Ethereum's largest protocol upgrade since The Merge is now running on multi-client development networks and targeting mainnet activation in Q3 2026. Glamsterdam introduces enshrined Proposer-Builder Separation (ePBS) via EIP-7732, Block-Level Access Lists (BALs) via EIP-7928, and state creation g...
"This [ePBS] ensures that block builder centralization does not creep into staking centralization, but it leaves the question: what do we do about block builder centralization?" — Vitalik Buterin, Ethereum Co-Founder, Block Building Roadmap Post, 2026
Ethereum's largest protocol upgrade since The Merge is now running on multi-client development networks and targeting mainnet activation in Q3 2026. Glamsterdam introduces enshrined Proposer-Builder Separation (ePBS) via EIP-7732, Block-Level Access Lists (BALs) via EIP-7928, and state creation gas repricing via EIP-8037 — a three-pronged overhaul designed to raise the block gas limit from 60 million to 200 million and push Layer 1 throughput toward 10,000 transactions per second.
The upgrade addresses a structural bottleneck: over 88% of Ethereum blocks are currently built through MEV-Boost relays, with the top three builders controlling more than 80% of all PBS blocks. Glamsterdam moves this off-chain relay infrastructure into the protocol itself, separating the roles of block proposers and block builders at the consensus layer. The economic implications are material — researchers estimate ePBS could reduce MEV extraction by up to 70%, while gas repricing under EIP-7904 targets a 78.6% reduction in fees for both simple transfers and complex smart contract interactions.
Scope management has been deliberate. FOCIL (censorship-resistance inclusion lists), Verkle Trees, and account-abstraction upgrades were moved to Hegotá, a follow-up fork targeting late 2026 or early 2027, after the Base engineering team warned that bundling FOCIL alongside ePBS could delay the entire upgrade beyond 2026.
Ethereum's current block production pipeline relies on an off-protocol mechanism called MEV-Boost, maintained by Flashbots. Validators outsource block construction to specialized builders who compete to assemble the most profitable blocks, then submit them through relay intermediaries.
The concentration data tells the story. According to RelaysScan data from April 2026, the relay landscape breaks down as follows:
| Relay | Payload Share | |-------|-------------| | relay.ultrasound.money | 33.92% | | titanrelay.xyz | 24.19% | | bloxroute.max-profit | 14.67% | | aestus.live | 10.03% | | bloxroute.regulated | 9.07% | | boost-relay.flashbots.net | 4.22% |
At the builder level, concentration is more severe. Titan controls 52.16% of blocks by count, BuilderNet holds 24.63%, and Quasar captures 15.06%. Three entities build over 90% of Ethereum blocks.
This represents improvement from 2022, when Flashbots alone processed approximately 80% of all MEV-Boost blocks. But the fundamental problem remains: critical infrastructure sits outside the protocol, dependent on trusted intermediaries. Over 50% of high-value Ethereum transactions were routed through private channels by mid-2025 to avoid MEV extraction, according to Bitfinex research.
Glamsterdam's core consists of three interdependent Ethereum Improvement Proposals.
EIP-7732 formalizes the proposer-builder handoff directly in the consensus protocol. Under the current system, validators trust external relays to faithfully relay builder bids. Under ePBS, the protocol itself mediates the exchange.
The mechanism works through a commit-reveal flow. Builders cryptographically seal blocks and commit to bids. Validators select the highest bid without seeing transaction contents. A new Payload Timeliness Committee (PTC) — a subset of validators — attests separately to whether the execution payload was delivered on time, introducing dual-deadline logic that separates consensus block attestation from execution payload verification.
A key consequence: the data propagation window expands from approximately 2 seconds to roughly 9 seconds, unblocking the network's ability to safely handle larger blocks and more data blobs for Layer 2 networks.
The "free option problem" — where builders can observe market movements after committing a bid and choose not to deliver — currently affects roughly 0.82% of blocks on average, rising to approximately 6% during volatile periods. ePBS's commit-reveal structure is designed to mitigate this, per Bitfinex's technical analysis.
Block-Level Access Lists function as a pre-declared map included in every block, identifying which database storage slots each transaction will touch before execution begins. Because the network can see in advance which transactions access non-overlapping state, nodes can perform parallel disk reads and process independent transactions simultaneously.
This is the mechanism that enables the gas limit increase. Without BALs, a 200 million gas block processed sequentially would overwhelm node hardware. With BALs, parallel execution keeps processing requirements within the capacity of standard physical infrastructure.
EIP-8037 introduces a dynamic pricing model for state creation operations, targeting a database growth rate of 100 GiB per year. The current gas schedule underprices operations that permanently expand the state database — creating new accounts, deploying contracts, writing new storage slots. EIP-8037 harmonizes costs by tying them proportionally to the amount of permanent data an operation creates.
Complementing this, EIP-7904 realigns gas costs with actual computational resources consumed. According to Ethereum Foundation documentation, the combined repricing produces a 78.6% reduction in gas costs for both simple ETH transfers and complex smart contract interactions under the new 200 million gas limit.
The gas limit increase will be phased, according to former Ethereum Foundation co-executive director Tomasz Stanczak. The first phase reaches 100 million per block; the second reaches 200 million once ePBS is fully operational. The 200 million target represents a 3.3x increase from the current 60 million ceiling.
