Ethereum's next hard fork, codenamed Glamsterdam, is entering its critical testing phase with a mainnet target of mid-2026. The upgrade introduces three structural changes to the protocol: enshrined Proposer-Builder Separation (ePBS) via EIP-7732, Block-level Access Lists (BALs) via EIP-7928, and...
"In Glamsterdam, Ethereum is getting ePBS, which lets proposers outsource to a free permissionless market of block builders." — Vitalik Buterin, Ethereum Co-Founder
Ethereum's next hard fork, codenamed Glamsterdam, is entering its critical testing phase with a mainnet target of mid-2026. The upgrade introduces three structural changes to the protocol: enshrined Proposer-Builder Separation (ePBS) via EIP-7732, Block-level Access Lists (BALs) via EIP-7928, and a comprehensive gas repricing package anchored by EIP-7904. Together, these changes aim to decouple block building from consensus validation, enable parallel transaction execution, and realign gas costs with modern hardware benchmarks.
The stakes are significant. Two block builders currently produce approximately 80% of Ethereum blocks. Five builders account for 99% of block production. Glamsterdam's ePBS would move the builder-proposer marketplace from an off-chain relay dependency (MEV-Boost) into the protocol itself, theoretically preventing builder concentration from contaminating validator neutrality. Simultaneously, the gas repricing package targets a 78.6% reduction in transaction costs, while BALs lay the groundwork for the gas limit to scale from the current 60 million toward 200 million.
Devnets are live. Epbs-devnet-0 launched on March 4, 2026, with Lodestar and Lighthouse achieving interoperability. The scope freeze for Q1 2026 is imminent. What follows will determine whether Ethereum can structurally decentralize its most concentrated layer before centralization becomes permanent.
Ethereum's Fusaka upgrade shipped on December 3, 2025, delivering PeerDAS (peer data availability sampling) and blob parameter adjustments that expanded rollup data throughput. The blob target can now double every few weeks up to a maximum of 48. The block gas limit increased from 30 million to 60 million between the Pectra (May 2025) and Fusaka upgrades.
Glamsterdam represents a different class of change. Where Fusaka optimized data availability for Layer 2s, Glamsterdam targets the structural mechanics of Layer 1 itself: who builds blocks, how transactions execute, and what gas costs actually reflect. Ethereum developers have described it as the most significant protocol-level restructuring since The Merge in September 2022.
The Ethereum Foundation's January 2026 checkpoint outlined three 2026 priorities: scaling the gas ceiling "toward and beyond" 100 million, enshrining builder-proposer separation, and beginning the transition toward real-time ZK verification. Glamsterdam addresses the first two directly.
The centralization of block building is Ethereum's most acute structural risk. Under the current MEV-Boost system, approximately 92% of Ethereum blocks are built through the MEV-Boost relay. Of those, two builders — Beaverbuild and Titan — control roughly 80% of block production. Beaverbuild alone holds approximately 50% market share. Sixteen active builders exist, but five produce 99% of blocks.
This concentration exists because block building is an economies-of-scale business. Builders with exclusive orderflow agreements, superior MEV extraction algorithms, and lower latency connections win more auctions. The current system relies on Flashbots' MEV-Boost relay as a trusted intermediary — a piece of off-chain infrastructure that the protocol does not control.
EIP-7732 moves this marketplace on-chain. Under enshrined PBS, the proposer and builder roles are formally separated at the protocol level:
The design eliminates the need for external relays. Block-building logic becomes part of Ethereum's consensus rules. A proposer cannot censor transactions it cannot see. A builder cannot leverage its role to influence validation.
However, ePBS introduces its own risks. Academic research on the "free option problem" estimates that approximately 0.82% of blocks would exercise timing options under normal conditions, rising to 6% during high-volatility periods. This occurs because builders can delay submitting blocks to capture last-moment MEV, creating an economic option at the proposer's expense.
Flashbots responded preemptively. In December 2024, the organization migrated all builders to BuilderNet, a decentralized block-building network using Trusted Execution Environments (TEEs), jointly operated with Beaverbuild and Nethermind. BuilderNet neutralizes exclusive orderflow deals and distributes MEV to users. Whether this off-chain solution coexists with or is replaced by ePBS remains an open design question.
Ethereum's current execution model is sequential. Every transaction could theoretically depend on any other transaction in the same block, forcing nodes to process them one at a time. This is the primary bottleneck preventing higher throughput at the base layer.
EIP-7928 introduces Block-level Access Lists (BALs), which require blocks to declare upfront which accounts and smart contract storage slots they will access. This pre-declaration enables:
The overhead is measurable. Compressed BALs average 70-72 KiB per block. This is a modest bandwidth cost for a capability that unlocks multi-core execution on commodity hardware.
