Ethereum's Glamsterdam upgrade — the largest protocol change since the 2022 Merge — activated on the Sepolia testnet on October 6, 2026, raising the block gas limit from 60 million to 200 million. Early test blocks used only 26% to 46% of that new capacity. A six-minute voting period produced all...
"I can just spin up a thousand builders, rotate them, offer very high bids, and not produce payloads." — Potuz, Ethereum Consensus Developer
Ethereum's Glamsterdam upgrade — the largest protocol change since the 2022 Merge — activated on the Sepolia testnet on October 6, 2026, raising the block gas limit from 60 million to 200 million. Early test blocks used only 26% to 46% of that new capacity. A six-minute voting period produced all 32 scheduled blocks, with 99.97% of eligible stake participating in finality. Mainnet deployment remains targeted for Q4 2026, but no date is set.
The upgrade bundles 18 EIPs across execution and consensus layers. Two headline changes dominate: Enshrined Proposer-Builder Separation (ePBS, EIP-7732), which moves block-building auctions from off-chain middleware directly into the consensus protocol; and Block-Level Access Lists (BALs, EIP-7928), which enable parallel transaction processing by pre-declaring which accounts and storage slots a block will touch. Together, they form the technical foundation for Ethereum to eventually target 200 million gas per block — roughly 3.3x current capacity — and a long-term path toward 10,000 TPS on Layer 1.
The stakes are material. MEV-Boost relays currently intermediate approximately 92% of Ethereum blocks. Neutral relays already build around 70% of blocks as of August 2026, but the entire relay layer — a $2.5 billion annual MEV market — faces architectural obsolescence under ePBS. Developers flagged a testnet-specific vulnerability in which fake builders can win auctions and withhold payloads at zero cost. The Hoodi testnet is tentatively scheduled for October 26. An All Core Devs consensus call on October 15 will coordinate next steps.
Glamsterdam activated on Sepolia at epoch 353,024, slot 11,296,768 (13:53:36 UTC, October 6, 2026). The fork combined the Amsterdam execution-layer changes with the Gloas consensus-layer changes, as confirmed by the Ethereum Foundation's September 17 testnet announcement.
Initial performance data, reported by CoinDesk on October 8, showed blocks running near the 200 million gas limit. However, early test blocks consumed only 26% to 46% of available capacity. A six-minute voting period produced all 32 scheduled blocks. Finality participation reached 99.97% of eligible testnet stake, according to TradingView's reporting of testnet metrics.
These results confirm that basic block production and consensus work under the new rules. They do not confirm that validators can sustain near-200 million gas blocks at scale or that the configuration is safe for mainnet. Validator strain under full-capacity loads remains an unresolved question.
The testnet operated with approximately 84,000 validators across multiple client implementations — roughly one-tenth the size of mainnet's active validator set.
EIP-7732 is the centerpiece of Glamsterdam. It brings Proposer-Builder Separation into Ethereum's consensus protocol, replacing the current off-chain MEV-Boost relay system that has intermediated block production since the Merge.
Under the current architecture, validators outsource block construction to specialized builders via MEV-Boost. Builders compete in off-chain auctions managed by relays — third-party intermediaries that verify builder commitments before forwarding them to proposers. The system works but introduces trust assumptions and centralization risk. As of August 2026, MEV-Boost adoption stood at approximately 92% of staked validators, according to relay tracking data.
The relay market has shifted substantially. Flashbots, which once built over 82% of relay blocks, produced just 2.4% of daily blocks by late August 2026. Ultra Sound Relay led with 35.1%, followed by Titan Relay at 29.8% and Aestus at 7.2%. Neutral relays — those not filtering transactions based on OFAC sanctions — accounted for more than 70% of blocks, up from roughly 30% in early 2024.
ePBS eliminates the relay layer entirely. Under EIP-7732, builders become staked consensus-layer actors. A proposer includes a builder's signed commitment to an execution payload in the beacon block. The builder subsequently reveals the payload. A Payload Timeliness Committee attests to whether the payload was revealed on time. The protocol handles proposer payment directly.
The economic implications are significant. Industry trackers estimate Ethereum's annual MEV extraction at roughly $2.5 billion in 2026, with daily activity ranging from $10–20 million on normal days to $40–50 million during volatility events. Titan Builder alone booked over $112 million in block-building revenue in H1 2026. ePBS does not eliminate MEV extraction — it restructures how the value flows. Builders still compete. But the relay infrastructure that currently mediates those auctions becomes unnecessary.
The shift also expands the payload distribution window from 2 seconds to 9 seconds, giving builders more time to construct blocks and reducing missed-slot risk.
EIP-7928 introduces Block-Level Access Lists (BALs), which record every account and storage slot accessed during block execution, along with post-execution values. A hash of this access list — the BAL root — is embedded directly in the block header.
The practical effect: clients can prefetch data and execute non-conflicting transactions in parallel across multiple CPU cores. Without BALs, Ethereum processes transactions sequentially because a client cannot know in advance which state a transaction will touch. With BALs, sequential execution becomes a dependency graph. Only transactions that genuinely conflict must wait for each other.
