OP Mainnet Keeps 321 for Every 1 It Pays Ethereum, Base Keeps 226, and ETH Supply Has Been Inflationary Since Dencun - The Parasitic Rollup Debate Explained With Numbers

General
1

The “parasitic rollup” debate has been simmering since Dencun shipped in March 2024, but the numbers have gotten stark enough that we need to have an honest conversation about what is happening to Ethereum’s economic model.

The Revenue Ratios

Using data from GrowThePie measuring retained fees versus L1 costs, the ratios are eye-opening:

L2 Revenue Retained per $1 Paid to Ethereum
OP Mainnet $321.31
Base $226.40
Arbitrum One $28.62
Scroll $3.54

These ratios represent the fees each L2 collects from users versus what they pay to Ethereum for data availability and settlement. OP Mainnet keeps $321 for every dollar it sends to Ethereum’s base layer. Before Dencun, these ratios were dramatically lower because L2s had to pay calldata costs on L1. Blob transactions reduced L2 data costs by 100-200x, and the L2s kept the savings rather than passing them entirely to users.

The Macro Picture

Ethereum L1 revenue: Daily network gas revenue fell from peaks of approximately $23 million down to roughly $6.3 million in 2025. The fee burn that powered the “ultrasound money” narrative has collapsed.

ETH supply: Since Dencun (April 2024), ETH supply has increased by 262,493 ETH worth approximately $621 million. Ethereum is no longer deflationary. The burn rate from L1 transactions cannot keep up with issuance because the highest-activity users have migrated to L2s that pay pennies for data availability.

L2 profitability: Base was the only L2 that turned a meaningful profit in 2025, earning around $55 million. Base captures over 80% of L2 transaction fee market share, with Arbitrum at 5-10% and Optimism at 3-5%.

Why This Happened

The Dencun upgrade did exactly what it was designed to do - reduce L2 costs by 10-100x. Arbitrum gas fees dropped from $0.37 to $0.012. Optimism dropped from $0.32 to $0.009. This was the explicit goal of EIP-4844.

The problem is that Ethereum’s economic model was not redesigned to account for this revenue shift. The protocol made L2 data posting cheap without creating a mechanism to capture value from the massive increase in L2 activity. L2 sequencers - which are still centralized in every major rollup - capture the margin instead.

The Two Camps

Camp 1: L2s are parasitic.
They use Ethereum’s security without paying for it proportionally. The $321:$1 ratio is exploitation. Ethereum should increase blob fees, implement revenue sharing, or force L2s to use based rollups that return value to ETH stakers.

Camp 2: This is working as intended.
Ethereum is becoming a settlement layer - like the internet backbone. TCP/IP does not charge based on the value of traffic it carries. Ethereum’s value comes from being the most secure, most decentralized base layer, not from extracting fees. ETH’s value accrues from demand for settlement security, not transaction fees.

The Reality

Both camps have valid points, and the truth is probably somewhere in between. Ethereum needs L2s to scale, and L2s need Ethereum for security. But the current economic arrangement - where L2s capture 99% of the value chain while Ethereum subsidizes their security - is not sustainable if ETH holders expect the token to accrue value.

The Fusaka upgrade is attempting to address this through EIP-7918 (blob pricing adjustments) and PeerDAS (increased blob capacity). But the fundamental question remains: can Ethereum be a low-margin settlement layer AND have a high-value native token?

I want to hear from L2 builders, ETH investors, and protocol researchers. Where do you stand on this?

2

The economic analysis here needs to go deeper than the headline ratios, because the $321:$1 number is misleading in isolation.

What the ratio actually measures:

The GrowThePie metric compares L2 user fees collected against L1 data posting costs. It does NOT measure the total economic value Ethereum provides to L2s. Ethereum provides:

  1. Settlement security - L2s inherit Ethereum’s $400B+ economic security for finality
  2. Liquidity access - L2 assets are bridged from Ethereum and derive value from the L1 ecosystem
  3. Brand and trust - “Built on Ethereum” is the strongest credibility signal in crypto
  4. Validator infrastructure - Thousands of independent validators secure every L2 transaction

None of these are captured in the fee ratio. A more honest comparison would be: what would it cost an L2 to replicate Ethereum’s security independently? The answer is billions of dollars in staked capital and years of trust-building. The $1 per $321 in retained fees is rent on a security guarantee that would cost orders of magnitude more to reproduce.

That said, the trend is concerning.

The issue is not that L2s are profitable - it is that the profit extraction is happening through centralized sequencers with zero revenue sharing. Every major L2 (Base, Arbitrum, OP Mainnet, zkSync, Starknet) runs a single sequencer controlled by the founding team. These sequencers capture:

  • Transaction ordering fees (MEV)
  • Priority fees from users
  • The spread between gas prices charged to users and actual L1 costs

This is not decentralized value extraction - it is a toll booth operated by a single entity. The DeFi purist in me finds this more concerning than the fee ratio itself.

What would fix this:

  1. Decentralized sequencers - Distribute MEV and sequencing revenue to a broader set of participants
  2. Based rollups - Use L1 validators for sequencing, sending revenue directly to ETH stakers
  3. Blob fee adjustments - Price blobs based on the economic value of settlement security, not just marginal data costs
  4. Revenue sharing mechanisms - L2s voluntarily or mandatorily returning a percentage of profits to L1

The market will eventually force a resolution. If ETH continues to underperform because value leaks to L2 tokens, capital will flow to based rollups or L1 competitors that solve the value accrual problem.

3

From an investment perspective, the parasitic rollup narrative is the single biggest headwind for ETH price action right now. Let me explain why the numbers matter even if the “working as intended” camp is technically correct.

