Ethereum's most consequential protocol change since the Merge is now nine weeks from mainnet. The Glamsterdam hard fork — targeted for May–June 2026 — enshrines Proposer-Builder Separation (ePBS) directly into Ethereum's consensus layer and introduces Block-level Access Lists (BALs), a seemingly ...
"With Block Access List, we are getting all the state that changes from transaction to transaction, and you are putting that information in the block... the biggest bottleneck we have [is sequential disk reads]." — Gabriel Trintinalia, Senior Blockchain Engineer, Consensys (Besu)
Ethereum's most consequential protocol change since the Merge is now nine weeks from mainnet. The Glamsterdam hard fork — targeted for May–June 2026 — enshrines Proposer-Builder Separation (ePBS) directly into Ethereum's consensus layer and introduces Block-level Access Lists (BALs), a seemingly obscure change that unlocks perfect parallel transaction execution. Together, these two EIPs address the network's most pressing structural vulnerability: two block builders, Beaverbuild and Titan, currently assemble roughly 90% of all Ethereum blocks, creating a de facto duopoly over which transactions get included and in what order.
This is not merely a technical upgrade. It is an economic redesign of Ethereum's $3–7 billion annual MEV extraction pipeline, a rearchitecting of L1 throughput from 60 million to potentially 200 million gas per block, and a precondition for the Layer 2 fee reductions that Coinbase's Base, Optimism, and Arbitrum need to compete with Solana on cost. With ETH trading at ~$2,024 and network activity at all-time highs despite a 30% price decline over six months, Glamsterdam is Ethereum's answer to the market's central question: does this network generate economic value, or merely redistribute subsidies?
Every 12 seconds, Ethereum produces a block. Today, approximately 92% of those blocks are constructed via MEV-Boost, the off-protocol relay system introduced by Flashbots after the Merge. Of those MEV-Boost blocks, 94% are dominated by just two entities: Beaverbuild and Titan Builder. At various points in 2025, a single builder controlled over 52% of all Ethereum blocks.
This concentration is not accidental — it is economically rational. Block builders with the largest order flow can extract the most MEV (Maximal Extractable Value), enabling them to pay proposers the highest bids and win more block auctions. The result is a self-reinforcing flywheel: more flow → more MEV → higher bids → more blocks → even more flow. Top searchers captured 90% of all MEV from 2022–2024, totaling 526,207 ETH.
The consequences are severe. Centralized builders create single points of failure for transaction censorship. When bloXroute Labs (holding 45.6% relay market share at its peak) chose to comply with OFAC sanctions, entire categories of transactions became excludable from Ethereum blocks. A network that bills itself as "credibly neutral" cannot afford a supply chain where two companies decide what gets included.
EIP-7732 — