Ethereum's Glamsterdam hard fork — a combined consensus-layer (Glam) and execution-layer (Amsterdam) upgrade — entered public testnet deployment in August 2026 after completing its final devnet stage in June. Sepolia forked on August 3; Hoodi is scheduled for August 17. If multi-client stability ...
"Almost every major piece of Ethereum will be replaced over the next three to four years." — Vitalik Buterin, Ethereum Co-Founder, July 2026
Ethereum's Glamsterdam hard fork — a combined consensus-layer (Glam) and execution-layer (Amsterdam) upgrade — entered public testnet deployment in August 2026 after completing its final devnet stage in June. Sepolia forked on August 3; Hoodi is scheduled for August 17. If multi-client stability holds across both networks, mainnet activation is targeted for Q4 2026, though the Ethereum Foundation has not committed to a fixed date.
The upgrade bundles ten EIPs, headlined by EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists). Together, these proposals move block construction into the protocol layer and enable parallel transaction execution — two structural changes that Buterin has described as the start of Ethereum's "third major iteration," following the original launch and the Merge. The target is a 200 million gas limit, more than triple the current 60 million, with projected throughput of approximately 10,000 transactions per second.
The stakes are material. Over 90% of Ethereum blocks currently flow through MEV-Boost relays, and as of mid-2026, two builders — Titan and Quasar — construct roughly 73% of all blocks. Glamsterdam's ePBS is designed to eliminate the trusted-relay dependency that underpins this concentration. Whether it succeeds will determine the network's censorship-resistance profile for the next cycle.
Ethereum L1 gas fees have fallen to $0.10–$0.25 per transfer as of mid-2026, down from $1.85 in mid-2025. Base fees sit at approximately 0.05 Gwei, a historic low. The network processed a record 2.89 million transactions in a single day in February 2026, and Q1 2026 volume hit 200.4 million transactions — a 43% quarter-over-quarter increase. Despite cheaper fees, the 60 million gas limit constrains throughput during demand spikes.
Layer 2 networks now process over 2 million daily transactions, exceeding mainnet activity. Fusaka, shipped in December 2025, scaled L2 data availability through PeerDAS. Glamsterdam shifts focus to L1 execution capacity — the layer where value settles, validators earn fees, and block production economics are determined.
The block-building market's concentration underscores the urgency. The Herfindahl-Hirschman Index for Ethereum's builder market stands at approximately 3,892 — well above the 1,800 threshold indicating high concentration. By mid-2026, Titan and Quasar construct roughly 73% of all blocks. Including Eureka and BuilderNet pushes the figure above 95%. This duopoly-to-oligopoly structure exists because MEV-Boost operates outside the protocol, creating trust assumptions and chokepoints that the original Ethereum design did not anticipate.
EIP-7732 is the consensus-layer headline. It moves proposer-builder separation from the off-protocol MEV-Boost stack into Ethereum's consensus rules.
How it works: The block production process splits into two on-chain roles. The proposer selects a builder's bid and publishes a signed commitment. The builder then reveals full block contents. The network's Payload Timeliness Committee validates that the builder's payload matches the commitment. This handshake happens at the protocol level, eliminating the need for trusted relays.
What changes structurally:
What it does not solve: ePBS does not eliminate MEV. It does not prevent builders from accessing exclusive order flow. It does not guarantee builder-market deconcentration. What it does is remove the trusted relay as a single point of failure and censorship vector. Builders still compete for block construction rights, but they do so within a protocol-enforced auction rather than through an off-chain marketplace.
According to research published on arXiv in June 2026 by Luca Zanolini and collaborators, ePBS has undergone formal security analysis proving its properties under the assumed threat model. The analysis confirms that the mechanism prevents proposers from stealing builder value and prevents builders from reneging on committed payloads — two attack vectors possible under MEV-Boost's trust model.
EIP-7928 is the execution-layer headline. It introduces Block-Level Access Lists (BALs) — enforced, block-wide declarations of every account, storage slot, balance, nonce, and code change that a block's transactions will touch.
Technical mechanism: A new field, block_access_list_hash, is added to the block header, containing the Keccak-256 hash of the RLP-encoded access list. Before execution begins, the network knows exactly which state each transaction will read or write.
What this enables:
Overhead: Average BAL size is approximately 35 KiB per block at 36 million gas, with worst-case sizes remaining below calldata worst cases.
