Ethereum is about to undergo its most consequential architectural overhaul since The Merge. The Glamsterdam hard fork — targeted for mid-2026 — bundles up to 22 Ethereum Improvement Proposals into a single coordinated upgrade that rewires how the network builds blocks, processes transactions, and...
"Both of these facts, for their own separate reasons, mean that the original vision of L2s and their role in Ethereum no longer makes sense, and we need a new path." — Vitalik Buterin, Co-founder, Ethereum
Ethereum is about to undergo its most consequential architectural overhaul since The Merge. The Glamsterdam hard fork — targeted for mid-2026 — bundles up to 22 Ethereum Improvement Proposals into a single coordinated upgrade that rewires how the network builds blocks, processes transactions, and scales its base layer. At its center sit two headlining changes: Enshrined Proposer-Builder Separation (ePBS, EIP-7732), which moves the MEV auction from off-chain relay infrastructure directly into the consensus layer, and Block-Level Access Lists (BALs, EIP-7928), which enable parallel transaction execution across multiple CPU cores. Together, they represent a philosophical U-turn for a network that spent 2021–2024 telling the world its future was rollups.
The timing is not accidental. Ethereum's Layer 1 generates roughly $500,000 per day in fees — a fraction of the $30 million daily peaks it once commanded. Meanwhile, Layer 2 sequencers like Base extract $185,000 per day while returning just 2.6% of their revenue to mainnet in blob fees. With L2 decentralization progressing "far slower than expected" by Vitalik Buterin's own admission, and 93% of Ethereum blocks running through a handful of MEV-Boost relays controlled by three dominant builders, the case for strengthening the base layer has become urgent. Glamsterdam is Ethereum's answer: scale L1, enshrine fairness, and reclaim the economic gravity that has been drifting upward to intermediaries.
For four years, Ethereum's official position was clear: the base layer would optimize for security and data availability while rollups handled execution at scale. The "rollup-centric roadmap," formalized in late 2020, positioned Layer 2 networks as branded shards — secure extensions inheriting Ethereum's security guarantees while processing the bulk of user transactions.
By early 2026, that thesis has cracked under the weight of two uncomfortable realities.
First, L2 decentralization stalled. Most major rollups still operate with centralized sequencers, single points of failure that can censor transactions, reorder them for profit, or go offline entirely. Buterin acknowledged in February 2026 that "progress among layer-2s toward later stages of decentralization has been slower and more difficult than expected." The vision of L2s as trustless extensions of Ethereum remains largely aspirational.
Second, L1 can now scale directly. Post-Dencun and Fusaka, Ethereum's fee environment has changed dramatically. Average L1 fees hover between $0.34–$3.78, and gas limits have already risen to 60 million. The Ethereum Foundation's 2026 protocol priorities document, published February 18, organizes development into three explicit tracks: Scale, Improve UX, and Harden the L1 — with Glamsterdam as the vehicle for the first two.
The result is a strategic pivot: rather than outsourcing execution to L2s and accepting the economic leakage that entails, Ethereum is investing in making its own base layer fast enough, cheap enough, and fair enough to recapture activity.
Today, Ethereum validators who propose blocks typically don't build them. Instead, they outsource block construction to specialized "builders" who compete to assemble the most profitable block, with the transaction passed through off-chain "relays" — middleware infrastructure that sits between builders and proposers. This system, known as MEV-Boost, works but introduces critical trust assumptions and centralization vectors.
As of late February 2026, three entities — Titan (50.15% of blocks), BuilderNet (27.94%), and Quasar (16.22%) — control over 94% of Ethereum's block building. On the relay side, ultrasound.money (33%), titanrelay.xyz (21%), and two BloXroute relays (combined 31.7%) handle the vast majority of block routing. Roughly 93% of all Ethereum blocks use MEV-Boost.
EIP-7732 eliminates the relay layer entirely by moving the proposer-builder handoff into the consensus protocol itself. The mechanism works through a commit-reveal flow:
This removes the need for trusted intermediaries, standardizes MEV distribution rules at the protocol level, and strengthens censorship resistance by making it harder for any single entity to filter transactions. Flashbots, which pioneered the current relay infrastructure, has acknowledged this transition, having already migrated its operations to the decentralized BuilderNet in December 2024.
Ethereum currently processes transactions sequentially — one at a time, in order. This means a block containing 1,000 transactions that touch entirely different parts of the state still processes them as if they were dependent on each other.
BALs change this by requiring blocks to declare in advance which accounts and smart-contract storage slots they will access. This "access map" allows execution clients to:
The impact is substantial. Tomasz Stanczak, co-director of the Ethereum Foundation, stated at the Bankless Summit that the gas limit would rise to 100 million in H1 2026, with a path to 200 million following ePBS deployment — and potentially 300 million by year-end. That represents a 233–400% increase from the current 60 million gas limit, translating to dramatically higher throughput without requiring users to change anything about how they interact with the network.
Beyond the two headliners, developers narrowed an initial list of nearly 50 proposed EIPs down to 17 high-impact changes. Base (Coinbase's L2) published a detailed analysis identifying six proposals they consider most promising, organized into three categories:
Blob Scaling: Increased blob capacity directly benefits L2 economics. When blob usage is consistently saturated, L1 data availability becomes a bottleneck for L2s to support more on-chain activity. Expanding blob capacity provides the data availability headroom necessary for continued L2 growth.
Gas Repricing: Recalibrating the gas costs of specific operations to reflect their actual computational burden, ensuring that cheap operations aren't overcharged and expensive ones aren't underpriced.
