Ethereum's next hard fork, Glamsterdam, is behind schedule. The Ethereum Foundation's April 10 development brief — Checkpoint #9 — confirmed that enshrined Proposer-Builder Separation (ePBS), the upgrade's consensus-layer headliner, is proving more complex than anticipated. Devnet-4 is complete; ...
"ePBS splits block production into two parties acting in sequence inside consensus… every part of the stack has to reason about partial blocks and two-party coordination, a change that touches practically everything." — Ethereum Foundation, Checkpoint #9 Development Brief, April 10, 2026
Ethereum's next hard fork, Glamsterdam, is behind schedule. The Ethereum Foundation's April 10 development brief — Checkpoint #9 — confirmed that enshrined Proposer-Builder Separation (ePBS), the upgrade's consensus-layer headliner, is proving more complex than anticipated. Devnet-4 is complete; Devnet-5 is underway. The original H1 2026 target now looks aspirational, with realistic estimates shifting to Q3.
The stakes are structural. Two block builders — Titan (52% market share) and beaverbuild — currently produce the majority of Ethereum's MEV-Boost blocks, an arrangement that exists entirely outside the protocol. Glamsterdam's EIP-7732 would move this separation into the consensus layer itself, reducing reliance on external relays that seven operators control 99% of today. On the execution side, EIP-7928 introduces Block-Level Access Lists (BALs) enabling parallel transaction processing, targeting a 10x throughput increase to approximately 10,000 TPS and a roughly 79% reduction in gas fees. ETH trades at approximately $2,200 with a $265 billion market cap as the network processes gas at 0.05 Gwei — down from 1.67 Gwei a year ago.
Ethereum's block production pipeline has a centralization problem that exists outside the protocol. Since the Merge in September 2022, block building has operated through MEV-Boost, an out-of-protocol sidecar developed by Flashbots. The system works: validators outsource block construction to specialized builders who compete to assemble the most profitable blocks. But the market structure it created is concentrated.
According to relayscan.io data, Titan holds 52.16% of blocks among top builders as of April 2026. Between October 2023 and March 2024, three builders — beaverbuild, rsync, and Titan — produced approximately 80% of all MEV-Boost blocks. Titan's rise from less than 1% to over 40% market share was driven largely by an exclusive order flow deal with the Banana Gun Telegram bot starting February 2024.
On the relay side, seven unique operators control 99% of the MEV-Boost network, per Flashbots data. ESMA has explicitly flagged centralization at relay and builder levels as a structural risk. Flashbots responded by launching BuilderNet in December 2024, migrating all its builders to a TEE-based multi-operator system. But BuilderNet addresses the builder layer — it does not change Ethereum's consensus rules.
Glamsterdam addresses both the builder and execution layers at the protocol level.
Vitalik Buterin outlined eight Ethereum Improvement Proposals defining Glamsterdam's scope in late February 2026. Two anchor the release:
Supporting EIPs include:
The scope is the most aggressive structural overhaul since the Merge, according to multiple developer commentary sources. Over 25 EIPs were initially proposed, though the final set was narrowed to eight to manage complexity.
EIP-7732 is the consensus-layer headliner. It separates the Ethereum block into consensus and execution components, adding an in-protocol mechanism for the consensus proposer to select an execution proposer (block builder).
Under the current MEV-Boost system, the process works as follows: builders assemble blocks off-chain, submit them to relays, and relays forward them to proposers who select the highest-bid block. The proposer cannot see block contents — a trust assumption enforced by the relay, not the protocol.
EIP-7732 moves this trust assumption into the consensus layer. Block builders assemble and cryptographically seal blocks. Proposers select the highest-paying block without seeing contents. Transactions are revealed only after block finalization. The relay middleman is eliminated.
The implications for MEV distribution are direct. With relays removed from the trust chain, the builder-proposer relationship becomes protocol-enforced. This reduces the attack surface — no relay operator can selectively censor transactions or front-run builders — but it concentrates the competitive dynamics on builders themselves. Whether ePBS reduces or merely reshuffles centralization pressure remains an open question among researchers.
According to an April 2026 analysis published by BlockEden, ePBS could reduce MEV extraction costs by eliminating relay fees and latency arbitrage. However, it may shift centralization pressure from a handful of relay operators to a smaller set of competitive builders with proprietary order flow.
EIP-7928 is the execution-layer headliner. It introduces Block-Level Access Lists (BALs) — enforced, block-wide records of every account and storage location accessed during block execution, along with post-execution values.
The technical rationale: Ethereum currently executes transactions sequentially. Each transaction must wait for the previous one to complete before state can be updated. This is inherently serial and limits throughput.