The throughput implications at 200 million gas:
| Metric | Current | Post-Glamsterdam | |--------|---------|-----------------| | Block Gas Limit | 60M | 200M (phased) | | Effective L1 TPS | ~1,000 | ~10,000 (target) | | Gas Cost Reduction | Baseline | ~78% | | State Growth Target | Unmanaged | 100 GiB/year |
The 10,000 TPS figure is a theoretical target contingent on BAL-enabled parallel execution reaching design specifications. Actual throughput will depend on transaction mix, state access patterns, and the proportion of parallelizable transactions in any given block.
One stated rationale for the capacity increase: Buterin has framed it as preparation for the "Agentic Economy," in which AI agents execute millions of daily micro-transactions on the base layer. Whether this demand materializes is speculative, but the infrastructure capacity is being built ahead of it.
Note: EIP-7782 (6-second slot times) was removed from Glamsterdam's scope due to conflicts with ZK proving timelines. This removes a raw throughput multiplier that faster slots would have provided.
Glamsterdam arrives during a period of structural revenue decline on Ethereum's L1. Mainnet fee revenue has dropped to approximately $500,000 per month as Layer 2 networks capture an estimated 94% of transaction margins. Ethereum still leads all blockchains in annual transaction fees at $2.73 billion, but the composition has shifted: the base layer increasingly functions as a settlement and data availability layer rather than a direct execution environment.
EIP-4844 (implemented in 2024's Dencun upgrade) made Layer 2 the default execution venue by cutting L2 data posting costs by 90-99%. By 2026, up to 80% of Ethereum-related transactions occur on Layer 2 solutions, according to CoinLaw data. Base leads L2 fee revenue at $49.4 million annually, followed by Arbitrum at $33.2 million and Optimism at $27.9 million.
The question Glamsterdam implicitly raises: can a 78% fee reduction and 10x throughput increase pull meaningful transaction volume back to L1, or does it primarily benefit the L2 ecosystem by reducing settlement costs? The economic value distribution across the Ethereum stack may shift materially depending on the answer.
As of early June 2026, the Ethereum Foundation's protocol update confirms the following milestones:
The original target was H1 2026. Current consensus among developers and commentators points to Q3 2026 — likely late August — as the realistic mainnet activation window, per CryptoTimes reporting from June 5, 2026. The Glamsterdam upgrade requires updates to both execution clients and consensus clients, meaning node operators must ensure both layers are upgraded to handle the new block production mechanics.
Three significant features were moved from Glamsterdam to Hegotá, a follow-up fork targeting late 2026 or early 2027:
FOCIL (Forward Obligatory Commitment to Inclusion Lists): A censorship-resistance mechanism requiring 16 randomly selected attesters to nominate transaction sets that must appear in a block. Blocks get rejected if nominated transactions are absent. The Base engineering team publicly warned that adding FOCIL alongside ePBS could delay the upgrade beyond 2026 entirely.
Verkle Trees: A storage structure replacement projected to cut node storage requirements by up to 90% and enable stateless clients. The implementation complexity was deemed too high for concurrent deployment with ePBS.
Account Abstraction: Native smart-contract wallet support at the protocol level.
The scope reduction was a deliberate trade-off: ship ePBS and the gas limit increase on a predictable timeline, defer censorship resistance and storage optimization to the next fork. The Glamsterdam-then-Hegotá sequence represents a two-phase approach — first formalize the block production market (ePBS), then enforce inclusion guarantees (FOCIL).
Centralization Transfer: ePBS prevents builder centralization from infecting staking centralization, but does nothing to address builder centralization itself. With three builders producing over 90% of blocks, the power dynamic shifts rather than resolves. Buterin has acknowledged this gap explicitly and proposed "Big FOCIL" and encrypted mempools as future mitigations, according to The Block.
Phased Gas Limit: The 100M-then-200M approach introduces uncertainty about the second phase timeline. If ePBS encounters stability issues post-mainnet, the 200M target could slip indefinitely, meaning throughput improvements underperform projections.
Client Diversity: Glamsterdam requires updates to both execution and consensus clients. The coordination burden is non-trivial — node operators must upgrade both layers, and any client-specific bugs could fragment the network during activation.
State Growth vs. Fee Revenue: EIP-8037 makes state creation more expensive while EIP-7904 makes execution cheaper. If the net effect discourages on-chain activity rather than redirecting it, L1 revenue could decline further from already-compressed levels.
L2 Competition: A 10x throughput increase on L1 at 78% lower fees narrows the cost advantage of Layer 2 networks. This could create competitive tension between L1 and its own L2 ecosystem — an outcome with unclear net effects on the Ethereum value stack.
Glamsterdam represents Ethereum's most consequential architectural change since the Proof-of-Stake transition in September 2022. The upgrade converts an off-protocol dependency — MEV-Boost relays — into a native consensus mechanism, while enabling the gas limit headroom for meaningful L1 throughput scaling.
The economic implications cut in multiple directions. Lower L1 fees and higher throughput could recapture transaction volume from Layer 2 networks, or they could primarily reduce settlement costs for those same L2s — reinforcing rather than reversing the current value distribution. The 78% fee reduction arrives while L1 revenue sits at approximately $500,000 per month, a fraction of historical peaks.
The scope management — shipping ePBS without FOCIL — is a calculated bet that the benefits of faster deployment outweigh the risks of operating without protocol-level censorship resistance during the interim period. Whether that bet pays off depends on how quickly Hegotá follows, and whether builder centralization creates exploitable censorship vectors before inclusion lists arrive.
Development networks are stable. The EIP specifications are finalized. The Q3 2026 window looks achievable. What remains uncertain is not whether Glamsterdam ships, but whether 10,000 TPS on a structurally reformed L1 changes Ethereum's economic trajectory — or merely makes the existing settlement-layer model more efficient.