L2s Extracted All the Value, Blob Fees Collapsed, and ETH Became a Gas Token for Chains That Dont Pay Back

defi_diana · February 6, 2026, 1:56pm

I’ve been running the numbers on Ethereum’s fee economics for the past quarter, and I need to share what I’m seeing because the data is genuinely alarming. This isn’t FUD - I’m a DeFi builder who’s deeply invested in this ecosystem. But the economic model that was supposed to make ETH a deflationary, value-accruing asset has broken down in a way that I’m not sure can be easily fixed.

The EIP-4844 Inflection Point

Let’s rewind to March 2024. The Dencun upgrade introduced EIP-4844, which created “blob space” - a dedicated, cheap data availability layer for L2 rollups. The goal was noble: reduce L2 settlement costs so users could transact for fractions of a cent instead of dollars. Mission accomplished. L2 transaction costs dropped by 90-99% overnight.

But here’s what the Ethereum Foundation seemingly didn’t model properly: when you reduce the cost of your primary product by 99%, you need 100x more volume just to break even. That volume never came - at least not in sufficient magnitude.

Pre-Dencun Ethereum fee revenue (annualized): ~$2.1 billion in L1 fees, with meaningful burn from L2 settlement transactions.

Post-Dencun reality: Blob fees are contributing negligible revenue to Ethereum. In many weeks throughout late 2025 and early 2026, total blob fee revenue has been measured in the tens of thousands of dollars. Not millions. Not hundreds of thousands. Tens of thousands - for a network that secures over $300 billion in value.

The Base Problem (And It’s Not Just Base)

As we discussed in the Base thread, Base keeps approximately $226 for every $1 it pays to Ethereum in settlement costs. That ratio should horrify anyone who holds ETH as an investment. Coinbase has built a massively profitable business on top of Ethereum’s security while contributing almost nothing back to the network that makes it possible.

But singling out Base is misleading - every major L2 operates with similar economics. Arbitrum, Optimism, zkSync, Starknet - they all settle on Ethereum for pennies while generating millions in sequencer revenue. The L2 ecosystem collectively extracted over $1.2 billion in revenue in 2025, while Ethereum’s blob fee revenue was barely a rounding error.

The Deflationary Narrative Is Dead

Remember “ultra-sound money”? The pitch was simple and compelling: EIP-1559 burns base fees, reducing ETH supply. If burns exceed issuance, ETH becomes deflationary. During the 2021-2022 bull market and even into early 2024, this worked. ETH supply was genuinely declining.

Post-Dencun, the math reversed. With L2 transactions no longer contributing meaningfully to L1 fee burns, and with beacon chain issuance continuing at roughly 800,000 ETH per year, ETH has become inflationary again. The net issuance is currently running at approximately +0.5% annually. That’s not catastrophic in isolation, but it’s a complete reversal of the narrative that helped ETH reach $4,800.

Vitalik’s Admission

In February 2026, Vitalik published a post acknowledging that the current L2-centric roadmap “no longer makes sense” from an economic perspective. This was extraordinary. The architect of Ethereum’s scaling strategy essentially admitted that optimizing for cheap L2 transactions had come at the expense of L1 value accrual. He proposed several potential fixes, but the damage to market confidence was immediate.

ETH/BTC ratio dropped to 0.019. Market dominance fell from 17% to 8%. The market was telling us something: ETH had become a gas token for chains that extract value rather than return it.

The Structural Problem

Here’s what makes this difficult to fix. The L2 roadmap created a structural misalignment of incentives:

L2s want cheap settlement - Every dollar they save on Ethereum fees is a dollar of margin. They have no incentive to pay more.
Users want cheap transactions - Users don’t care about ETH value accrual. They want $0.001 transactions.
ETH holders want fee revenue - They need burns to exceed issuance for the deflationary thesis to work.
Validators want sustainable yields - Without fee revenue, staking yields depend entirely on issuance, which dilutes non-stakers.

These four groups have fundamentally conflicting interests, and the current design resolves the conflict entirely in favor of L2s and users at the expense of ETH holders and validators.

The Numbers That Matter

Let me put this in perspective with some concrete data:

Metric	Pre-Dencun (Annualized)	Post-Dencun (Current)	Change
L1 Fee Revenue	~$2.1B	~$800M	-62%
L2 Settlement Fees to L1	~$350M	~$15M	-96%
Blob Fee Revenue	N/A	~$5M	New
ETH Net Issuance	-0.3% (deflationary)	+0.5% (inflationary)	Reversed
ETH/BTC Ratio	0.053	0.019	-64%

Can This Be Fixed?

I genuinely don’t know. The proposals I’ve seen fall into a few categories:

Increase blob fees through governance - Possible but risks driving L2s to Celestia/Avail
Gas limit increases to bring activity back to L1 - Happening with Pectra/Fusaka, but limited impact
Shared sequencing with fee redistribution - Promising but years away and Astria just shut down
L2 “security tax” - Would require social consensus that doesn’t exist

The uncomfortable truth is that Ethereum gave away its most valuable asset - data availability - for nearly free, and now it needs to claw back value from entities (L2s) that have built entire business models around cheap settlement.

What’s your take? Is the economic damage permanent, or can Ethereum find a way to realign incentives? I’m especially interested in hearing from L2 builders about whether they’d actually tolerate higher settlement costs.

Sharing this because I think the community deserves an honest assessment. I’m not selling my ETH, but I’ve significantly reduced my position size over the past quarter based on this analysis.

blockchain_brian · February 6, 2026, 1:58pm

Diana, your numbers are accurate but I think the framing misses some critical economic dynamics that haven’t played out yet. Let me push back on a few points as someone who’s been working on Ethereum’s protocol layer for nine years.

Blob Fees Are Low Because Blob Space Is Abundant

This is the part that keeps getting lost in the narrative. EIP-4844’s blob fee market works exactly like EIP-1559’s gas fee market - there’s a target utilization rate, and fees increase exponentially as demand exceeds the target. Right now, blob space is underutilized. We’re averaging about 3 blobs per block against a target of 3 and a max of 6 (pre-Fusaka). When demand consistently pushes past the target, blob fees will rise naturally through the same mechanism that made L1 gas fees expensive in 2021.

Think of it like early internet bandwidth pricing. In 1995, bandwidth was essentially free because nobody was using it. By 2005, ISPs were charging premium prices because demand had caught up with supply. We’re in the 1995 phase of blob space.

With Fusaka doubling the blob target to 6 and max to 9, there’s even more headroom. But as L2 transaction volume grows - and it is growing, 4x year-over-year - we’ll eventually hit the point where blob fee auctions become meaningful. The math suggests that at 10x current L2 volume, blob fees could contribute 200-400 million dollars annually to Ethereum. At 50x, we’re talking multiple billions.

Gas Limit Increases Are Already Changing the Equation

The gas limit increase from 30M to 36M gas that shipped with Pectra is already bringing some activity back to L1. More importantly, the ongoing conversation about pushing to 60M or even 100M gas would fundamentally change the value proposition. If L1 can handle 1,000+ TPS natively, many applications that moved to L2s purely for cost reasons would return to L1 where composability is native and settlement is instant.

This doesn’t replace the L2 ecosystem - it complements it. L2s handle the highest-throughput use cases (gaming, micropayments, social), while L1 handles the highest-value use cases (large DeFi trades, NFT marketplaces, institutional settlement). Each pays appropriate fees.

Shared Sequencing Creates New Fee Markets

I know Astria shut down, but that doesn’t invalidate the thesis - it just means the first implementation didn’t work. Espresso’s approach using EigenLayer restaking is fundamentally different and has better economic alignment. When shared sequencing eventually works (and I believe it will within 18-24 months), it creates an entirely new fee market where L2s pay for cross-chain atomic composability.