BALs change this by enabling:
According to analysis on ethresear.ch, 60-80% of Ethereum transactions access disjoint storage slots, enabling effective parallelization without conflict. The remaining 20-40% benefit from post-transaction state diffs that allow dependent transactions to be resolved efficiently.
Historical data shows an average BAL size of approximately 70 KiB per block. A new block header field, block_access_list_hash, containing the Keccak-256 hash of the RLP-encoded access list, anchors the system.
Combined with the gas limit increase to 200 million, the parallel execution primitive targets approximately 10,000 TPS on Layer 1 — compared to roughly 15-30 TPS today. Developer documentation from BlockEden projects a 78.6% reduction in gas fees across both simple transfers and complex smart contract interactions.
The current gas environment is already historically cheap. Average gas prices sit at 0.052 Gwei as of April 2026, down from 1.67 Gwei in April 2025, according to CoinLaw data. The 60-million gas limit (2x year-ago levels) and expanded blob capacity (14 blobs per block, up from 6) have already pushed Ethereum L1 into low-fee territory. Glamsterdam would extend this further.
The Ethereum Foundation's Checkpoint #9, published April 10, 2026, provides the most current status:
The primary bottleneck is ePBS. According to the Foundation's brief, the implementation "splits block production into two parties acting in sequence inside consensus," meaning the protocol must handle disagreement or failure between proposer and builder. Every client must reason about "partial blocks" and two-party coordination — a change described as touching "practically everything."
Non-headliner features add scope risk. Gas repricing (EIP-7904) has its own complexities. The Base engineering team has warned about adding FOCIL (Fork-Choice enforced Inclusion Lists, EIP-7805) alongside ePBS, leading to FOCIL being deferred to the subsequent Hegota upgrade.
The remaining path to mainnet: stable multi-feature devnet → client releases → security reviews → testnet deployment → testnet stability confirmation → mainnet fork date announcement. The original H1 2026 target has a tentative June window, but developers have been explicit that shipping correctly takes priority over hitting a date. Q3 2026 remains a realistic possibility.
As of April 14, only three of the eight proposed EIPs have been tested on Devnet-4.
Glamsterdam's economic impact flows through three channels:
1. Fee reduction. The 79% projected gas fee cut, on top of already-low 0.05 Gwei fees, would make L1 transactions near-zero cost. This challenges the L1 fee revenue model. Ethereum L1 hit 2.2 million daily transactions on December 30, 2025, with average fees at $0.17 — already down from $2.15 six months prior. Further compression raises questions about validator economics, though the gas limit increase is designed to compensate via higher transaction volume.
2. MEV redistribution. By enshrining PBS, the protocol codifies the builder market rather than relying on an external relay ecosystem. This formalizes MEV as protocol-level infrastructure, potentially enabling future governance over MEV distribution. Currently, MEV flows to builders, relays, and searchers in proportions that are opaque and market-driven.
3. L1 vs. L2 competition. Layer 2 networks now account for approximately 95% of Ethereum's total transaction throughput. If Glamsterdam delivers 10,000 TPS on L1 at near-zero fees, the economic rationale for L2 execution shifts. L2s would retain advantages in application-specific customization, governance, and sequencer revenue — but the cost argument weakens substantially. L2 data costs, already reduced by blob expansion, would face further pressure as L1 becomes viable for a broader range of transactions.
The Ethereum Foundation opened non-headlining EIP proposals for the subsequent Hegota upgrade on April 9, 2026. Hegota's confirmed consensus-layer headliner is FOCIL (EIP-7805), Fork-Choice enforced Inclusion Lists.
FOCIL addresses censorship resistance — a different dimension of the same block production problem. While ePBS restructures who builds blocks, FOCIL ensures that certain transactions cannot be excluded from blocks by builders or proposers. The two features were originally considered for the same fork but were separated after complexity concerns.
The Glamsterdam → Hegota sequence represents a deliberate two-phase approach to restructuring Ethereum's block production: first formalize the market (ePBS), then enforce inclusion guarantees (FOCIL).
Glamsterdam is the largest structural change to Ethereum since the Merge. It does not add a new feature so much as formalize and optimize existing infrastructure: block building moves from an external relay system into the protocol; transaction execution moves from sequential to parallel. Both changes address real, measurable problems — builder centralization and throughput limits — with specific technical mechanisms.
The delay is telling. ePBS touches every layer of the client stack, and the Ethereum Foundation has chosen correctness over speed. Three of eight EIPs are on devnet; five remain. The June target is aspirational; Q3 is probable.
For the broader Ethereum economy, Glamsterdam's implications are structural. Near-zero L1 fees and 10,000 TPS would redefine the L1-L2 relationship, formalize MEV as protocol infrastructure, and compress validator fee revenue further. Whether this strengthens or weakens Ethereum's economic model depends on whether volume growth compensates for per-transaction fee decline — a question the data cannot yet answer.