Ethereum's Layer 1 fee revenue has collapsed more than 90% year-over-year, falling from $113 million in L2-to-L1 payments in 2024 to approximately $10 million in 2025. Monthly mainnet fee revenue now sits below $15 million. Gas fees averaged 3 gwei in mid-March 2026, the lowest sustained level in...
"The year ahead is likely to mark Ethereum's L2 consolidation: a leaner, more resilient layer anchored by ETH-aligned, exchange-backed, and high-performance networks." — 21Shares, Ethereum 2026 Outlook Research Report
Ethereum's Layer 1 fee revenue has collapsed more than 90% year-over-year, falling from $113 million in L2-to-L1 payments in 2024 to approximately $10 million in 2025. Monthly mainnet fee revenue now sits below $15 million. Gas fees averaged 3 gwei in mid-March 2026, the lowest sustained level in over two years, with intraday lows touching 0.055 gwei on March 12. ETH trades at $1,983 as of March 27, 2026 — down from $2,250 weekly highs and carrying a market capitalization of approximately $233 billion.
The cause is structural, not cyclical. EIP-4844, activated with Ethereum's Dencun upgrade in March 2024, introduced blob transactions that reduced L2 data-posting costs by over 90%. Blob utilization currently sits at approximately 29% of the 14-blob target — massive surplus capacity that keeps pricing near floor levels. Meanwhile, the three dominant L2 sequencers — Base, Arbitrum, and Optimism — collectively earned approximately $161 million in annualized revenue, with Base alone capturing $93 million. The spread between what L2s charge users and what they remit to L1 has become Ethereum's most consequential economic divide.
The network has flipped from deflationary to inflationary. ETH supply now grows at approximately 0.23% annually, with validator issuance running at roughly 1,700 ETH per day while daily burns have fallen to approximately 3.26 ETH — a 71% decrease following the Pectra upgrade in May 2025. The "ultrasound money" thesis, predicated on fee-driven deflation, no longer holds under current network conditions.
Ethereum's mainnet transaction count has declined from 1.2 million daily a year ago to 1.05 million daily in March 2026. The migration is quantifiable: combined L2 transaction counts rose from 8 million to over 13 million daily in the same period. Total transactions across Ethereum and its L2 ecosystem reached 172 million per week in early March 2026, an all-time high. Yet the mainnet captures a diminishing fraction of the economic activity it enables.
The practical impact on users is dramatic. A simple ETH transfer now costs approximately $0.15. A Uniswap token swap runs $1.00–$2.00. Complex multi-step DeFi interactions remain under $5.00. For end users, this represents a significant improvement. For the network's economic model, it represents a structural revenue problem.
According to Blockworks, the ETH fee burn has dropped 78% year-over-year, pushing the network back to net inflationary supply dynamics. The forces driving that decline — cheaper L2 data availability, growing L2 adoption, and surplus blob capacity — are accelerating, not reversing.
Ethereum's quarterly fee revenue peaked at $4.3 billion in Q4 2021, according to Pine Analytics. The current annualized run rate is a fraction of that figure. The gap between peak and trough illustrates the magnitude of value that has migrated up the stack to L2 sequencers, MEV supply chains, and application-layer participants.
The three dominant L2 networks — Base, Arbitrum, and Optimism — process approximately 90% of all L2 transactions, according to 21Shares. The remaining 50-plus chains share approximately 10% of activity.
Sequencer economics tell the story. Base generated approximately $93 million in annualized revenue. Arbitrum produced roughly $42 million. Optimism contributed around $26 million. These figures represent the spread between user-paid fees and near-zero blob posting costs, plus MEV and priority fee extraction.
The L1 capture rate is minimal. Base, the single largest L2 by revenue, paid only $4.9 million in blob fees to Ethereum L1 in 2024 — a 5% capture rate on $92 million in revenue. According to KuCoin, Base contributed 71% of all OP Chain sequencer fees but paid only 2.5% back to the Optimism Collective. Coinbase captured 28 times more value than it contributed.
The economic structure functions as follows: L2s aggregate user fees, pay near-zero data availability costs to L1, and retain the spread. Ethereum provides settlement security and data availability. The L2 operators capture the margin. This is the inverse of what the "ultrasound money" thesis anticipated — that increased network utility would drive proportionally increased L1 fee revenue.
On February 18, 2026, Coinbase's Base announced a transition from the open-source Optimism Stack to a proprietary codebase. The move ends a three-year dependency and, critically, terminates Base's sequencer revenue sharing with the Optimism treasury.
The financial impact is material. In 2025, Base generated $74 million in chain revenue, accounting for over 71% of all OP Chain sequencer fees. Base's departure removes the largest single revenue source for the Optimism Collective treasury, which funds ecosystem incentives and a recently approved OP token buyback program. The Optimism governance had passed a proposal in February 2026 to allocate 50% of net Superchain sequencer revenue to monthly OP token purchases — a program now diminished by Base's exit.
Base's stated rationale is technical: greater control over its development roadmap, targeting six hard forks per year. The economic subtext is straightforward. Base was subsidizing a collective it no longer requires. The departure signals that the largest, most profitable L2 sees more value in independence than in shared infrastructure costs.
For Ethereum's L1, the Base exit is neutral in direct terms — Base still settles on Ethereum mainnet and pays blob fees. But it demonstrates a broader pattern: L2s that achieve scale have little economic incentive to share revenue horizontally with other L2 frameworks, let alone vertically with L1.
According to 21Shares, usage across smaller L2s has declined 61% since June 2025. Many operate as "zombie chains" — running infrastructure with minimal user activity and evaporating liquidity.