The Real Counterargument: Network Effects

Here’s what your analysis misses, Diana. ETH’s value doesn’t come solely from fee burns. It comes from being the collateral asset of an entire financial system. ETH backs over 43 billion dollars in DeFi TVL, is the primary collateral in lending protocols, and is the settlement currency for NFT markets. Even if fee revenue dropped to zero, ETH has monetary premium from its role as programmable collateral.

Bitcoin generates essentially zero fee revenue relative to its market cap and trades at 100K+. The “ETH needs fee burns to be valuable” thesis was always a bonus narrative, not the core value proposition.

I’m not saying the economics are perfect - they clearly need work. But calling ETH a “gas token” ignores the 300+ billion dollars in value that the network secures and the unique role ETH plays as on-chain money. The economic model can be tuned. The network effects can’t be replicated.

layer2_lisa · February 6, 2026, 1:59pm

Speaking as someone who has spent six years building L2 infrastructure - including time at both Polygon Labs and Optimism Foundation - I want to provide the operator perspective here, because the “parasitic L2” framing misses important nuance.

Yes, L2s Are Paying Minimal Fees. That Was The Goal.

Diana, I am not going to pretend the numbers you cited are wrong. L2s are paying pennies to settle on Ethereum, and that is exactly what Ethereum’s roadmap was designed to achieve. The entire point of EIP-4844 was to make L2 settlement cheap enough that end users could transact for fractions of a cent. Declaring this a failure because it worked too well is a strange argument.

But I will be honest about something the L2 ecosystem has not reckoned with: the relationship between L2s and Ethereum is currently one-directional in economic terms. We take security, we take the trust assumptions, we take the developer ecosystem - and we pay almost nothing for it. That is not sustainable long-term, and the smart L2 operators know it.

The Volume Projections Tell a Different Story

Let me share some numbers from our internal modeling that I think reframe the discussion. The key insight is that blob fees are non-linear - they increase exponentially once blob utilization exceeds the target. Here is what settlement costs would look like at scale:

At 10x current L2 volume:

Blob utilization hits approximately 80% of capacity
Annual blob fees to Ethereum: roughly 150-250 million dollars
Per-L2 cost increase: 5-10x current levels, still cheap for operators

At 50x current L2 volume:

Blob space becomes consistently congested
Annual blob fees to Ethereum: roughly 1.5-3 billion dollars
Per-L2 cost increase: 50-100x current levels, still manageable with revenue growth

At 100x current L2 volume:

Blob fee auctions become intensely competitive
Annual blob fees to Ethereum: roughly 5-10 billion dollars
L2s would need to pass some costs to users, still sub-cent transactions

The question is not whether Ethereum’s economic model works - it is whether L2 volume grows fast enough to fill blob space before the narrative damage becomes irreversible.

L2s Cannot Exist Without Ethereum

Here is what the “parasitic L2” critics never address: L2s derive 100% of their security from Ethereum. If Ethereum’s validator set weakened because ETH lost value, L2s would lose their entire security model. Base’s 47% L2 TVL share is only valuable because it is secured by Ethereum’s 300+ billion dollar validator set. No L2 operator with half a brain wants Ethereum to fail.

This is why I think L2s should voluntarily contribute to Ethereum’s security budget - not because governance forces them to, but because it is rational self-interest. Think of it as an insurance premium. If you are a major L2 generating hundreds of millions in revenue, paying 5-10% back to Ethereum is cheap insurance against the collapse of the security layer that makes your entire business possible.

A Concrete Proposal

I would advocate for an informal “Ethereum Security Contribution” where the top L2s by revenue voluntarily commit to either:

Direct ETH burns - Periodically buying and burning ETH proportional to sequencer revenue
Validator support - Running Ethereum validators with a portion of L2 revenue
Ecosystem funding - Contributing to Ethereum Protocol Guild or core development

This would not require a hard fork or governance vote. It would be a social commitment, similar to how major tech companies contribute to open-source projects they depend on. The L2s that refuse would face community pressure, and the ones that participate would build goodwill.

