Ethereum is executing the most significant strategic reversal in its history. After five years of championing a rollup-centric roadmap — the thesis that Layer 2 networks would handle the bulk of transaction volume while the base layer served as a passive settlement engine — Vitalik Buterin declar...
"If you create a 10,000 TPS EVM where its connection to L1 is mediated by a multisig bridge, then you are not scaling Ethereum." — Vitalik Buterin, Ethereum Co-Founder
Ethereum is executing the most significant strategic reversal in its history. After five years of championing a rollup-centric roadmap — the thesis that Layer 2 networks would handle the bulk of transaction volume while the base layer served as a passive settlement engine — Vitalik Buterin declared on February 27, 2026, that this model "no longer makes sense." The new direction: scale Ethereum's Layer 1 directly, with the Glamsterdam hard fork targeting a 3.3x gas limit increase and the introduction of parallel transaction processing in H1 2026.
This is not merely a technical pivot. It is a fundamental rethinking of where economic value accrues in the Ethereum ecosystem. The rollup-centric era inadvertently created a value extraction machine where Layer 2 operators — particularly Coinbase's Base — captured the lion's share of fee revenue while contributing minimal economic value back to L1 validators and ETH holders. With ETH trading below $2,000 and annual L1 fee revenue collapsing to roughly $65 million against $4-5 billion in staking inflation subsidies, the urgency is existential. Glamsterdam represents Ethereum's attempt to recapture its own economic destiny.
The stakes extend beyond Ethereum. If successful, this upgrade could demonstrate that a decentralized network can scale without sacrificing its core properties — a proposition that, as of today, remains unproven at the scale Ethereum is targeting.
The numbers tell the story of a strategy that worked technically but failed economically. In 2025, Layer 2 networks handled between 58% and 65% of all Ethereum transaction volume. Average gas prices on L1 plummeted from ~72 gwei in early 2024 to ~2.7 gwei by March 2025 — a 95% collapse. Daily network gas revenue fell from peaks of ~23 million gwei to ~6.3 million. The "ultrasound money" thesis, which depended on robust fee burn to keep ETH deflationary, was functionally dead: post-Dencun, only ~40,000 ETH are burned annually while ~960,000 ETH are newly issued to stakers, producing net inflation of approximately 920,000 ETH per year.
The concentration was stark. Base alone captured over 80% of L2 fee market share in 2025, generating approximately $55 million in profit — extracted almost entirely from the Ethereum ecosystem. Arbitrum held 5-10% of the remainder, followed by Abstract and Optimism at 3-5%. Base's average daily revenue exceeded $185,000, compared to Arbitrum's ~$55,000 per day. Meanwhile, most smaller L2s became what industry analysts call "zombie chains," with usage dropping 61% year-over-year.
The value extraction was one-directional. L2s paid minimal data availability fees back to Ethereum L1, while users, liquidity, and developer attention migrated off the base layer. A 21Shares research note warned that most Ethereum L2s may not survive 2026 as activity concentrates overwhelmingly on three dominant platforms.
The Ethereum Foundation's February 18, 2026 Protocol Priorities Update made the strategic shift official, collapsing the previously separate "Scale L1" and "Scale Blobs" tracks into a single unified "Scale" track. Co-executive director Tomasz Stanczak — who would announce his departure from the EF just days earlier on February 13 — had restructured the Foundation around three pillars: scaling mainnet, expanding blob capacity, and improving cross-chain user experience.
Buterin's February 27 blog post crystallized the new doctrine. Layer 2s, he argued, should focus on providing value "beyond basic scaling" — privacy features, application-specific design, ultra-fast confirmations, or non-financial use cases. Scaling itself would be reclaimed by L1.
Glamsterdam, targeted for H1 2026, is built on three interconnected technical pillars.
Parallel Transaction Processing (EIP-7928). Currently, Ethereum processes transactions sequentially — a single-lane highway regardless of how many CPU cores a validator runs. EIP-7928 introduces Block-Level Access Lists (BALs), which allow a block to declare in advance which accounts and smart-contract storage slots it will access. Transactions that touch independent portions of state — such as a token transfer and an unrelated DEX swap — can then execute simultaneously across multiple CPU cores.
