast enough to offset the supply increase. If it doesn’t, blobs stay free and Ethereum L1 earns nothing from them.

The maxi defense: it’s a feature, not a bug

The ETH maximalist position is that this isn’t a crisis. Ethereum isn’t trying to be a monolithic execution layer. It’s becoming a settlement layer, and settlement layers don’t need to capture every dollar of fee revenue.

Value accrues in other ways. Over 30 million ETH is staked, earning 4-6% yield. That’s $58 billion locked up, not counting DeFi collateral across L2s. Users bridge ETH to Layer 2s, creating lockup and scarcity on L1. Based rollups let validators capture L2 MEV. The network effect of dozens of L2s using ETH as their base asset matters more than blob fee income.

The historical precedent: AWS makes money on cheap compute. The internet protocols (TCP/IP, HTTP) generated zero direct revenue but enabled trillion-dollar ecosystems. Ethereum’s job is to be secure and credibly neutral infrastructure. Let L2s handle transactions. Let ETH’s value come from being the reserve asset of crypto.

It’s a compelling argument. It also requires faith that non-fee value drivers will sustain ETH’s price in perpetuity.

The skeptic’s worry

The counterpoint: ETH’s original value thesis was the “triple-point asset.” Capital asset (staking yield). Consumable commodity (gas for transactions). Store of value (deflationary via EIP-1559 burn).

L2 migration killed the consumable commodity part. L1 gas fees are low because most activity moved off-chain. Blob fees are near zero because supply exceeds demand. EIP-1559 burn, which briefly made ETH deflationary in 2021-2022, collapsed. Ethereum is net inflationary again.

If ETH’s only value driver is staking yield, it competes with Cosmos, Cardano, Solana, every other proof-of-stake chain offering similar returns. The moat is network effects and security budget, which are real. But they’re not infinite.

And if L2s decide to use alternative data availability layers like Celestia or EigenDA because they’re even cheaper, Ethereum loses the minimal blob revenue it has left.

What happens at 100,000 TPS

The Ethereum roadmap is ambitious. PeerDAS enables 128 blobs per block. Data compression and better L2 designs push toward 100,000 transactions per second. If that works, Ethereum becomes the settlement layer for global finance.

But if blobs stay virtually free at 128 per block, where does the revenue come from? Staking yield alone? DeFi lockup? Hope that L2s voluntarily burn fees back to L1?

Vitalik’s January 2025 post made it clear: this is an open question. The Ethereum community needs to figure out how ETH accrues value in a world where L1 executes almost no transactions and L2 data costs nearly nothing.

March 2025’s $6,000 blob fee week isn’t a bug. It’s a preview. Ethereum solved its scaling problem. Now it has to solve its business model.

Sources

Published February 24, 2026