Ethereum's Glamsterdam Upgrade: ePBS, Parallel Execution, and the Biggest Protocol Change Since The Merge

Ethereum's next major network upgrade, Glamsterdam, is currently in active development and is documented as a Q4 2026 plan on ethereum.org. The upgrade is defined by meta EIP-7773 and centers on two confirmed proposals: EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists). ePBS restructures how block builders and validators interact by introducing in-protocol builder payments through a Payload Timeliness Committee of 512 validators, while BALs enable parallel transaction execution by tracking which accounts and storage slots are accessed in each block. Together, these changes support a design target of a 200M gas limit floor and aim to reduce database bloat. Several claims circulating online, including a "78.6% fee cut," "10,000 TPS," or an August 2026 activation date, lack confirmation from primary Ethereum sources.

Key Takeaways

Glamsterdam is a confirmed Ethereum upgrade defined by Draft meta EIP-7773, with a Q4 2026 plan per ethereum.org.
The two headliner proposals are EIP-7732 (ePBS) and EIP-7928 (BALs), both in Review status.
ePBS removes the ExecutionPayload from the consensus layer beacon block and replaces it with a SignedExecutionPayloadBid, managed by a 512-validator Payload Timeliness Committee.
BALs add a block_access_list_hash to the header and enable parallel disk reads, parallel transaction execution, and executionless state sync.
The 200M gas limit is a verified design target from the EF Soldogn interop recap (May 2026), not a fork-enforced rule.
The actual gas repricing package (EIP-8037/8038/7976/7981/2780) mostly increases costs for state creation and access. This is opposite of a blanket fee cut.
No primary source confirms an August 2026 activation date; the activation table in EIP-7773 remains empty.
The 32 ETH solo-staker minimum is explicitly NOT being changed by Glamsterdam.

What is Glamsterdam?

Glamsterdam is an Ethereum network upgrade currently in development. It is the successor fork to Fusaka, which activated on December 3, 2025, and the predecessor to Hegotá. The upgrade is documented on the canonical roadmap page at ethereum.org/roadmap/glamsterdam, last updated August 6, 2026. It is defined by the Draft meta EIP EIP-7773, which lists the proposals scheduled for inclusion, those under consideration, and those declined.

The meta EIP carries the mascot "Gloas" (from Gloas the polar bear, representing the consensus layer, and Amsterdam, representing the execution layer). Its activation table is currently empty. No Sepolia, Hoodi, Holesky, or Mainnet activation timestamp has been set. The upgrade exists as a target on the roadmap rather than a scheduled fork.

EIP-7732: Enshrined Proposer-Builder Separation

EIP-7732 is the first of Glamsterdam's two headliner proposals. Authors include D'Amato, Flaig, Monnot, Neuder, Potuz, Traglia, and Tsao, with the proposal reaching Review status in June 2024. It fundamentally restructures how block production works on Ethereum by separating consensus-layer validation from execution-layer validation.

Before ePBS, proposers built blocks themselves or relied on MEV-Boost relays to receive payloads from external builders. ePBS removes the ExecutionPayload field from BeaconBlockBody and replaces it with a SignedExecutionPayloadBid. Builders compete by submitting bids that include the payload and a payment amount. The proposer selects the highest bid, and the payment is trustlessly deducted from the builder's consensus-layer balance. This removes reliance on MEV-Boost relays as intermediaries.

A key addition is the Payload Timeliness Committee (PTC), a group of 512 validators (where PTC_SIZE = 2^9) that attest to the timeliness and validity of builder payloads. PTC members do not validate the execution payload itself; they confirm that the payload arrived within acceptable time bounds. The proposer gets 6 seconds to process the payload after selection, while other validators get 9 seconds. This expanded propagated window replaces the previous 2-second window, giving clients more time to handle complex execution data.

The proposal introduces a new entity type: the Builder. Builders are staked entities that register separately from validators, using the BUILDER_WITHDRAWAL_PREFIX (0x03). The minimum stake to become a builder is 1 ETH, which is distinct from the 32 ETH validator minimum. A sentinel value BUILDER_INDEX_SELF_BUILD allows a validator to build its own payload when no external bid is selected. The change is backwards-incompatible on the consensus layer and requires a hard fork.

EIP-7928: Block-Level Access Lists

EIP-7928 is the second headliner proposal, authored by Wahrstätter, Feist, D'Amato, Brouwer, Hagopian, Selmo, Raxhvl, and Qu0b. It reached Review status in March 2025. Unlike ePBS, BALs require no consensus-layer changes and are implemented purely as an execution-layer Engine API addition.