The Dencun upgrade's 90% fee reduction triggered aggressive fee wars that pushed most rollups into losses. According to multiple analyses, Base was the only L2 that turned a profit in 2025, earning approximately $55 million. Arbitrum's Q4 gross profit was approximately $6.5 million ($26 million annualized), sustained partly by a DAO treasury exceeding $150 million in non-native assets. Most other chains operate at a loss.
21Shares projects that "most Ethereum L2s may not survive 2026" as activity concentrates in Base, Arbitrum, and Optimism. The firm expects a consolidation around three categories: ETH-aligned networks that redirect fees back to Ethereum through burns or validator rewards; exchange-backed networks with built-in distribution (Base, Linea); and high-performance entrants such as MegaETH targeting near-real-time execution.
The L2 proliferation that began in 2023 — with over 50 chains launching or announcing — is entering a contraction phase. The economics are unforgiving: near-zero blob fees eliminate the margin available to smaller operators, while network effects in liquidity and developer tooling favor incumbents.
Ethereum's supply dynamics have fundamentally shifted. Post-Merge in September 2022, the network briefly achieved deflationary status as EIP-1559 fee burns exceeded validator issuance. That condition no longer holds.
As of February 2026, ETH supply grows at approximately 0.23% annually, according to 21Shares. Validator issuance runs at approximately 1,700 ETH per day, based on roughly 14 million ETH staked. Daily burns have collapsed to approximately 3.26 ETH per day — a 71% decrease following the Pectra upgrade. The math is unambiguous: 1,700 ETH issued versus 3.26 ETH burned equals net inflation of approximately 1,697 ETH per day, or roughly 619,000 ETH annually.
The "ultrasound money" narrative relied on sustained, high mainnet fee activity to drive burns above issuance. That activity has migrated to L2s. According to CoinTribune, $13.5 billion in ETH has been burned cumulatively, yet supply continues to grow. The mechanism works as designed — it simply requires more mainnet activity than currently exists.
Staking yields now rest primarily on consensus rewards and MEV rather than token burn dynamics. For ETH holders, the asset's monetary premium — to the extent it was priced on deflationary supply — faces re-evaluation.
The Ethereum community has not been passive. The Fusaka upgrade, activated in December 2025, introduced EIP-7918: a minimum blob base fee anchored to the L1 execution base fee. The mechanism sets a floor at 1/15.258 of the execution base fee, replacing the previous 1-wei minimum that left nodes inadequately compensated for KZG verification costs.
According to Fidelity Digital Assets research by Max Wadington, had this mechanism existed since Dencun, Ethereum would have accrued approximately $78.6 million in additional revenue. On 93% of days since Dencun, the adjusted fee would have exceeded the observed fee, according to Fidelity's analysis.
The implications extend to ETH's burn mechanism. Blob fees now contribute to burns, with projections showing an eightfold increase in blob-driven burns that could represent 30–50% of total burns by late 2026, contingent on L2 transaction volume growth.
EIP-7918 represents a pricing-power assertion in the data availability market. It does not solve the value capture problem entirely — L2s still retain the vast majority of user-paid fees — but it establishes a floor that prevents Ethereum from providing settlement services at effectively zero marginal cost.
The upcoming Glamsterdam upgrade, targeting mid-2026, includes enshrined Proposer-Builder Separation (ePBS) and Block-level Access Lists, further refining L1's role. Whether these changes materially alter the revenue split between L1 and L2 remains to be demonstrated.
Pine Analytics argues this is not unique to Ethereum. Bitcoin, Ethereum, and Solana have followed the same arc: fee revenue spikes, attracts attention, and gets competed away by L2s, private order flow, MEV-aware routing, or application-layer extraction. The pattern appears structural to open, permissionless networks.
Ethereum's case is the most documented. Quarterly fee revenue climbed from $231 million in Q4 2020 to $4.3 billion in Q4 2021. Then the compression cycle began. Post-Dencun, Arbitrum's per-transaction fees dropped from $0.37 to $0.012. Optimism fell from $0.32 to $0.009. The cost savings accrued to users and L2 operators, not to L1.
By 2026, the market has shifted from pricing L1s on fee capture to evaluating them on staking yields, ETF fund flows, real-world asset tokenization narratives, protocol upgrade expectations, and macro liquidity conditions, according to Pine Analytics. This represents a fundamental repricing of what L1 security layers are worth — and how that worth is measured.
The economic-value question is whether Ethereum's L1 can sustain its security budget as fee revenue trends toward a negligible share of total ecosystem value. Bitcoin faces an analogous challenge, requiring $54–72 billion annually to secure $115 million in fees, as documented in prior research. Ethereum's version of this challenge is arriving faster than expected because it explicitly designed for activity to move to L2s.
Ethereum's L2 scaling strategy has succeeded on its stated terms. Transactions are cheap. Throughput is high. User experience has improved materially. The cost of that success is a fundamental restructuring of where economic value accumulates in the Ethereum ecosystem.
The data shows value migrating from L1 to L2 sequencer operators — primarily Coinbase (Base), Offchain Labs (Arbitrum), and the Optimism Collective. Ethereum's mainnet increasingly functions as a low-margin settlement layer: essential for security, minimal in revenue capture. The network's response — EIP-7918's blob fee floor, upcoming ePBS, and gas limit increases toward 100 million — represents an attempt to reclaim some pricing power without undermining the scaling properties that drove L2 adoption.
Whether these measures are sufficient depends on two variables: the rate of L2 transaction growth (which increases blob demand and therefore blob fees), and the extent to which new mainnet-native use cases — real-world asset settlement, institutional custody operations, AI agent transactions — generate direct L1 fee revenue.
The market, at $1,983 per ETH, appears to be pricing in uncertainty about the answer. The economic model that justified Ethereum's valuation in 2021 — high fees, deflationary supply, direct fee-to-value linkage — no longer operates under current conditions. A new model is forming, but its contours remain incomplete.