Ethereum processed a record 200.4 million transactions in Q1 2026, a 38.0% quarter-over-quarter increase and 81.5% year-over-year gain. Monthly active addresses averaged 13.2 million, up 85.9% from a year earlier. By every usage metric, the network is at all-time highs. Fee revenue tells the oppo...
"The original vision of L2s and their role in Ethereum no longer makes sense, and we need a new path." — Vitalik Buterin, Ethereum Co-Founder, February 2026
Ethereum processed a record 200.4 million transactions in Q1 2026, a 38.0% quarter-over-quarter increase and 81.5% year-over-year gain. Monthly active addresses averaged 13.2 million, up 85.9% from a year earlier. By every usage metric, the network is at all-time highs.
Fee revenue tells the opposite story. Base-layer transaction fees totaled $39.9 million in Q1 2026, down 47.9% from the prior quarter and 81.9% year-over-year. The network's annual fee run-rate has collapsed from over $1 billion in 2022-2023 to roughly $160 million. Ethereum's circulating supply is mildly inflationary at approximately 0.23% annually, with total supply at ~121.5 million ETH, having expanded by roughly 950,000 ETH since the Merge. The "ultrasound money" thesis is, for now, structurally broken.
The cause is not declining demand. It is a deliberate architectural choice. The rollup-centric roadmap, executed through EIP-4844 and subsequent blob parameter expansions, shifted user activity to Layer 2 networks that pay Ethereum a fraction of what calldata fees once generated. Blob fee revenue has frequently measured in single-digit ETH per day. In 2025, Ethereum sacrificed over $100 million in fee revenue, according to CryptoSlate analysis, while Coinbase's Base network alone generated approximately $75 million in sequencer revenue. The question is no longer whether Ethereum's scaling strategy works. It does. The question is whether the base layer can recapture enough economic value to justify its $290+ billion market capitalization.
The data from Ethereum's Q1 2026 performance report illustrates a structural disconnect that has no precedent in crypto markets. Network throughput rose to 25.78 transactions per second, a 41.2% increase. The number of ETH holders grew to 292.8 million, up 24.9% year-over-year. Tokenized assets on Ethereum reached $203.4 billion, a 42.9% annual increase, according to Etherealize's quarterly report.
Meanwhile, fee revenue per transaction has compressed to near-zero in relative terms. A 38% increase in transaction volume produced a 48% decline in total fees. The mechanism behind this is straightforward: EIP-4844, activated in March 2024, introduced a separate blob data market for rollup data posting, reducing L2 data-posting costs by 80-90%. Average L2 transaction fees compressed from the $2-5 range to $0.001-$0.05 across major networks.
Prior to EIP-4844, Ethereum's L1 regularly collected more than $1 billion in annualized fee revenue, with peak months in 2023 exceeding $300 million in a single calendar month, according to Yellow Research. Post-EIP-4844, monthly protocol revenue fell by 60-80%. The second Blob Parameters Only (BPO) fork in January 2026 expanded data capacity further, compounding the decline.
The result is an anomaly in blockchain economics. The most-used smart contract platform, by a wide margin, is generating less fee revenue than multiple individual DeFi protocols operating on its own Layer 2 networks. GMX alone generates over $180 million in annualized protocol revenue on Arbitrum.
The economic value did not disappear. It migrated upward in the stack to Layer 2 sequencer operators.
Layer 2 networks collectively process transactions at a 5:1 to 10:1 ratio versus Ethereum mainnet, with aggregate daily counts exceeding 10 million on peak days against the mainnet's approximately 1 million daily ceiling, according to L2Beat data. Users pay transaction fees to L2 sequencers, not to Ethereum validators. The sequencer batches transactions, posts compressed data to Ethereum's blob space, and retains the margin.
The economics are stark. Fidelity Digital Assets Research estimated that Base, Ethereum's largest L2 by transaction volume, would incur approximately $30.6 million in annual blob fees to Ethereum under the new EIP-7918 fee floor, compared to $5.2 million paid over the prior year. In the same period, Base generated roughly $94 million in user-facing transaction revenue. The sequencer margin — the spread between what users pay and what the L2 pays Ethereum — accrues to Coinbase's balance sheet.
This is not unique to Base. The top three Ethereum L2s by TVL control approximately 80% of all sequencer fee revenue as of May 2026, according to Yellow Research. MEV extracted on L2 blocks accrues to L2 block builders or sequencer operators, not Ethereum validators. The gas fees paid by DeFi users on Base are denominated in ETH but collected by Coinbase's centralized sequencer infrastructure first.
The combined daily transaction count across all L2s has consistently exceeded Ethereum mainnet by 5-10x. Total value locked across all Ethereum L2 networks surpassed $45 billion in 2026. Yet the blob fees flowing back to Ethereum for this activity have been negligible. Outside of brief spikes from events like Blobscriptions and the LayerZero airdrop, the blob base fee spent most of 2024 and 2025 at or near 1 wei — effectively zero.
The blob fee market was, by design, too generous to rollups. EIP-4844 created a separate fee market for blob data with its own pricing mechanism. In practice, demand remained so low relative to supply that blob fees collapsed to 1 wei for extended periods. Rollups received near-free data availability. Ethereum captured almost none of the value it enabled.
The Fusaka upgrade, activated December 3, 2025, introduced EIP-7918 as a partial remedy. The proposal establishes a minimum blob base fee tied to 1/15.258 of the L1 execution base fee, preventing blob prices from falling to meaningless levels.