BALs are the prerequisite for scaling the gas limit beyond current levels. The current coordination target is 60 million gas per block, producing approximately 238 simple transactions per second or 42 complex (120,000 gas) transactions per second. With BALs enabling parallelism:
| Scenario | Gas Limit | Simple Tx/sec | Complex Tx/sec | |----------|-----------|---------------|----------------| | Current | 60M | ~238 | ~42 | | 2× (near-term target) | 120M | ~476 | ~83 | | Glamsterdam target | 200M | ~793 | ~139 |
The Ethereum Foundation has stated its intention to push the gas ceiling toward and beyond 100 million in 2026. External analyses project Glamsterdam could enable up to 200 million, representing a 3.3× increase from current levels. Base, Coinbase's Layer 2, has publicly endorsed BALs as critical infrastructure for L2 performance improvements.
Ethereum's gas cost schedule is outdated. Many opcode prices were set years ago and no longer reflect actual computational costs on modern hardware. Some operations are overpriced by orders of magnitude; others are underpriced, creating exploitable imbalances.
EIP-7904 anchors the repricing effort by benchmarking every opcode against actual execution costs on reference hardware and realigning gas charges accordingly. The overall package is tracked under EIP-8007, which groups 10 repricing proposals:
Compute and Memory:
State and Storage:
Data and Calldata:
The net effect is a 78.6% reduction in gas costs for both simple ETH transfers and complex smart contract interactions. A Uniswap trade that currently costs $3-8 in gas is projected to fall below $1. However, the repricing is not uniformly cheaper. State creation costs rise tenfold under EIP-8037, deliberately increasing the cost of growing Ethereum's state to prevent bloat.
This is a deliberate economic design: make computation cheaper, make storage more expensive. The incentive shifts favor stateless or state-minimal contract architectures and penalize contracts that create excessive permanent storage.
As of mid-March 2026, Glamsterdam development is in active devnet testing:
The scope freeze for Glamsterdam was set for late Q1 2026. Features not achieving interoperability by end of February faced deferral to the subsequent Hegota fork (targeted for late 2026). Early indications suggest both ePBS and BALs cleared this threshold.
Projected timeline:
The Hegota fork, expected in late 2026, will complement Glamsterdam with FOCIL (Forced Inclusion Lists) — a censorship resistance mechanism where 16 randomly selected attesters nominate transactions that must be included, with block rejection for non-compliance. Vitalik Buterin described the combination of ePBS and FOCIL on February 19, 2026, as providing "guaranteed rapid inclusion" for smart accounts and privacy protocols.
ePBS centralization paradox. Enshrining PBS may formalize rather than fix builder concentration. If the protocol-level marketplace exhibits the same winner-take-all dynamics as MEV-Boost, concentration simply moves from an off-chain relay to an on-chain mechanism — with the added difficulty that protocol-level changes require hard forks to adjust.
Free option problem. The 0.82% average (6% during volatile markets) block option exercise rate represents a direct economic cost to proposers. Mitigations exist (timing constraints, penalties), but the optimal parameters remain under active research.
Gas repricing disruption. The 10-EIP repricing package will alter the economics of every deployed smart contract. Contracts optimized for current gas schedules may become more or less expensive to operate. A tenfold increase in state creation costs will particularly impact protocols with heavy storage patterns.
ZK verification dependency. Higher gas limits beyond 200 million depend on the eventual adoption of real-time ZK proving for state verification. The Ethereum Foundation's requirements — 128-bit security, proof sizes under 300 KiB, no recursive wrappers with trusted setups — have not yet been met by any production proving system. This creates an implicit ceiling on how far Glamsterdam's throughput gains can scale.
Execution risk. Glamsterdam is the most complex Ethereum upgrade since The Merge. Three major protocol changes shipping simultaneously increases the interaction surface for bugs. The two 30-day audit windows are a mitigation, not a guarantee.
Glamsterdam is Ethereum's attempt to solve a problem it created. The Merge shifted Ethereum to proof of stake. The subsequent PBS marketplace emerged to handle MEV extraction efficiently — and promptly centralized into a two-builder duopoly. Fusaka scaled data availability for rollups. Now Glamsterdam must restructure the base layer itself: separate building from proposing at the protocol level, enable parallel execution, and reprice gas to reflect 2026 hardware realities.
The economic logic is straightforward. Cheaper computation incentivizes on-chain activity. Expensive storage prevents state bloat. Enshrined PBS prevents builder concentration from becoming validator concentration. Each component addresses a specific failure mode in the current architecture.
Whether all three changes can ship simultaneously without introducing new failure modes is the central execution question. The devnets are running. The scope freeze is approaching. Mid-2026 will determine whether Ethereum's base layer keeps pace with the rollup ecosystem it has spent two years optimizing for.