This is not theoretical parallelism. BALs enable what engineers call "perfect parallel processing," where the scheduler knows at block-assembly time which transactions are independent. The result is a multi-lane execution model that converts available hardware capacity into actual throughput gains, rather than leaving most processor cores idle.
Gas Limit Increase to 100-200 Million. The current gas limit stands at approximately 60 million. The Foundation's 2026 priorities explicitly target "toward and beyond" 100 million in the first half of the year. Stanczak told the Bankless Summit that the limit would increase to 100 million initially, with a path to 200 million following ePBS deployment, and potentially 300 million before year-end. This represents a 3.3x to 5x increase in per-block computation capacity — the single largest throughput expansion since the Merge.
Combined with parallel processing, the target is to push Ethereum L1 toward 10,000 transactions per second, up from the current effective throughput of approximately 15-30 TPS.
Storage Cost Restructuring. Buterin's February 27 proposal introduces a critical economic distinction: temporary computation versus permanent state storage. Not all on-chain activity imposes equal cost on the network. A swap that reads and writes to existing contract storage imposes a fundamentally different burden than a transaction that creates new permanent state entries that every node must store indefinitely.
Under the proposed model, long-term storage becomes more expensive while ephemeral computation becomes cheaper. This restructuring aligns fee economics with actual resource consumption — a principle that current flat gas pricing systematically violates.
Perhaps the most consequential component of Glamsterdam is Enshrined Proposer-Builder Separation (ePBS), tracked as EIP-7732. MEV extraction on Ethereum is estimated at over $3 billion annually, representing a hidden tax that costs the average DeFi user 0.5% to 3% of transaction value.
The current MEV supply chain relies on external infrastructure: MEV searchers identify profitable transaction reorderings, builders assemble optimized blocks, and relays coordinate between builders and proposers. This ecosystem, dominated by platforms like Flashbots, operates outside of Ethereum's protocol — meaning it is neither censorship-resistant nor trust-minimized.
ePBS moves block-building logic directly into Ethereum's core protocol. Under the new model, builders assemble and cryptographically seal blocks, while proposers select the highest-paying block without seeing or tampering with its contents. Transactions are revealed only after finalization. This architectural change achieves three objectives simultaneously:
First, it democratizes MEV revenue by creating a protocol-level auction accessible to any builder, rather than a cartel-dominated off-protocol market. Second, it hardens censorship resistance by removing centralized relay operators from the critical path. Third, it enables parallel processing by guaranteeing that block structure is known at assembly time, allowing the BAL system to schedule parallel execution.
The economic implications are significant. If ePBS successfully captures and redistributes even a fraction of the $3 billion annual MEV extraction back to validators and the burn mechanism, it would meaningfully offset Ethereum's current inflation dynamics — addressing the core sustainability gap that our foundational economic analysis identified.
The Glamsterdam upgrade must be understood through the lens of Ethereum's economic sustainability crisis. As documented in our October 2025 economic value analysis, Ethereum generates approximately $65 million in annual chain fee revenue while distributing $4-5 billion in staking inflation. The network operates at an estimated subsidy ratio of 60:1 — for every dollar of fee revenue, sixty dollars of inflationary rewards are distributed to validators.
This dynamic has three potential resolution paths, and Glamsterdam addresses all of them:
Path 1: Increase Fee Revenue. A 3-5x gas limit increase, combined with parallel processing that attracts computation-intensive applications back to L1, could substantially increase absolute fee volume even at low per-transaction costs. If the gas limit reaches 200 million and utilization stabilizes at 50-70%, annualized fee revenue could plausibly reach $200-500 million — a 3-8x improvement over the current $65 million baseline.
Path 2: Recapture MEV Revenue. ePBS creates a mechanism for MEV revenue to flow through the protocol rather than around it. If the burn mechanism captures even 20% of the estimated $3 billion in annual MEV, that represents $600 million in deflationary pressure — potentially restoring the "ultrasound money" thesis.
Path 3: Attract Activity from L2s. If L1 becomes fast and cheap enough for mainstream DeFi, gaming, and agentic AI transactions, the value leakage to L2 operators narrows. This does not mean L2s disappear — Buterin explicitly endorses their continued role for privacy and specialized applications — but the default settlement location shifts.