BALs enforce a block-level access list that tracks every account and storage slot accessed during transaction execution, along with their post-execution values. The list is included in the block header as a new field: block_access_list_hash. The average BAL size is approximately 70–72 KiB.

This design enables several optimizations. Clients can perform parallel disk reads by knowing which storage slots are needed upfront. Transactions that access disjoint sets of storage slots can be executed in parallel rather than sequentially. State-root computation can also be parallelized. Executionless state sync is also possible. A node can validate the state root without executing transactions if it has the BAL and knows the post-execution values.

Research indicates that approximately 60–80% of transactions access disjoint storage slots, which is the primary condition that makes parallelization effective at scale.

The 200M Gas Limit Target

One of the most widely discussed aspects of Glamsterdam is the target gas limit. As of early 2026, Ethereum's mainnet gas limit hovers around 60M. Post-Glamsterdam, developers are targeting a floor of 200M gas per block.

This target was confirmed in the Ethereum Foundation's Soldogn Interop Recap, published on the EF blog on May 2, 2026. The authors wrote that the team "converged on a credible 200M post-Glamsterdam gas limit floor." They explained that this is possible because ePBS structures the slot to give execution more time, BAL optimizations give clients throughput headroom, and EIP-8037 ensures that a higher gas limit does not translate into runaway state growth.

The 200M figure is a design target, not a consensus-enforced rule. EIP-8261 (Gas Limit Schedule, Informational status) defines an optional epoch-based GAS_LIMIT_SCHEDULE that can be configured in config.yaml. It is a recommendation for operators, not a protocol mandate. Validators and operators remain free to configure any gas limit value they choose. During devnet testing, developers use a 150M reference block to derive state pricing before scaling toward the 200M target.

Gas Repricing: What Actually Changes

The research brief flags an important correction regarding the task's original framing of "EIP-7904 gas repricing." EIP-7904 (Compute Gas Cost Analysis) is an Informational proposal that concludes no compute-gas repricing is needed. Its empirical analysis of EVM compute opcodes found that all examined opcodes and precompiles already meet the 100 Mgas/s throughput target at current costs, especially with EIP-7928 optimizations in place. The specification calls for no changes to the compute gas schedule.

The actual gas repricing proposals for Glamsterdam are packaged under EIP-8007, a Draft meta EIP that is a directory. This package includes:

EIP-8037 : State creation gas cost increase. Introduces a separate state-gas dimension (CPSB = 1530 gas/byte) to cap state growth at approximately 120 GiB/year at a 150M reference block (scaled to 200M). The existing GAS_CREATE cost of 32,000 is replaced by 120 × CPSB = 183,600 state gas. The GAS_NEW_ACCOUNT cost of 25,000 receives the same treatment. Backwards-incompatible.
EIP-8038 : State-access gas cost update. Raises COLD_ACCOUNT_ACCESS from 2,600 to 3,000. Introduces STORAGE_WRITE and ACCOUNT_WRITE surcharges. EXTCODESIZE and EXTCODECOPY are charged WARM_ACCESS for a second database read. Backwards-incompatible.
EIP-7976 : Increase calldata floor cost from 10/40 per byte to 64/64 per byte.
EIP-7981 : Price access lists with a 64 gas/byte new surcharge.
EIP-2780 : Decompose the flat 21,000 intrinsic transaction cost into explicit resource primitives.

The direction of these changes is mostly increases for state creation, state access, and calldata. This is the opposite of a blanket fee reduction. Any fee reductions would come from efficiency gains due to parallel execution (BALs) and reduced relay dependency (ePBS), not from lowering individual opcode costs.

Timeline: What Primary Sources Say

The official ethereum.org roadmap page states that Glamsterdam is "planned for Q4 2026." This is the most authoritative timeline source available. The activation table in EIP-7773 is empty for all testnets and mainnet, confirming no date has been locked in.

The EF Soldogn Interop Recap (May 2, 2026) notes that devnet work has "hardened Glamsterdam implementations" and that the team converged on the 200M gas-limit target. It does not state a mainnet activation date.

Secondary sources diverge. IG UK and Everstake cite an "end of August 2026" or "Q3 2026" target, but frame these as internal working targets or aspirational goals, not confirmed dates. Binance Academy describes a Q3 2026 timeline as "aspirational, not confirmed." CoinMarketCal claims an August 24, 2026 date but provides no Ethereum source. The Ethereum Foundation has not endorsed any of these dates.

Development status: the ePBS implementation tracker (ethpandaops HackMD) shows that ePBS is still not fully implemented across clients at the devnet-0 stage. Testnet deployment is ongoing on Sepolia and Hoodi. Node operators are advised to run supported client versions and test upgrades on testnets before mainnet activation.