The investment case for ETH pre-Dencun:

  1. ETH is money (used for gas, staking, collateral)
  2. EIP-1559 burns ETH proportional to usage, making it deflationary
  3. More usage = more burn = less supply = higher price
  4. This is the “ultrasound money” thesis

The investment case for ETH post-Dencun:

  1. ETH is money (still true)
  2. EIP-1559 burns are negligible because high-volume activity moved to L2s paying minimal blob fees
  3. ETH supply is now inflationary (262K ETH added since Dencun)
  4. The “ultrasound money” thesis is effectively dead until fee dynamics change

What the market is pricing:

ETH has been the worst-performing major crypto asset in 2026, down 36% while BTC and SOL outperform. The market is telling you that investors have noticed the value accrual problem. When Base generates $55M in profit that flows to Coinbase shareholders, not ETH holders, institutional allocators ask: why hold ETH instead of COIN stock?

The comparison to internet infrastructure is flawed because internet backbone providers (Level 3, Cogent, etc.) were terrible investments precisely because they became commoditized low-margin businesses. If Ethereum becomes the “internet backbone of value,” ETH the token could follow the same path - essential but not valuable.

My positioning:

I have reduced ETH allocation from 35% to 15% of my crypto portfolio over the past six months. The value accrual problem is structural, not cyclical. Until one of the following happens, ETH will continue to underperform:

  1. Based rollups gain significant market share (sends sequencing revenue to L1)
  2. Blob fees are repriced to capture more value
  3. A new narrative emerges that does not depend on fee burn (ETH as collateral/money premium)

The bull case for ETH is that it becomes the reserve asset of a $10T+ on-chain economy. But that thesis does not require high L1 fees - it requires ETH to be essential collateral. I am watching restaking growth and institutional staking demand as the key metrics, not L1 fee revenue.

4

I need to push back on the “parasitic” framing because it fundamentally misunderstands what L2s do for Ethereum.

The counterfactual matters:

Without L2s, where does the activity go? Not to Ethereum L1 - the gas costs would be prohibitive for most use cases. It goes to Solana, BSC, Avalanche, or centralized solutions. The relevant comparison is not “L2 fees vs L1 fees” - it is “Ethereum ecosystem with L2s vs Ethereum ecosystem without L2s.”

Before Dencun, L2 transactions cost $0.30-$1.00. That was too expensive for social applications, gaming, micropayments, and high-frequency DeFi. Post-Dencun, transactions cost under $0.01. This opened up entirely new use cases that would never have existed on L1 at any price.

The activity numbers:

Ethereum L2s collectively process more transactions than every other blockchain ecosystem combined. Base alone handles more daily transactions than Ethereum L1. This activity generates demand for ETH as the gas token on L2s, increases bridged TVL, and expands the Ethereum ecosystem’s footprint.

The revenue ratio context:

Yes, OP Mainnet keeps $321 per $1 paid to L1. But OP Mainnet also processes millions of transactions that generate $0 in L1 revenue if they do not exist. The $1 Ethereum receives is $1 more than it would get from those transactions on Solana.

Where I agree with critics:

Sequencer centralization is a legitimate problem. Every major L2 (including the one I work on) runs a centralized sequencer. The sequencer captures MEV and ordering revenue that arguably should flow to a more distributed set of participants. This is not parasitic - it is a temporary architectural choice during early rollup development. But “temporary” has lasted longer than anyone promised.

The path forward:

  • Shared sequencers (though Astria’s shutdown shows this is harder than expected)
  • Based rollups for applications that prioritize Ethereum alignment
  • Decentralized sequencer sets with revenue distribution
  • L2 native staking that uses ETH as collateral

The solution is not to make L2s more expensive - that kills adoption. The solution is to evolve the architecture so that value naturally flows back to the base layer through security demand, ETH collateral requirements, and L1-aligned sequencing.

5

Let me reframe this debate in business terms because the protocol economics arguments sometimes miss the forest for the trees.

L2s are businesses. Ethereum is infrastructure.

Base is a business unit of Coinbase that generated $55M in profit in 2025. Arbitrum is governed by a DAO with a treasury and a token. OP Mainnet is backed by the Optimism Collective with its own token economy. These are not charitable organizations building for the good of Ethereum - they are businesses optimizing for their own revenue and growth.

There is nothing wrong with this. Businesses should maximize revenue. But let us stop pretending that L2s are purely altruistic scaling solutions. They are profit-seeking entities that happen to use Ethereum for security.

The platform economics parallel:

This is the same dynamic that played out between Apple and app developers, AWS and SaaS companies, or Google and Android OEMs. The platform (Ethereum) provides foundational infrastructure. The businesses (L2s) build on top and capture most of the end-user revenue.

In each case, the platform eventually had to assert economic power:

  • Apple takes 30% of App Store revenue
  • AWS raises prices on profitable services
  • Google controls Android distribution terms

Ethereum has not yet asserted its economic power over L2s. The blob fee market is priced at marginal data cost, not at the value of security provided. This is like AWS charging based on electricity costs rather than the value of its reliability, scalability, and ecosystem.

What I think happens:

  1. Short term (2026): The debate intensifies as ETH continues to underperform. Ethereum researchers will resist “extractive” pricing because they prioritize adoption.

  2. Medium term (2027-2028): Market pressure forces a compromise. Based rollups gain traction as an “Ethereum-aligned” alternative. Blob pricing gets adjusted upward. Some L2s voluntarily implement revenue sharing to maintain Ethereum ecosystem goodwill.

  3. Long term (2029+): The market bifurcates between “Ethereum-aligned” L2s that share revenue and “independent” L2s that optimize for their own token. The ETH price reflects which camp wins.

The 21Shares prediction that most L2s will not survive 2026 is relevant here. Consolidation around Base, Arbitrum, and OP Mainnet makes the revenue concentration problem worse, not better.