Brian’s point about network effects is exactly right. L2s have a vested interest in ETH maintaining its monetary premium. If we do not contribute to Ethereum’s security voluntarily, we risk governance-imposed solutions that would be far less favorable to L2 operators.

crypto_chris · February 6, 2026, 1:59pm

I want to step back from the technical debate and talk about what this means for anyone allocating capital to ETH, because that is ultimately what drives price and narrative.

The Fat Protocol Thesis Is Dead

Remember Joel Monegro’s “fat protocols” thesis from 2016? The core argument was that in crypto, unlike the internet, value would accrue to the base protocol layer rather than applications built on top. Ethereum was the poster child for this thesis - the idea was that as more applications launched on Ethereum, demand for ETH (for gas, staking, collateral) would drive its value up.

Ten years later, we can conclusively say this thesis has not materialized for Ethereum. Value has accrued to applications - specifically L2 rollups, which are generating billions in revenue while Ethereum’s own fee revenue has collapsed. Base, Arbitrum, and Optimism are the “fat applications” extracting value from a “thin protocol.” The exact opposite of what was predicted.

This is not a temporary market dislocation. It is a structural outcome of Ethereum’s own design choices. When you give away your most valuable resource (data availability) for near-zero cost and allow L2s to capture 99% of user-facing revenue, you have chosen to be a thin protocol.

The Investment Case Has Fundamentally Changed

Brian makes a valid point about ETH’s role as programmable collateral, but the market is not pricing ETH on collateral utility alone. Here is how I think about it:

The bull case for ETH in 2021-2023:

Deflationary supply via fee burns (broken)
Fee revenue growing with network usage (broken)
L2s would pay meaningful settlement fees (not happening)
“Ultra-sound money” narrative (dead)

What remains of the bull case in 2026:

ETH as DeFi collateral (valid but priced in)
Staking yield of ~3.5% (available elsewhere without inflation risk)
Network effects and developer ecosystem (real but declining)
Potential governance fixes (uncertain and contentious)

The remaining bull case is significantly weaker than what the market was pricing in at ETH’s peak. And critically, every proposed fix - minimum blob fees, security taxes, governance changes - requires hard forks or social consensus that could take years and might drive L2s to alternative DA layers.

My Current Position

I have been trading crypto for seven years and I will be transparent about my positioning. I am currently underweight ETH relative to my historical allocation:

ETH went from 35% of my portfolio in early 2024 to approximately 12% today
I rotated primarily into SOL, BTC, and select L2 tokens (particularly OP and ARB which are capturing the value ETH is losing)
I maintain some ETH exposure because the network effects are real and a governance fix is possible, but I am not betting on it

The ETH/BTC ratio at 0.019 tells you everything. This is not just a bear market - it is a repricing of ETH’s fundamental value proposition. Bitcoin at 100K generates negligible fees but has a clear narrative (digital gold, store of value). ETH at sub-2000 was supposed to have revenue-based valuation support that no longer exists.

What Would Change My Mind

I would increase my ETH allocation if any of the following happened:

Blob fees sustainably exceed 500 million dollars annualized - This would prove the economic model works at scale
Major L2s voluntarily commit to security contributions - Lisa’s proposal is interesting but I will believe it when I see it
Gas limit increases bring meaningful activity back to L1 - The 36M gas limit increase has not moved the needle
A credible governance proposal for minimum blob fees passes - Without driving L2s to Celestia

Until then, the risk-reward favors assets with clearer value accrual mechanisms. I say this as someone who genuinely wants Ethereum to succeed - I have built my career on this ecosystem. But the market does not care about loyalty, and right now the data says ETH is overvalued relative to its economic fundamentals.

Diana’s analysis is sobering but necessary. The community needs to stop treating criticism of ETH’s economics as FUD and start treating it as a problem to solve.

dao_david · February 6, 2026, 2:00pm

Governance is a marathon, not a sprint - and the ETH value accrual problem is fundamentally a governance challenge, not a technical one. Let me walk through what I see as the governance landscape for fixing this, because I spend my days in the trenches of DAO proposals and on-chain voting.