Validator and Operator Impact

For solo stakers running 32 ETH, the minimum stake requirement is unchanged. Mike Neuder (Ethereum protocol) explicitly stated in a notes.ethereum.org post: "we do not propose increasing the 32 ETH minimum required to become a validator." Glamsterdam does not alter the staking requirements for validators.

The upgrade introduces a new actor type: the Builder. Builders must stake a minimum of 1 ETH and register with the BUILDER_WITHDRAWAL_PREFIX (0x03). This is separate from the validator role. A validator can also act as a builder, but the roles are distinct in the protocol.

Node operators must update to Glamsterdam-ready client versions for both execution and consensus layers. Gas-limit voting will follow the optional EIP-8261 schedule if operators configure it. The upgrade also brings faster exit processing, driven by EIP-8061 which increases the churn rate for exits and consolidations. This benefit applies to all validators and is not specific to ePBS.

Unverified Claims

Several quantitative claims about Glamsterdam appear in secondary sources but lack corroboration from primary Ethereum documentation. The following should be treated as unverified:

"78.6% lower L1 fees" : This figure appears in Binance Academy, IG UK, Phemex, and Reddit. EIP-7904 explicitly recommends no compute-gas repricing, and the actual gas repricing proposals (EIP-8037/8038) increase state creation and access costs. No EF or EIP source states a 78.6% reduction.
"10,000 TPS" : This figure appears in IG UK, Phemex, and Reddit. No primary source cites a specific TPS number for Glamsterdam. BALs enable parallelization, but the protocol does not specify a throughput target in transactions per second.
"Up to 70% MEV reduction" : This figure appears in IG UK, Phemex, and CoinMarketCap. Primary sources describe ePBS as reducing reliance on MEV-Boost relays and introducing in-protocol builder payments, but no quantified MEV-reduction percentage is stated in EIP-7732 or on ethereum.org.
"August 2026 activation" : Multiple secondary sources cite this date. The primary source (ethereum.org) states Q4 2026, and EIP-7773's activation table is empty.

Conclusion

Glamsterdam represents a significant architectural shift for Ethereum. The introduction of ePBS moves builder payments into the protocol, replacing the MEV-Boost relay model with an in-protocol bidding system managed by a 512-validator Payload Timeliness Committee. BALs unlock parallel execution by making storage access patterns explicit at the block level, enabling throughput improvements without changing individual opcode costs. The 200M gas limit target, verified through the Soldogn interop, reflects confidence that these optimizations can support a substantially larger block without proportional state growth. EIP-8037 caps state creation to keep growth around 120 GiB per year at current rates.

The upgrade is not yet scheduled for mainnet. The Q4 2026 window on ethereum.org is the best available timeline, and the empty activation table in EIP-7773 confirms no firm date exists. Developers should monitor testnet progress on Sepolia and Hoodi. Solo validators can expect no change to their 32 ETH requirement. The gas repricing package moves in the direction of higher state and access costs, not lower fees.

Frequently Asked Questions

Q: What is the difference between ePBS and MEV-Boost?

MEV-Boost uses external relays to match proposers with builders. ePBS removes the relay layer entirely: builders submit signed bids directly to proposers, and payments are deducted from the builder's consensus-layer balance through an in-protocol mechanism managed by the Payload Timeliness Committee.

Q: Does Glamsterdam change the 32 ETH staking requirement?

No. Mike Neuder (Ethereum protocol) explicitly stated that the 32 ETH minimum for validators is not being increased. The upgrade introduces a separate Builder role with a 1 ETH minimum stake, but this does not affect existing validators.

Q: When is Glamsterdam activating?

Ethereum.org lists a Q4 2026 plan. No specific date has been set in the activation table of EIP-7773. Secondary sources citing August 2026 are unverified by primary Ethereum documentation.

Q: Will Glamsterdam make Ethereum transactions cheaper?

The gas repricing proposals in Glamsterdam mostly increase costs for state creation and access (EIP-8037, EIP-8038). Any fee reductions would come from efficiency gains due to parallel execution (BALs) and reduced relay overhead, not from lowering individual opcode prices. EIP-7904 explicitly recommends no compute-gas repricing.

Q: What are BALs and why do they matter?

Block-Level Access Lists (EIP-7928) record every account and storage slot accessed during a block's execution. This allows clients to parallelize disk reads, execute independent transactions in parallel, and perform executionless state sync. Approximately 60–80% of transactions access disjoint storage slots, which is the condition that makes parallelization effective.