Ethereum's Glamsterdam hard fork — the network's most significant protocol overhaul since the September 2022 Merge — has entered its final devnet phase. Two headline EIPs anchor the upgrade: EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists). Together, they r...
"Glamsterdam is probably the largest fork we've had since the Merge. It will change a lot of assumptions about Ethereum and set us up for much more scaling in the future." — Parithosh Jayanthi, Ethereum Foundation Core Developer
Ethereum's Glamsterdam hard fork — the network's most significant protocol overhaul since the September 2022 Merge — has entered its final devnet phase. Two headline EIPs anchor the upgrade: EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists). Together, they restructure how Ethereum produces blocks, prices state, and executes transactions.
The upgrade targets a 200 million gas-limit floor, up from the current 60 million — a 3.3x increase in raw throughput capacity. Mainnet activation is internally targeted for late August 2026, though core developers describe this as aspirational. A firmer base case, according to multiple staking providers, falls in the September–December 2026 window. The scope is large: ePBS alone eliminates the trust dependency on third-party MEV relays that currently intermediate over 90% of Ethereum block production, while EIP-8037 introduces a dedicated state-gas reservoir to keep annual state growth below 120 GiB even at the higher gas limit.
The economic stakes are material. Over $550 million in MEV is extracted from Ethereum annually. More than 95% of the network's 1.2 million validators use MEV-Boost, a relay-dependent system that adds 10–30% to base staking rewards. Glamsterdam moves this infrastructure on-chain — but academic research warns that doing so may amplify builder centralization rather than reduce it.
Ethereum's block production pipeline currently depends on MEV-Boost, an off-chain system created by Flashbots in 2022. In this system, specialized builders assemble transaction bundles, relays verify and transmit them, and proposers (validators) select the highest-paying block. The relay sits in the middle as a trusted intermediary: builders trust relays not to steal their transaction ordering strategies; proposers trust relays to deliver valid blocks.
EIP-7732 eliminates this intermediary layer. Under ePBS, the block production auction moves into the consensus protocol itself. Builders assemble blocks and cryptographically seal their contents. Proposers select the highest bid without seeing the block's internal transaction ordering. Transactions are revealed only after the block is finalized.
As of July 19, 2026, the relay market shows the following distribution: Ultrasound relay holds 26.03% of payloads, Titan relay 21.51%, BloXroute max-profit 20.95%, BloXroute regulated 17.91%, Aestus 7.54%, and Flashbots — the original relay operator — just 3.33%. This fragmentation has occurred organically, but the underlying trust model remains unchanged: validators must trust at least one relay to deliver valid, non-manipulated blocks.
Vitalik Buterin has framed ePBS narrowly: "ePBS only prevents builder centralization from spilling over into the staking layer, but the issue of builder centralization itself remains." This is a containment measure, not a cure.
While the relay market has diversified, the builder market has moved in the opposite direction. Current data from Relayscan shows Titan Builder commanding 48.74% of Ethereum blocks, followed by Quasar at 23.24% and Eureka at 16.17%. Three builders produce approximately 88% of all Ethereum blocks.
A January 2026 research paper by Bruno Mazorra et al. ("The Free Option Problem of ePBS," arXiv: 2509.24849) models the economic dynamics under EIP-7732 and finds that ePBS may "significantly amplify profit and content centralization." The mechanism: under an on-chain auction, builders with superior MEV extraction capabilities and lower latency consistently outbid competitors, creating a positive-feedback loop where dominant builders accumulate more capital to bid even more aggressively.
This concentration has censorship implications. According to MEV Watch, OFAC-compliant relays (which exclude sanctioned addresses from blocks) have historically carried a significant share of block production. Under ePBS, the censorship question shifts from relays to builders — and three builders controlling 88% of blocks represents a narrower chokepoint than the current relay distribution.
ePBS introduces a structural vulnerability that does not exist in the current MEV-Boost system. Under EIP-7732's dual-deadline design, a builder commits to a block bid and, if selected, must later reveal the full block payload. However, the builder retains unilateral control over whether to distribute the payload data. If market conditions shift between commitment and reveal — for example, a large price movement makes the committed transaction ordering suboptimal — the builder can withhold the payload without protocol-level penalty.
The result: the slot produces an empty block, degrading network liveness. The builder's commitment functions as a free option — the right, but not the obligation, to fill the slot.
Flashbots has documented this risk. Shutter Network has proposed a cryptographic mitigation using silent threshold encryption: builders encrypt the block to a set of decryption parties (Keypers), attach a zero-knowledge proof of consistency with their commitment, and the block is collaboratively decrypted after the proposer signs. This eliminates the builder's ability to selectively withhold, but introduces additional cryptographic infrastructure and trust assumptions around the Keyper set.
As of July 2026, no mitigation for the free option problem is included in the Glamsterdam EIP bundle. Core developers have opted to ship ePBS without this fix and address it in subsequent upgrades.
EIP-7928 introduces Block-Level Access Lists (BALs), a mechanism that maps which parts of Ethereum's state database each transaction will read or write — before execution begins. This pre-declaration enables client software to identify non-conflicting transactions and execute them in parallel rather than sequentially.
Current Ethereum execution is strictly serial: each transaction is processed one after another, even when they touch entirely different parts of the state. BALs provide the metadata layer needed for clients to safely parallelize execution without risking state conflicts.
The throughput implications compound with the gas-limit increase. At 200 million gas with parallel execution, Ethereum L1 could process substantially more transactions per block than the current sequential model at 60 million gas allows. Developer documentation references a long-term path toward 10,000 TPS on L1, though this figure depends on additional optimizations beyond Glamsterdam.