Fidelity Digital Assets Research modeled the counterfactual. Had EIP-7918 been in effect since the Dencun upgrade, Ethereum would have generated an additional $78.6 million (24,641 ETH) in cumulative revenue. On 93% of days in that period, the fee floor would have produced higher fees than the market actually generated. The average additional cost per blob would have been approximately $6.02.
The Fusaka upgrade also boosted blob base fees by approximately 15 million times from their previous near-zero levels, according to KuCoin research data. Blob fees are now included in ETH's burn mechanism, with estimates suggesting they could contribute 30-50% of total ETH burn, depending on L2 transaction growth trajectories.
The subsequent blob parameter expansions — targeting 48 blobs per block by mid-2026, with a long-term target of 128 blobs per slot under full Danksharding — create a tension: more supply reduces per-blob pricing, but higher throughput could generate more total revenue if demand scales proportionally.
The L2 landscape has bifurcated into a clear winner-take-most structure. As of late May 2026, the top three chains by TVL are: Arbitrum One at approximately $18 billion, Base at approximately $13.5 billion, and ZKSync Era at approximately $4.5 billion. Together, these three hold around $36 billion of the roughly $45 billion tracked across the Ethereum scaling stack.
Base, Arbitrum, and Optimism process nearly 90% of all L2 transactions. Base achieved over 4 million daily transactions on multiple occasions in early 2026. The remaining 50+ tracked chains compete for residual value.
The consolidation follows a predictable pattern. Mid-tier L2s with $200 million to $2 billion in TVL showed net capital outflows in Q1 2026. Chains below $500 million TVL largely stagnated or declined. Developer concentration data reinforces the trend: chains with 50+ monthly active developers retained growth, while those below saw median developer counts decline year-over-year. Over 100 application-specific rollups have been deployed via OP Stack or Arbitrum Orbit frameworks, further fragmenting the long tail.
The winning L2s won on distribution, not technology. Coinbase funnels retail users to Base. Arbitrum captured institutional rails and the largest DeFi protocols. The OP Stack federation gave Optimism a platform play. Technology — optimistic versus ZK — mattered less than go-to-market strategy.
Stablecoin yield products collectively represent over $8 billion in L2 TVL, anchoring sticky capital on the winning chains. The economic moats are widening as network effects compound: more users attract more protocols, which attract more liquidity, which reduces slippage, which attracts more users.
Ethereum's developer community is not unaware of the value capture problem. The next hard fork, Glamsterdam, represents the most direct attempt to pull economic activity back to the base layer since the Merge.
The upgrade introduces Enshrined Proposer-Builder Separation (EIP-7732) and Block-Level Access Lists (EIP-7928). Together, these move block building into the protocol and unlock parallel execution, laying groundwork for roughly tripling Layer 1 capacity. Roadmap materials outline a post-upgrade gas-limit target near 200 million gas per block, versus approximately 60 million today.
Additionally, Glamsterdam supports based rollups using preconfirmation mechanisms. This architectural shift would route MEV tips and transaction sequencing value to Ethereum validators rather than L2 sequencer operators — a direct reversal of the current value flow. If the design works as intended, some portion of the sequencer margin currently accruing to entities like Coinbase could flow back to the Ethereum validator set.
Timeline estimates vary. Some staking providers cite an internal target around Q3 2026. Given recent fork testing timelines, September to December 2026 is the firmer base case, according to Everstake analysis.
The strategic logic is coherent. If Glamsterdam delivers the projected 78% gas cost reduction and up to 70% MEV reduction on L1, some activity currently routed through L2s may find it economically rational to execute directly on Ethereum's base layer. Combined with the blob fee floor from EIP-7918 and projected blob count increases, the upgrade cycle aims to create a two-sided value recapture: higher L1 throughput attracts direct activity, while L2s pay more for the data availability they consume.
The risk is that L2 ecosystems have already accumulated sufficient network effects and user lock-in that base layer improvements arrive too late. Base has 4 million daily transactions and Coinbase's distribution engine. Arbitrum has $18 billion in TVL and entrenched DeFi protocols. Reversing capital and user flows is harder than preventing them.
Ethereum faces a paradox of its own engineering. The rollup-centric roadmap achieved its stated objective: cheap, scalable execution for hundreds of millions of users. It did so at the cost of economic capture by the base layer.
The numbers describe a network that is simultaneously more useful and less profitable than at any point in its history. Daily active addresses approach 2 million. L2 combined transactions exceed 10 million on peak days. Tokenized assets surpass $200 billion. And the base layer that secures all of it generates less revenue than a single derivatives protocol on one of its rollups.
Vitalik Buterin's February 2026 admission that the original L2 vision "no longer makes sense" marked a strategic pivot. The Fusaka blob fee floor and the upcoming Glamsterdam upgrade represent the corrective measures. Whether they arrive in time is an open question. L2 ecosystems have accumulated years of user adoption, liquidity, and developer tooling. Capital and users follow incentives, and the current incentive structure directs economic value away from ETH holders toward L2 sequencer operators.
The investment case for ETH currently rests on a bet that Ethereum can recapture value from its own scaling layer — a bet that requires both technical execution (Glamsterdam) and market dynamics (L1 activity growth) to converge. The data available today does not yet support that convergence.