World Chain, an L2 built by Worldcoin's parent entity, deployed EIP-7928 ahead of Ethereum mainnet on August 17, 2026, reporting throughput of 1 gigagas per second without higher hardware demands. For Ethereum L1, the target is a 200 million gas limit — approximately 16.7 Mgas/s of execution throughput — up from the current 60 million.
BALs and ePBS are complementary. ePBS separates consensus validation from execution validation, giving validators more time to process execution payloads. BALs make that processing parallelizable. Together, they make it technically defensible to raise the gas limit without making blocks too slow to verify.
Glamsterdam restructures Ethereum's gas pricing to better reflect actual resource costs. Key changes include:
The Ethereum Foundation's announcement explicitly warned developers to "review contracts with fixed gas stipends or hardcoded limits," as these gas accounting changes will alter the cost profile of existing smart contracts.
Additional EIPs include EIP-7708 (ETH transfers emit a log), EIP-7843 (SLOTNUM opcode), and EIP-7954 (increased maximum contract size). The full upgrade includes 18 core EIPs plus supporting networking and state-healing specifications.
Developers identified a testnet-specific vulnerability in the ePBS auction mechanism. On Sepolia, where test ether is free, a malicious actor can generate thousands of fake builder identities, place winning bids with inflated offers, then withhold the required execution payloads.
Consensus developer Potuz described the attack vector during pre-launch discussions: an attacker could spin up a thousand builders, rotate identities, and offer arbitrarily high bids while never producing payloads. Existing protocol defenses activate only after multiple payload failures from a single identity. Identity rotation renders those defenses ineffective on a testnet where creating new identities costs nothing.
On mainnet, builders must stake ETH, raising the cost of this attack. However, the testnet vulnerability complicates Ethereum's ability to fully stress-test ePBS under adversarial conditions before mainnet deployment. Developers indicated that builder-specific banning mechanisms remain under development.
All six consensus-layer clients (Grandine, Lighthouse, Lodestar, Nimbus, Prysm, Teku) and all six execution-layer clients (Besu, Ethrex, Erigon, Geth, Nethermind, Reth) shipped Glamsterdam-compatible releases ahead of the Sepolia activation.
Prysm, maintained by Offchain Labs, required a last-minute patch. Version 7.2.0 defaulted to a 60 million gas limit post-fork, requiring manual reconfiguration for the 200 million target. Offchain Labs released v7.2.1 on October 5 at 21:18 UTC — approximately 17 hours before activation — correcting the default and enabling partial data columns.
Teku 26.9.1 still defaults to 60 million gas and requires operators to set the higher limit via command-line flag. This divergence in default behavior across clients introduces operational risk if node operators do not update configurations.
The Ethereum Bug Bounty Program was activated for Glamsterdam specifications upon confirmation of client releases.
Glamsterdam arrives as Ethereum's validator set undergoes structural compression. Active validators have fallen from approximately 975,000 at the start of 2026 to roughly 841,332 as of October 14, 2026 — a decline of approximately 14%. Yet total staked ETH has risen from 36.3 million to 43.7 million over the same period, an increase of 7.4 million ETH.
The driver is EIP-7251, implemented earlier in 2026, which raised the maximum effective balance per validator from 32 ETH to 2,048 ETH. Lido, the largest liquid staking protocol, is consolidating over 265,000 individual validators into approximately 4,000 larger ones. If completed, this consolidation alone could reduce the total validator count by roughly one-third.
Fewer, larger validators reduce consensus overhead — fewer attestation messages, faster finality — which supports Glamsterdam's objective of increasing L1 throughput. The 43.7 million staked ETH represents approximately 35.8% of Ethereum's total supply.
| Milestone | Date | Status | |-----------|------|--------| | Devnet-11 test | September 16, 2026 | Completed | | Sepolia activation | October 6, 2026 | Completed | | Client update target | October 16, 2026 | Pending | | ACDC #189 call | October 15, 2026 | Scheduled | | Hoodi testnet | October 26, 2026 | Tentative | | Mainnet | Q4 2026 | No date set |
The All Core Devs Consensus call (#189) on October 15 will review Sepolia results and coordinate the Hoodi timeline. According to the Ethereum Foundation, "Hoodi and mainnet activation dates have not yet been decided." A subsequent upgrade, Hegotá, is in early planning with a tentative mainnet target of Q2 2027.
Glamsterdam is the most structurally significant Ethereum upgrade since the Merge. ePBS rewrites how blocks are built. BALs enable how they are processed. Gas repricing adjusts how they are costed. Taken together, these changes lay the infrastructure for Ethereum L1 to operate at 200 million gas per block — a 3.3x increase that, combined with parallel execution, targets an eventual throughput of 10,000 TPS.
The Sepolia results are encouraging but incomplete. Full-capacity stress testing has not occurred. Builder attack mitigations need further work. Client divergence on gas limit defaults introduces operational risk. The next 30 days — through the Hoodi testnet and the mainnet date decision — will determine whether Q4 2026 deployment is realistic or whether Glamsterdam pushes into 2027.
The economic implications extend beyond technical capacity. A $2.5 billion annual MEV market currently flows through a small number of relay operators. ePBS does not eliminate MEV extraction, but it eliminates the intermediaries. For relay operators, builders, and validators, the revenue map is being redrawn.