The gas-limit increase from 60 million to 200 million cannot ship without addressing state growth. Under current pricing, a 200 million gas limit would generate over 380 GiB of new state data per year — crossing the performance degradation threshold for consumer-grade node hardware in under 12 months.
EIP-8037 (State Creation Gas Cost Increase) separates the pricing of temporary computation from permanent data storage. It introduces a Cost Per State Byte (CPSB) mechanism that targets a sustainable growth rate of 120 GiB per year. Critically, EIP-8037 creates a dedicated state_gas_reservoir — a separate gas pool for state-creating operations, distinct from the standard execution gas limit.
This decoupling means that applications generating large amounts of persistent data cannot monopolize the execution gas budget. Computation-heavy but state-light transactions (such as DeFi swaps that modify existing balances) become cheaper relative to state-heavy operations (such as deploying new contracts or creating new storage slots).
EIP-8037 was finalized in May 2026. Client teams describe it as the "final piece" that made the 200 million gas target viable without compromising node decentralization.
As of mid-July 2026, the Glamsterdam upgrade status is:
Ethereum Foundation contributors have noted that Glamsterdam is "proving trickier and slower than Fusaka," and that "getting ePBS right outranks any fixed date." The original June 2026 target has already slipped.
Glamsterdam's scope was narrowed during development. Two significant features were deferred to the subsequent Hegotá fork, targeted for late 2026 or early 2027:
FOCIL (Fork-Choice Inclusion Lists): A censorship-resistance mechanism that would allow validators to force-include transactions that builders are excluding. FOCIL was removed from Glamsterdam to avoid delaying the fork and to keep scope manageable. Its absence means that ePBS ships without a protocol-level anti-censorship backstop — a noted gap given the builder concentration data above.
Verkle Trees: A state storage optimization that could reduce node storage requirements by up to 90% and enable stateless clients. Verkle Trees were moved to Hegotá to decouple the state storage overhaul from the block production overhaul.
The long-term Ethereum roadmap (the "Strawmap Vision") targets approximately 10,000 TPS on L1, native privacy features, and post-quantum cryptographic security through 2029.
The economic redistribution from Glamsterdam is significant across several stakeholder groups:
Validators: MEV-Boost currently adds 10–30% to base validator rewards, equivalent to roughly 0.28–0.83% additional APR. Under ePBS, this revenue stream moves on-chain but the distribution mechanics change. Validators no longer select blocks based on relay trust relationships; they select based on protocol-enforced auction bids. Whether this increases or decreases average validator revenue depends on the competitiveness of on-chain auctions versus current relay-mediated auctions.
Relay Operators: Flashbots, BloXroute, Ultrasound, Titan, and Aestus collectively operate infrastructure that intermediates over 90% of Ethereum blocks. ePBS renders this relay infrastructure unnecessary at the protocol level. Relay operators will need to find new value propositions — likely in MEV strategy optimization, builder services, or adjacent infrastructure.
Builders: The three dominant builders (Titan, Quasar, Eureka) may see their market position reinforced under ePBS, according to the Mazorra et al. research. Without FOCIL or other anti-concentration measures, the on-chain auction format may favor incumbents with superior MEV extraction capabilities.
Application Users: The 3.3x gas-limit increase combined with parallel execution should reduce transaction fees and increase throughput on L1. EIP-8037's state-gas separation further benefits computation-heavy transactions (DeFi, DEX trades) while appropriately pricing state-heavy operations (contract deployments, NFT mints).
Glamsterdam targets a 3.3x gas-limit increase (60M to 200M) and introduces parallel transaction execution via Block-Level Access Lists (EIP-7928). Both changes require EIP-8037's state pricing reform to remain sustainable.
ePBS (EIP-7732) eliminates the trusted relay layer that currently intermediates over 90% of Ethereum block production. This removes a trust dependency but does not address underlying builder concentration: three builders currently produce 88% of blocks.
The free option problem — a structural vulnerability where builders can withhold block payloads without penalty — ships unresolved. Proposed cryptographic fixes (Shutter Network's threshold encryption) are not included in the Glamsterdam bundle.
FOCIL, the censorship-resistance mechanism, was deferred to the Hegotá fork. ePBS therefore launches without a protocol-level tool to counteract builder-level transaction censorship.
Mainnet activation is internally targeted for late August 2026 but realistically falls in the September–December 2026 window, based on historical testnet timelines.
The annual MEV extraction market on Ethereum exceeds $550 million. ePBS restructures how this value flows but does not necessarily reduce concentration among the entities capturing it.
Glamsterdam represents Ethereum's most ambitious protocol change since the Merge, touching block production, transaction execution, and state pricing simultaneously. The upgrade addresses real structural problems: relay trust dependencies, sequential execution bottlenecks, and unsustainable state growth under higher gas limits.
The tradeoffs are equally concrete. ePBS ships without censorship-resistance tooling (FOCIL) and without a fix for the free option problem. Academic research suggests the on-chain auction format may deepen builder concentration rather than alleviate it. The gas-limit increase depends entirely on EIP-8037's untested state-pricing model holding up under production workloads.
For the Ethereum ecosystem, Glamsterdam is a calculated bet: accept known risks in block production economics to unlock material improvements in throughput and state management. Whether that bet pays off depends on the Hegotá fork delivering the censorship-resistance and state-storage reforms that Glamsterdam deliberately deferred.