Builder UX Improvements: Technical changes that make it easier and more predictable for builders to construct efficient blocks, reducing wasted gas and improving overall network utilization.
Maximum Extractable Value — the profit validators and builders capture by reordering, inserting, or censoring transactions — represents an estimated $3–7 billion annually across the blockchain ecosystem. On Ethereum specifically, MEV has created a three-layer centralization problem:
Glamsterdam attacks all three layers simultaneously. ePBS eliminates relays as trusted intermediaries. The protocol-level commit-reveal mechanism ensures builders cannot renege on commitments. And while Fork-Choice Inclusion Lists (FOCIL, EIP-7805) — which would force builders to include specific transactions — were ultimately deemed too complex to ship alongside ePBS in Glamsterdam, they remain a priority for the subsequent Hegota upgrade in late 2026.
The net effect: Ethereum's MEV supply chain moves from a trust-based, off-chain oligopoly to a trustless, on-chain market enforced by consensus rules.
Ethereum L1 as a platform. A 3–5x throughput increase at the base layer makes L1 competitive for use cases that currently default to L2s or alternative L1s. If Ethereum can process transactions at sub-dollar costs with 200 million gas blocks, the economic case for routing through a centralized L2 sequencer weakens.
Solo validators. ePBS levels the playing field by removing the need for validators to run sophisticated MEV extraction software or trust relay operators. Any validator can participate in the proposer role without technical disadvantage.
Application developers. Parallel execution and higher gas limits mean more complex smart contract interactions become feasible on L1, reducing the need to architect around L2 fragmentation.
Relay operators. The entire relay business model — including legacy relay infrastructure from Flashbots, BloXroute, and ultrasound.money — becomes obsolete once ePBS is enshrined. These entities have had years of notice and are already pivoting, but the revenue stream disappears.
Dominant builders. The current builder oligopoly benefits from exclusive orderflow deals and relay relationships. Protocol-enforced PBS standardizes the auction mechanism, potentially reducing the advantage of scale in block building.
L2 sequencer economics. This is the most significant long-term risk. If L1 becomes cheap and fast enough for a broader set of transactions, L2 sequencers lose their primary value proposition — cheap execution. Base's $94 million in annual sequencer profit (against just $4.9 million returned to Ethereum in blob fees) illustrates the magnitude of value at stake. That said, L2s offering genuine differentiation — privacy, application-specific optimization, or alternative execution environments — retain strong positions.
A subtler economic consequence: a dramatically higher gas limit means lower base fees per transaction, which means less ETH burned via EIP-1559. Ethereum shifted from deflationary to approximately 0.8% inflation post-Dencun. A 200 million gas limit could push net issuance higher still, unless transaction volume scales proportionally to offset the per-unit fee decline. This is an open debate among Ethereum researchers and has direct implications for ETH's monetary properties.
As of early March 2026, the development picture is nuanced:
The primary risk is ePBS complexity. If ePBS proves too challenging to finalize in time, it could be deferred to Hegota — the second 2026 upgrade targeting late in the year. This would preserve Glamsterdam's timeline but significantly reduce its impact, as BALs without ePBS delivers parallel execution but not the MEV reform that makes the throughput increase trustworthy.
A secondary risk is coordination across client teams. Glamsterdam requires synchronized updates across all execution and consensus clients — a process that historically takes months of testing. The Ethereum Foundation's shift to a biannual upgrade cadence (Glamsterdam in H1, Hegota in H2) was designed to create predictable shipping windows, but the ambition of the EIP list will test that framework.
Glamsterdam is Ethereum's biggest architectural change since The Merge, bundling 22 EIPs that rewrite block building (ePBS), enable parallel execution (BALs), and target a gas limit increase from 60M toward 200M.
The rollup-centric roadmap is officially being revised. Vitalik Buterin's February 2026 acknowledgment that the L2 vision "no longer makes sense" marks a strategic pivot toward L1 scaling.
93% of Ethereum blocks run through off-chain MEV infrastructure controlled by three builders. ePBS eliminates this dependency by enshrining the proposer-builder handoff in consensus.
L2 sequencer economics face disruption. Base earns $185K/day while returning 2.6% to Ethereum — a dynamic that L1 scaling directly challenges.
The June 2026 target is aspirational. BALs are on track, but ePBS complexity could force a split across Glamsterdam and Hegota.
Glamsterdam represents Ethereum's admission that decentralization cannot be delegated. For years, the network bet that Layer 2s would inherit its security while scaling its execution — a bet that produced fast, cheap rollups but also centralized sequencers, fragmented liquidity, and an L1 bleeding fee revenue to intermediaries.
The upgrade's dual architecture — ePBS for fairness, BALs for speed — addresses both the political economy of block production and the raw throughput limitations that pushed users off the base layer. If successfully deployed, Ethereum's L1 transforms from a slow, expensive settlement layer into a high-throughput execution environment where block construction is trustless and MEV distribution is protocol-enforced.
The stakes are proportional to the ambition. A 200 million gas limit with parallel execution could make Ethereum's base layer competitive with the throughput claims of rival chains while maintaining validator decentralization that justifies its market premium (over 1 million validators versus Solana's approximately 1,900). But ePBS is the most complex consensus change since proof-of-stake itself, and the June target leaves little margin for the kind of extended devnet debugging that Ethereum upgrades typically require.
What is clear is that Ethereum's core developers have decided the base layer matters too much to leave underbuilt. Glamsterdam is the beginning of that reconstruction.