Ethereum's Glamsterdam hard fork, targeted for H1 2026, represents the network's most significant Layer 1 execution overhaul since The Merge. The upgrade bundles 12 Ethereum Improvement Proposals anchored by two structural changes: EIP-7732 (Enshrined Proposer-Builder Separation, or ePBS) and EIP...
"2026 is the year that we take back lost ground in terms of self-sovereignty and trustlessness." — Vitalik Buterin, Co-Founder, Ethereum
Ethereum's Glamsterdam hard fork, targeted for H1 2026, represents the network's most significant Layer 1 execution overhaul since The Merge. The upgrade bundles 12 Ethereum Improvement Proposals anchored by two structural changes: EIP-7732 (Enshrined Proposer-Builder Separation, or ePBS) and EIP-7928 (Block-Level Access Lists). Together, they enable a phased gas limit increase from 60 million to 200 million per block — a 3.3x capacity expansion — and set Ethereum on a path toward 10,000 transactions per second on Layer 1.
The timing is deliberate. As of April 2026, the top three block builders — Titan, BuilderNet, and Quasar — control 86.8% of MEV share across Ethereum blocks. Over 80-90% of block production depends on off-chain relays. Private transaction routing exceeds 50% of high-value activity. Glamsterdam's ePBS moves this trust-dependent infrastructure on-chain, replacing relay intermediaries with cryptographic commitments and protocol-enforced payments. Whether this reduces or entrenches builder concentration remains an open empirical question.
Devnet-4 testing is complete. Devnet-5 is underway. Spring 2026 testnet phases with dual audits are planned before a tentative June 2026 mainnet activation. Ethereum's track record on timeline adherence — Pectra shipped May 2025, Fusaka activated December 2025 — provides some basis for confidence, though scope creep risk is non-trivial given the 25+ non-headliner EIPs under consideration.
The core throughput improvement centers on a phased gas limit increase. Tomasz Stańczak, former Ethereum Foundation co-executive director, told the Bankless Summit that the limit would rise to 100 million per block in the first phase and double to 200 million once ePBS is fully operational. The current gas limit stands at approximately 60 million.
A 200 million gas limit translates to a theoretical throughput ceiling of roughly 10,000 TPS on Layer 1 — up from the current 15-20 TPS. This figure represents a long-term target, not an immediate outcome. The phased approach reflects engineering pragmatism: the initial 100M ceiling can be supported by existing client infrastructure, while the 200M target requires ePBS to extend the propagation window from approximately 2 seconds to 9 seconds, giving nodes more time to validate larger blocks.
Gas costs would fall correspondingly. Phemex research projects a 78.6% decrease in gas fees across standard transfers and smart contract calls. EIP-2780, included in the upgrade, specifically targets simple ETH transfers with fee reductions of up to 71%.
For context, Ethereum's gas limit has historically increased through community-driven validator signaling rather than hard-coded protocol changes. Glamsterdam's approach — tying gas limit increases to structural protocol changes — represents a shift toward more deliberate capacity planning.
EIP-7732 enshrines Proposer-Builder Separation into the protocol, replacing the current MEV-Boost relay system that has operated as critical but unprotected middleware since September 2022.
Under the current system, proposers (validators) outsource block construction to specialized builders via off-chain relays. The architecture works but introduces three structural problems:
ePBS introduces a commit-reveal pipeline: builders seal blocks and submit cryptographic commitments with bids. Validators select the highest bid without seeing block contents. A new Payload Timeliness Committee (PTC) verifies delivery. Builders who fail to deliver or reveal late face protocol-enforced penalties.
The architecture formalizes builders as staked protocol participants with on-chain identity. This eliminates the relay layer entirely but does not, by itself, reduce builder concentration — a distinction worth emphasizing.
EIP-7928 introduces Block-Level Access Lists (BALs), which declare every account and storage slot a block will access before execution begins. This seemingly bureaucratic addition unlocks a fundamental capability: parallel transaction processing.
Currently, Ethereum executes transactions sequentially because any transaction could potentially modify state that subsequent transactions depend on. BALs make these dependencies explicit. Validators can identify non-conflicting transactions and execute them simultaneously, enabling what developers project as 10-30x improvements in execution throughput.
The mechanism works as follows: each block header includes a new block_access_list_root hash field. Execution clients can pre-fetch the entire working state before processing begins. EIP-8159, a companion networking proposal, enables nodes to share access lists across the peer-to-peer protocol via the eth/71 wire extension.
BALs also enable "executionless syncing" — new nodes can verify block validity without re-executing every transaction, reducing the time required to join the network. EL clients must store a minimum of 3,533 epochs of BAL data.
The practical impact for decentralized application developers is minimal in the short term. BALs are infrastructure-level changes that primarily affect client teams and block producers. However, they lay the foundation for future parallelization strategies and are a prerequisite for the ZK-verified execution framework targeting 1,000x improvements on Ethereum's longer-term roadmap.
Three EIPs address the accumulation of technical debt in Ethereum's gas pricing:
EIP-7904 (General Repricing) recalibrates gas costs for EVM opcodes using modern hardware benchmarks. Many current gas prices were set years ago and no longer reflect actual computational costs. The repricing aligns fees with resource consumption, removing subsidies for operations that are cheaper in practice and increasing costs for operations that are underpriced.
EIP-8037 (State Creation Gas Cost Increase) adjusts fees proportional to permanent data size, targeting a state database growth rate of approximately 100 GiB per year. This addresses a long-standing concern: every piece of state written to Ethereum persists indefinitely, and the current gas cost for state creation does not adequately reflect this permanence.
