Ethereum's next hard fork, Glamsterdam, entered its first unified testing phase on April 24, 2026, when core developers launched glamsterdam-devnet-0 — the first environment where all upgrade components run together. The fork bundles three structural changes: Enshrined Proposer-Builder Separation...
"2026 is the year that we take back lost ground in terms of self-sovereignty and trustlessness." — Vitalik Buterin, Co-founder, Ethereum
Ethereum's next hard fork, Glamsterdam, entered its first unified testing phase on April 24, 2026, when core developers launched glamsterdam-devnet-0 — the first environment where all upgrade components run together. The fork bundles three structural changes: 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. If it ships on schedule, the upgrade would raise the block gas limit from 60 million to 200 million, target approximately 10,000 transactions per second on Layer 1, and reduce gas costs by an estimated 78.6%.
June 2026 remains the aspirational mainnet date. According to the Ethereum Foundation's Checkpoint #9 blog post dated April 10, developers acknowledged that ePBS implementation "is proving to be trickier than anticipated." Q3 2026 is the more realistic window. The path ahead requires stable generalized devnets, client releases, security audits, and public testnet activations on Holesky and Sepolia before a mainnet fork date is announced.
The stakes extend beyond throughput. Approximately 90% of Ethereum blocks are currently built through off-chain relay infrastructure, with Titan Builder controlling roughly 52% and BuilderNet (the decentralized builder launched by Flashbots, Beaverbuild, and Nethermind) holding approximately 25%. Glamsterdam's ePBS would move this coordination on-chain, eliminating the trust assumptions embedded in the current relay system.
Glamsterdam — a portmanteau of Amsterdam (Execution Layer) and Gloas (Consensus Layer) — is the successor to the Fusaka upgrade and the third hard fork in the post-Merge era, following Pectra (May 2025) and Fusaka (late 2025). Vitalik Buterin outlined eight EIPs defining the upgrade's scope in late February 2026. Two were designated as "headliners": EIP-7732 and EIP-7928. The remaining proposals cover gas repricing (EIP-7904, EIP-8037, EIP-8038), maximum contract size increases (EIP-7954), and related adjustments.
Over 30 proposals were explicitly excluded from the upgrade scope, including FOCIL inclusion lists (deferred to the subsequent Hegotá fork), reduced slot times, and multidimensional gas metering.
The current block production pipeline works as follows: specialized builders assemble transaction bundles optimized for Maximal Extractable Value (MEV), submit them to relays, and validators blindly select the highest-paying block via MEV-Boost. According to data from Rated Network, approximately 90% of Ethereum blocks are produced through this off-chain pipeline.
The concentration is severe. As of early April 2026, Titan Builder produces roughly 52% of blocks. BuilderNet, the decentralized block builder Flashbots launched in December 2024 with Beaverbuild and Nethermind, handles approximately 25%. Beaverbuild independently accounts for under 2%. The remaining share is fragmented among smaller builders.
EIP-7732 moves this coordination on-chain. Under ePBS:
The relay layer — currently a trust-dependent intermediary — becomes unnecessary. The protocol itself enforces the separation. This removes the single-point-of-failure risk inherent in relay infrastructure and addresses censorship concerns. According to prior research, approximately 30% of blocks built through OFAC-compliant relays have excluded sanctioned transactions.
However, as Buterin himself has noted, ePBS prevents builder centralization from spilling into the validator layer, but builder centralization itself persists. The subsequent Hegotá upgrade is expected to address this through FOCIL (Forced Inclusion Lists) via EIP-7805.
Ethereum's execution bottleneck is not computation — it is state access. The EVM can process instructions quickly, but each storage read requires a round trip to a growing on-disk database. Nodes discover which accounts and storage slots a block touches only during execution, making parallelization impossible.
EIP-7928, authored by Toni Wahrstätter, Dankrad Feist, and several other core researchers, introduces Block-Level Access Lists (BALs). Each block must declare in advance every account and storage slot it will access, along with post-execution state diffs.
This pre-declaration enables four types of parallelism:
The BAL root is embedded in the block header. EL clients must store BALs for a minimum of 3,533 epochs. A new wire protocol, eth/71, handles peer-to-peer BAL distribution.
Short-term execution improvements are estimated at 10–30x. Combined with the gas limit increase, this forms the technical basis for the 10,000 TPS target.
EIP-7904 recalibrates gas costs to reflect actual computational costs on modern hardware. The current gas schedule dates to an era when Ethereum ran on consumer hardware with different performance characteristics. The repricing produces a projected 78.6% reduction across both simple ETH transfers and complex smart contract interactions.
Companion proposals EIP-8037 and EIP-8038 adjust costs in the opposite direction — increasing gas costs for account creation, storage creation, and cold storage reads. This decouples state growth from execution gas, addressing a long-standing concern: scaling compute capacity previously meant inflating Ethereum's state, which every full node must store indefinitely.
The net effect: computation gets cheaper, but state expansion does not. A Uniswap trade currently costing $3–8 in gas could drop below $1, according to developer estimates. Multi-step DeFi operations involving several contract calls would see proportionally larger savings.
