Ethereum's Glamsterdam upgrade — the largest protocol change since the September 2022 Merge — entered public testing on August 20, 2026, when the Platåberget testnet activated the fork's ten Ethereum Improvement Proposals. The Sepolia testnet fork is provisionally scheduled for September 28, 2026...
"Any tool that relies on a hardcapped maximum gas limit — think wallets, indexers, and gas estimators — will break and needs to be updated." — Ethereum Foundation, Platåberget Testnet Announcement, August 17, 2026
Ethereum's Glamsterdam upgrade — the largest protocol change since the September 2022 Merge — entered public testing on August 20, 2026, when the Platåberget testnet activated the fork's ten Ethereum Improvement Proposals. The Sepolia testnet fork is provisionally scheduled for September 28, 2026, with Hoodi on October 5 and mainnet targeted for November 4.
The upgrade's two headline changes — Enshrined Proposer-Builder Separation (EIP-7732) and Block-Level Access Lists (EIP-7928) — address a structural deficiency that has worsened since the Merge: as of mid-2026, two to three block builders control 80–95% of Ethereum L1 blocks, routing through three relays that handle roughly 88% of all MEV-Boost payloads. Glamsterdam moves block building into the consensus layer itself, making the relay middleware that currently mediates billions of dollars in Maximal Extractable Value optional rather than essential. Simultaneously, BALs unlock parallel transaction execution, clearing the technical path for validators to signal the gas limit from 60 million to a 200 million target — roughly tripling L1 throughput capacity.
The stakes are material. Ethereum processes approximately $2.76 trillion in market capitalization and hosts the majority of DeFi total value locked. A failed fork carries systemic risk; the December 2025 Fusaka upgrade saw a Prysm client bug cause 248 missed blocks across 42 epochs, reducing network participation to 75% and costing validators approximately 382 ETH. How Glamsterdam navigates its testnet gauntlet will determine whether Ethereum can reclaim L1 capacity lost to rollups and competing chains.
Glamsterdam ships ten EIPs under the Meta EIP-7773 umbrella. The package modifies both Ethereum's consensus layer (how validators agree on blocks) and execution layer (how transactions are processed). The two headline proposals — EIP-7732 and EIP-7928 — represent the deepest structural changes to Ethereum's block production pipeline since Proof of Stake replaced Proof of Work.
The full EIP roster:
| EIP | Change | Layer | |-----|--------|-------| | EIP-7732 | Enshrined Proposer-Builder Separation (ePBS) | Consensus + Execution | | EIP-7928 | Block-Level Access Lists (BALs) | Execution | | EIP-7688 | Forward-compatible consensus data structures | Consensus | | EIP-7708 | ETH transfers emit ERC-20-style logs | Execution | | EIP-7778 | Block gas accounting without refunds | Execution | | EIP-7843 | SLOTNUM opcode | Execution | | EIP-7954 | Contract size limit increase (24 KiB → 64 KiB) | Execution | | EIP-8007 | Coordinated gas repricings | Execution | | EIP-8024 | Backward-compatible SWAPN, DUPN, EXCHANGE opcodes | Execution | | EIP-8037 | State-cost gas dimension | Execution |
Since the Merge, Ethereum block production has operated through an informal middleware layer. Validators (proposers) outsource block construction to specialized builders via MEV-Boost, a system maintained by Flashbots. Relays — third-party intermediaries — shuttle sealed blocks between builders and proposers. None of this exists in the Ethereum protocol specification; it runs as sidecar software.
EIP-7732 moves this arrangement into the protocol itself. Under ePBS:
The practical effect: the relay layer that currently handles 88% of Ethereum blocks becomes optional. Builders compete in a permissionless, protocol-enforced auction rather than through opaque bilateral relationships with relay operators.
The urgency behind ePBS is quantifiable. Block builder market concentration has tripled since 2022, according to on-chain data tracked by multiple analytics providers:
This concentration is self-reinforcing. Exclusive order flow deals between builders and transaction originators create a feedback loop: more order flow produces more profitable blocks, which wins more auctions, which attracts more exclusive order flow. Without protocol-level intervention, the market trends toward monopoly.
Vitalik Buterin has stated that Glamsterdam brings ePBS specifically so that "builder centralization does not leak into validator centralization" via the current permissioned relay system. The concern is not merely theoretical — if builders can selectively exclude transactions, censorship resistance degrades. In September 2022, during OFAC sanctions compliance, relay-level filtering briefly censored over 70% of Ethereum blocks.
EIP-7928 addresses a different bottleneck: sequential transaction execution. Currently, every Ethereum transaction must be processed one after another because the EVM cannot know in advance which state (account balances, storage slots) each transaction will read or write. This sequential constraint limits throughput regardless of how much hardware validators run.
BALs solve this by recording, at the block level, every account and storage location accessed during execution, along with post-execution values. With this information made explicit:
According to Toni Wahrstätter, an Ethereum researcher, BALs enable "parallel disk reads, parallel transaction validation, parallel state root computation, and executionless state updates." World Chain has already adopted EIP-7928 for its mainnet, streaming access-list data in 200-millisecond slices rather than waiting for full block completion.