EIP-8038 (State-Access Gas Cost Increase) raises gas costs for database lookups. This change is defensive — it eliminates a class of denial-of-service attacks that exploit underpriced state access operations.
Additional proposals round out the upgrade's scope. EIP-7954 increases the maximum contract size from approximately 24 KiB to 32 KiB. EIP-8045 prevents slashed validators from proposing future blocks. EIP-8080 allows standard exits to access the consolidation queue, increasing exit speeds by up to 2.5x. EIP-7708 mandates log emissions for all ETH transfers and burns, improving on-chain auditability.
The centralization Glamsterdam aims to address is measurable. As of April 2026, according to RelaysScan data:
| Builder | Block Share | Notes | |---------|------------|-------| | Titan (titanbuilder.xyz) | 52.16% | 72.99 ETH profit in 24 hours | | BuilderNet | 24.63% | Flashbots-backed decentralized builder | | Quasar | 15.06% | — | | All others | ~8.15% | — |
The Herfindahl-Hirschman Index for the builder market stands at 3,186 — well above the 2,500 threshold that regulators in traditional markets consider "highly concentrated." The top three builders control 85.9% of all blocks and 86.8% of MEV share.
On the relay side, ultrasound.money leads with 33.92% of payload share, followed by Titan Relay at 24.19%, bloXroute at 14.67%, and Aestus at 10.03%.
Research published in January 2026 found that exclusive order flows and non-atomic MEV accounted for 71% and 23% of trading-related builder revenue, respectively. Exclusive order flow access — not superior algorithms — is the primary driver of builder dominance.
ePBS removes the relay layer but does not directly address exclusive order flow arrangements. Academic modeling from January 2026 suggests ePBS may "significantly amplify profit and content centralisation" among builders by lowering the cost of sophisticated block construction strategies. The net effect on decentralization remains contested.
BuilderNet, backed by Flashbots with BeaverBuild and Nethermind participation, produced 25.5% of blocks by January 2026 as a partial decentralization countermeasure. Post-Glamsterdam inclusion lists (FOCIL) are planned for a subsequent upgrade to enforce transaction inclusion requirements.
Glamsterdam positions Ethereum's L1 against a competitive field that has moved aggressively on execution speed:
| Metric | Ethereum (Post-Glamsterdam) | Solana | Monad | |--------|---------------------------|--------|-------| | Target TPS | ~10,000 | 1,000-1,500 | 10,000+ | | EVM Compatibility | Native | None | Full | | Block Time | ~12 seconds | ~400ms | ~400ms | | Parallelism Approach | Pre-declared access lists | Native Sealevel scheduler | Optimistic parallel execution | | Validator Set | 500,000+ | ~1,500 | Emerging |
The comparison reveals trade-offs rather than a clear winner. Ethereum's validator set is orders of magnitude larger than competitors, providing superior decentralization but at the cost of slower block times. Solana's native parallelism produces lower latency but with a smaller validator set. Monad matches Ethereum's throughput target with full EVM compatibility but lacks an established validator network.
The combined L2 ecosystem currently processes approximately 5,600 TPS, with projections of 24,000+ TPS as the blob parameter optimization schedule progresses through 2026. For Ethereum's total throughput (L1 + L2), Glamsterdam's 10,000 TPS L1 target would bring aggregate capacity above 30,000 TPS.
Scope creep. Beyond the headline EIPs, 25+ additional proposals remain under consideration. Vitalik Buterin outlined eight EIPs defining the upgrade's full scope in late February 2026, but the final list is not frozen. Every additional EIP increases testing surface and delay risk.
Timeline slippage. Ethereum hard forks have historically shipped later than initial targets. While Pectra and Fusaka both landed approximately on schedule, the development cadence — three hard forks in 13 months — is unprecedented and may strain client teams.
MEV infrastructure disruption. Existing relay operators and builders have built businesses around the current MEV-Boost architecture. ePBS eliminates their structural position. How these participants adapt — and whether new centralization vectors emerge — is uncertain.
Application breakage. Hardcoded gas estimates in deployed contracts and front-end applications may break when gas costs are repriced. EIP-7904's recalibration, while necessary, introduces non-trivial migration work for application developers.
Academic concerns. January 2026 modeling suggests ePBS could amplify rather than reduce builder centralization. The protocol change removes relays but may lower barriers for sophisticated builders to consolidate further. FOCIL inclusion lists, the proposed countermeasure, are not part of Glamsterdam and remain on a future upgrade timeline.
Glamsterdam is an infrastructure upgrade, not a user-facing feature launch. Its effects will manifest as cheaper transactions, faster block validation, and a restructured block production market. Whether it reduces or merely reshuffles centralization in block building depends on dynamics that the protocol change alone cannot determine — exclusive order flow arrangements, builder business models, and the eventual deployment of FOCIL inclusion lists.
The phased gas limit increase reflects a lesson Ethereum's developers have internalized: capacity expansion works when it tracks infrastructure readiness, not aspirational targets. The 100M initial ceiling, expandable to 200M once ePBS stabilizes, is a conservative engineering choice for a network securing over $300 billion in value.
For the broader Ethereum ecosystem, Glamsterdam marks the beginning of L1 reclamation. After two years focused on L2 scaling — blob data, PeerDAS, BPO schedules — the protocol's attention returns to the base layer. The upgrade does not solve Ethereum's execution speed deficit relative to monolithic competitors like Solana. It narrows it, while preserving the decentralization properties that 500,000+ validators provide.
The next upgrade, Hegota, is planned for H2 2026. Its focus — Verkle Trees and a projected 90% reduction in node storage requirements — addresses the complementary challenge: making Ethereum's expanding state manageable for the validator set that Glamsterdam depends on.