The development path to devnet-0 was fragmented. Until late April 2026, ePBS and BALs were tested on separate networks (epbs-devnet and bals-devnet respectively). The generalized devnet launched on April 24 merges all components into a single test environment for the first time.
| Milestone | Status | |-----------|--------| | Devnet-4 (individual EIPs) | Complete | | Devnet-5 (individual EIPs) | In progress | | Glamsterdam-devnet-0 (unified) | Launched April 24, 2026 | | Client releases | Pending | | Security audits | Pending | | Holesky testnet activation | Pending | | Sepolia testnet activation | Pending | | Mainnet fork | Aspirational: June 2026; Realistic: Q3 2026 |
The Ethereum Foundation's Checkpoint #9 post noted that ePBS requires coordinating "partial blocks" and two-party interactions across the protocol stack — a level of complexity that has slowed implementation. Non-headliner features like gas repricing have added further complexity. Non-headlining EIP proposals have been open for submission since April 9, 2026.
Despite the slow pace, the Ethereum Foundation reported that client developers "don't seem to be particularly disheartened."
Glamsterdam's economic impact spans several dimensions:
Fee Revenue Redistribution. Gas repricing will reduce total fee revenue per transaction by approximately 79%. At current Ethereum L1 usage levels (roughly 1.1 million transactions per day), this implies a significant reduction in per-transaction validator revenue. However, the gas limit increase from 60M to 200M could accommodate 3x–10x more transactions per block, partially or fully offsetting the per-unit decline if demand materializes.
MEV Pipeline Restructuring. ePBS moves the builder-proposer interaction on-chain. Today, relays capture value as trusted intermediaries. Post-Glamsterdam, this intermediary layer is eliminated. Builders bid directly into the protocol. The economic question is whether the current relay operators (Ultra Sound at 32%, Titan at 25%, bloXroute at 26%, Flashbots at 3%) can transition their business models, or whether their revenue streams evaporate.
Validator Economics. The gas limit increase to 200 million will be phased — 100 million initially, 200 million once ePBS is fully operational. Higher throughput means more transactions per block, more MEV opportunities, and more base fee revenue. But ePBS constrains validators to selecting sealed blocks, removing their ability to extract MEV directly. The net effect on validator profitability is uncertain and will depend on builder competition dynamics.
Ethereum Foundation Treasury. The Foundation completed its 70,000 ETH staking target on April 3, 2026, depositing a final 45,034 ETH ($93 million) in uniform chunks of 2,047 ETH. The staking commitment generates an estimated $3.9–5.4 million in annual yield, shifting the Foundation from periodic ETH sales to protocol-native income. The Foundation retains over 100,000 ETH in unstaked liquid reserves.
Glamsterdam compounds the L2 cost reductions already delivered by Fusaka's PeerDAS implementation, which cut blob costs by approximately 40%. Glamsterdam expands blob capacity further — potentially up to 72 blobs per block — giving rollups more room to publish compressed transaction data.
For rollups like Arbitrum, Optimism, and Base, this translates to lower settlement costs on L1. End-user fees on L2s, already measured in fractions of a cent, would decrease further. The economic question for L2s is whether this makes them more sustainable (lower operating costs) or more commoditized (lower barriers to competition).
The gas repricing and higher gas limits also make Ethereum L1 more competitive with L2s for certain use cases. If a Uniswap trade on L1 drops below $1, the cost advantage of executing on an L2 narrows. This creates a potential tension in Ethereum's rollup-centric roadmap: an L1 that is too cheap could undermine the economic rationale for rollups.
Schedule Risk. ePBS is the first time Ethereum's consensus layer has been designed around two-party interactions (proposer and builder). The Ethereum Foundation has acknowledged the complexity. June is aspirational; a slip to Q3 or Q4 is plausible.
State Growth. A 200 million gas limit means 3x more state-altering operations per block. While EIP-8037 and EIP-8038 increase the cost of state creation, the net effect on state growth rate is unclear. Faster state growth makes running a full node more expensive, which could reduce the validator set over time.
Builder Centralization Persistence. ePBS prevents builder centralization from infecting the validator set, but Titan's 52% block production share is unchanged by the upgrade. The deeper structural fix — FOCIL — is deferred to Hegotá.
Client Diversity. The new eth/71 wire protocol and BAL storage requirements add implementation burden to all execution clients. Bugs in any single client handling these features could create consensus failures.
Glamsterdam is the most structurally ambitious Ethereum upgrade since the Merge. It does not merely adjust parameters — it rewrites how blocks are built, how state is accessed, and how gas is priced. The technical scope is commensurate with the risk: ePBS is proving difficult to implement, the timeline is uncertain, and the economic consequences for validators, builders, and L2 operators are not fully modeled.
What is clear from the data: Ethereum's current block production pipeline is centralized, relay-dependent, and trust-laden. Glamsterdam addresses the first two problems. The third — builder concentration — remains an open question for Hegotá. The market will price the upgrade's credibility based on devnet stability over the coming weeks. Until then, June remains a number on a whiteboard.