Ethereum's gas limit — the maximum computational work per block — has increased in steps: 30 million through most of 2024, then 60 million following the Fusaka upgrade in December 2025. Glamsterdam does not directly set a new gas limit; validators signal their preferred limit through a voting mechanism that requires no hard fork.
However, Glamsterdam removes the technical barriers that currently prevent validators from safely signaling above 60 million:
The Ethereum Foundation has set 200 million as the design target — roughly 3.3x the current limit. Under realistic workloads, proponents estimate this translates to up to 10,000 transactions per second equivalent throughput on L1. The increase would be gradual; validators would step the limit up only as node operators confirm propagation and execution remain stable at each threshold.
Beyond the two headliners, several supporting changes have significant operational implications:
EIP-2780 (Resource-Based Intrinsic Gas) breaks the flat 21,000 gas base fee for transactions into measured components. Under the new model, a zero-value transaction costs approximately 15,000 gas; a self-transfer approximately 12,000 gas. Plain ETH transfers to existing accounts remain near 21,000 gas. These figures are provisional and subject to change during testing.
EIP-8037 (State-Cost Gas Dimension) introduces a separate gas meter for state-creating operations, charged at a fixed cost per state byte. Standard transfers to existing accounts are unaffected, but transfers to new accounts incur additional state gas charges. This creates economic pressure against state bloat — a prerequisite for safely raising the gas limit.
EIP-7954 (Contract Size Limits) increases the maximum deployed contract size from 24 KiB to 64 KiB, and the maximum initcode from 48 KiB to 128 KiB. This change addresses a persistent constraint for complex smart contract development, particularly for protocols deploying sophisticated DeFi logic or on-chain governance.
EIP-7708 (ETH Transfer Logs) makes native ETH transfers emit logs matching the ERC-20 Transfer event format. This standardizes event tracking across all token types, reducing indexer complexity and eliminating a class of integration bugs where ETH movements were invisible to standard event-monitoring infrastructure.
The current schedule, as confirmed by six client teams during the August 20, 2026, developer call:
| Milestone | Date | |-----------|------| | Platåberget testnet launch | August 17, 2026 | | Glamsterdam fork on Platåberget | August 20, 2026 | | Sepolia testnet fork | September 28, 2026 (provisional) | | Hoodi testnet fork | October 5, 2026 (provisional) | | Mainnet activation | November 4, 2026 (target) |
The schedule carries acknowledged risk. Developer call minutes note explicitly that the Sepolia date was "deferred to the next call" for final confirmation, and client problems or devnet issues could push timelines further. Historical precedent supports caution: the Fusaka upgrade's Prysm client bug caused 248 missed blocks, and past forks have typically required two to four months of public-testnet testing before mainnet deployment.
The Ethereum Foundation has issued a specific warning that wallets, indexers, and gas estimation tools that hardcode the gas limit as a fixed value "will break and need to be updated." This encompasses a wide range of ecosystem infrastructure — exchanges, DeFi frontends, block explorers, and analytics platforms — that must complete upgrades before mainnet activation.
Glamsterdam's changes redistribute economic value across Ethereum's stakeholder map:
Relay operators face diminished relevance. Flashbots, bloXroute, and Ultrasound relay — which collectively route approximately 88% of blocks — will see their intermediary role become optional. Their business models must pivot from block relay to other MEV-related services.
Block builders face a more competitive market. Protocol-enshrined auctions with on-chain collateral lower barriers to entry compared to the current system, where reputation and exclusive order flow create winner-take-most dynamics. The 1.4 ETH entry subsidy barrier may decrease.
Validators gain direct access to builder auctions without trusting relay middleware, and the expanded data propagation window (2s to 9s) reduces the advantage of low-latency co-located infrastructure.
L2 rollups face a more competitive L1. If gas limits reach 200 million, L1 transaction costs decline substantially, potentially reducing the economic case for rollup-based execution. This creates competitive pressure on Optimism, Arbitrum, Base, and other L2s that have built business models around L1 congestion.
Application developers benefit from higher contract size limits, standardized ETH transfer events, and lower execution costs — but must invest in upgrading gas estimation logic and infrastructure tooling.
Glamsterdam is a structural overhaul, not a feature release. It rewires how Ethereum builds blocks, processes transactions, and meters computational costs. The two headline changes — ePBS and BALs — address problems that have worsened measurably since the Merge: builder centralization has tripled, and sequential execution constrains throughput at a time when competing L1s and rollups have absorbed much of Ethereum's growth.
The November 4 mainnet target is aggressive given the scope of changes and Ethereum's history of fork delays. The Fusaka precedent — where a client bug disrupted 248 blocks — illustrates the operational risk of shipping consensus-layer and execution-layer changes simultaneously. The next eight weeks of testnet operation will determine whether the schedule holds.
What is not in question is the direction. Ethereum's core developers have committed to protocol-level solutions for problems previously managed by middleware and market structure. Whether the implementation timeline matches the ambition remains an open question.