However, there is a material risk. Increasing throughput does not guarantee proportional fee growth. Ethereum has demonstrated that capacity expansions often lead to lower unit prices that outpace volume gains — the Dencun upgrade's impact on blob fees being the most recent example.
Ethereum currently operates with approximately 975,000 active validators, with 35.8 million ETH staked (28.9% of total supply) across 80 countries. But centralization pressures are acute: Lido DAO controls 22.08% of staked ETH, Binance holds 9.15%, and the top five entities collectively control over 46% of all staked assets.
Increasing the gas limit to 200 million raises legitimate questions about hardware requirements. Larger blocks demand more bandwidth, storage, and processing power from validators. The Foundation's approach relies on two mitigations: parallel processing (which leverages existing multi-core hardware rather than requiring faster single cores) and the eventual deployment of ZK-EVM clients that allow validators to verify blocks using proofs rather than full re-execution.
The Hegota upgrade, planned for late 2026, would introduce Verkle Trees — replacing the existing Merkle Patricia Trees with a data structure that reduces proof sizes by an estimated 90%. This enables "stateless clients" that can verify the blockchain without storing hundreds of gigabytes of historical data, dramatically lowering the barrier to running a full node.
Still, the transition period between Glamsterdam's capacity expansion and Hegota's efficiency gains creates a window where node operation becomes more resource-intensive. Whether this window concentrates validation power further among well-capitalized operators or proves manageable for existing hardware remains an open and critical question.
Ethereum is abandoning the rollup-centric scaling thesis. Vitalik Buterin's February 27, 2026 blog post marks the official pivot to L1-first scaling, with Layer 2s repositioned for specialized functions rather than general-purpose computation.
Glamsterdam targets a 3.3-5x capacity increase. The combination of gas limit expansion (60M to 100-200M), parallel processing (EIP-7928), and ePBS (EIP-7732) represents the most ambitious single upgrade since the Merge.
The L2 economic extraction problem is real. With Base capturing 80%+ of L2 fee market share and generating $55 million in 2025 profit while L1 earned only $65 million in fees, the value leakage is no longer theoretical.
ePBS could reshape the $3 billion MEV economy. Moving block-building into the protocol creates a mechanism for MEV revenue to flow through Ethereum's fee-and-burn model rather than enriching off-protocol intermediaries.
The sustainability gap remains Ethereum's existential challenge. At $65 million in annual fee revenue against $4-5 billion in staking inflation, Glamsterdam needs to deliver a step-change in L1 economic activity, not merely a technical throughput improvement.
The decentralization risk window is real but time-limited. The period between Glamsterdam's capacity expansion and Hegota's Verkle Tree efficiency gains will test whether Ethereum's validator set can absorb higher resource requirements without further centralization.
Glamsterdam is Ethereum's most consequential bet since the Merge. The technical components — parallel processing, gas limit expansion, ePBS — are ambitious but grounded in years of research and development. The real question is whether scaling L1 capacity can reverse the economic dynamics that have eroded Ethereum's value proposition: collapsing fee revenue, persistent inflation, and systematic value extraction by Layer 2 operators.
The pivot carries risks. Ethereum is effectively admitting that its previous strategy created a parasitic layer of intermediaries that captured value without proportionally contributing to network security. The L2 ecosystem — particularly Base, which is operated by a publicly traded company generating $55 million in annual profit from Ethereum's infrastructure — has strong incentives to resist a shift that redirects economic activity back to L1.
But the status quo is untenable. A network that earns $65 million per year while distributing $4-5 billion in inflationary subsidies is not economically viable at any timescale. If Glamsterdam can push fee revenue toward $200-500 million annually and ePBS captures meaningful MEV revenue for the burn mechanism, Ethereum may finally close the sustainability gap that has defined — and undermined — its economic model since inception.
The market is skeptical, with ETH trading below $2,000 and down sharply from cycle highs. Whether Glamsterdam converts that skepticism into conviction will depend not on the elegance of the upgrade's engineering, but on whether the economic value actually flows.