The Economic Model Can Be Fixed Through Governance

I want to push back on Chris’s pessimism. The economic model is not permanently broken - it is miscalibrated. Ethereum has a robust governance process (EIPs, AllCoreDevs calls, community signaling) that has successfully shipped radical economic changes before. EIP-1559 fundamentally restructured Ethereum’s fee model. The Merge changed the entire consensus mechanism. These were not small tweaks - they were structural overhauls that required broad social consensus.

The challenge with fixing the L2 value accrual problem is not that governance cannot do it. It is that the stakeholders are more divided than they were for EIP-1559 or the Merge, where almost everyone agreed on the direction.

Three Governance Proposals That Could Work

Here are the most credible proposals I have seen circulating in governance forums and AllCoreDevs discussions:

1. Minimum Blob Base Fee

The simplest approach: set a floor price for blob space so that even when demand is low, L2s pay a minimum settlement cost. Think of it like a minimum tax. This could be implemented as a protocol parameter change through a hard fork. The numbers I have seen discussed range from 1 gwei to 10 gwei per blob, which would generate roughly 50-500 million dollars annually at current L2 volumes.

Governance feasibility: Moderate. Core devs are open to discussing it, but L2 teams will lobby hard against it. The key argument against is that it breaks the “credibly neutral” fee market design of EIP-4844.

2. L2 Security Contribution Tax

A more aggressive approach: require L2s that use Ethereum for data availability to contribute a percentage of their sequencer revenue back to Ethereum. This could be enforced at the protocol level by requiring L2 contracts to include a “security contribution” mechanism.

Governance feasibility: Low in the near term. This would be extremely contentious because it requires defining what counts as an L2, measuring revenue on-chain, and enforcing compliance. It also sets a precedent for protocol-level taxation that many in the community would oppose on philosophical grounds.

3. Enhanced Burning Through L1 Activity

Rather than taxing L2s directly, focus on making L1 more attractive for high-value transactions. This means aggressive gas limit increases, better L1 composability, and potentially reserving certain protocol functions (like large staking operations, governance votes, or cross-L2 settlements) for L1 execution. More L1 activity means more burns.

Governance feasibility: High. Gas limit increases are already happening, and there is broad consensus that L1 should handle more activity. This is the path of least resistance, though it alone may not be sufficient.

The Celestia Threat Is Real

Here is what worries me most about the governance process: any proposal that meaningfully increases L2 costs risks pushing L2s to alternative data availability layers. Celestia already offers competitive data availability at a fraction of Ethereum blob costs. Avail is launching with similar economics. If Ethereum governance pushes too hard on extracting value from L2s, L2s will simply stop posting data to Ethereum.

This is not hypothetical. Several smaller L2s have already switched to Celestia for data availability. The major L2s (Arbitrum, Optimism, Base) remain on Ethereum primarily for the trust and brand association, not because it is cheaper. If governance makes Ethereum settlement significantly more expensive, the economic calculus changes even for them.

The governance challenge is therefore threading a needle: extract enough value from L2s to make ETH economics work, without extracting so much that L2s defect to competitors. This is exactly the kind of multi-stakeholder coordination problem that governance is supposed to solve, but it requires diplomatic skill and careful calibration.

Timeline Expectations

Being realistic about governance timelines: even if the community reaches consensus on a value accrual fix today, implementation would take 12-18 months minimum. EIP drafting, AllCoreDevs approval, client implementation, testnet deployment, hard fork scheduling - this process does not move fast. Chris is right that from an investment perspective, the uncertainty itself is a drag on ETH price.

But I would urge patience. Ethereum governance has repeatedly demonstrated the ability to make difficult, correct decisions - just slowly. The question is whether the market gives Ethereum enough time to fix its economics before the narrative damage becomes self-fulfilling.

Lisa’s voluntary contribution proposal is actually the most pragmatic near-term solution precisely because it bypasses the slow governance process. If major L2s commit to security contributions voluntarily in the next few months, it buys time for the formal governance process to develop a long-term solution.