Toni Wahrstätter, the EIP's lead author and Ethereum Foundation researcher, noted that the proposal "received much love from core devs recently." The first devnet for EIP-7928 went live in late 2025, and the proposal was confirmed as the Glamsterdam execution-layer headliner during the ACDC-178 All Core Devs call in May 2026.
BALs are the prerequisite for the gas limit increase. Without parallel execution, raising the gas limit to 200 million would overwhelm single-threaded node operators. With BALs, the network can absorb higher throughput without proportionally increasing hardware requirements — at least in theory. Actual gains depend on client implementation quality and real-world state access patterns.
Beyond the two headliners, Glamsterdam bundles eight additional EIPs. Three carry significant economic weight:
EIP-7904 (Gas Cost Repricing): Realigns gas costs with actual computational resources. Many current prices were set years ago and no longer reflect execution costs on modern hardware. The repricing targets 60 million gas per second — a 3x increase from the current 20 Mgas/s performance baseline. The result: an estimated 78.6% fee reduction for both simple ETH transfers and complex smart contract interactions.
EIP-8037 (Cost Per State Byte): Introduces a per-byte pricing model for state creation, replacing the current flat-cost approach. At a 200 million gas limit, unchecked state growth would reach approximately 380 GiB per year — a rate the developers consider unsustainable. The sustainable target is 120 GiB per year.
EIP-7954 (Contract Size Limit): Increases maximum contract bytecode from approximately 24 KiB to 32 KiB. A targeted change, but one that unblocks deployment of more complex protocols without proxy-pattern workarounds.
Other bundled proposals include EIP-7843 (SLOTNUM opcode), EIP-7708 (standardized logs for ETH transfers and burns), EIP-2780 (up to 71% reduction in ETH transfer gas costs), and EIP-7997 (deterministic factory predeployment).
| Milestone | Date | Status | |-----------|------|--------| | Final devnet lock-in | June 2026 | Complete | | Sepolia testnet fork | August 3, 2026 | Active | | Hoodi testnet fork | August 17, 2026 | Scheduled | | Mainnet activation | Q4 2026 (target) | Pending |
Past Ethereum forks have required two to four months of public testnet validation before mainnet deployment. On that cadence, mainnet would land between October and December 2026.
Risk factors:
The fee reduction math is straightforward. At current 60 million gas and approximately 0.05 Gwei base fees, Ethereum L1 generates minimal fee revenue — a structural problem for a network whose security budget depends on transaction fees as issuance rewards decline. The 78.6% fee reduction from EIP-7904 would further compress per-transaction revenue, but the 3.3x capacity increase at 200 million gas is expected to generate higher aggregate fee volume through transaction count growth.
For validators, ePBS changes the MEV distribution. Currently, validators outsource block building to MEV-Boost builders and capture value through relay-mediated auctions. Under ePBS, the auction mechanism is enshrined, potentially reducing the builder's ability to extract rents from the proposer relationship. Whether this increases or decreases validator revenue depends on how competitive the enshrined auction proves to be relative to MEV-Boost.
For Layer 2 networks, Glamsterdam's L1 capacity increase raises a competitive question. If L1 transactions become cheaper and faster, some activity currently routed to L2s for cost reasons may migrate back. This could compress L2 sequencer revenue while increasing L1 fee capture — a rebalancing of economic value across the Ethereum stack.
For application developers, the gas repricing and parallel execution support lower the cost of deploying and operating on-chain logic. The contract size limit increase removes a constraint that has pushed developers toward proxy patterns and modular architectures for technical rather than design reasons.
Glamsterdam is Ethereum's largest single upgrade since the Merge. It addresses two systemic risks — block-builder centralization and single-threaded execution limits — through protocol-level changes rather than off-chain patches. The 200 million gas target represents a 3.3x capacity increase, and ePBS eliminates trusted intermediaries from the block production pipeline.
Whether the upgrade delivers on its targets depends on factors not yet resolved: client stability across public testnets, real-world parallel execution gains versus theoretical projections, and builder behavior under an enshrined auction mechanism. State growth at full capacity remains a concern without future state expiry mechanisms.
The market has not priced in the upgrade significantly. ETH traded at approximately $1,881 on August 12, 2026, and Bitcoin ETF flow reversals dominated weekly sentiment. Glamsterdam's impact will be measured not in immediate price action but in whether Ethereum L1 can recapture transaction volume from Layer 2 networks — and whether the enshrined PBS mechanism produces a more competitive, less concentrated block-building market than the relay system it replaces.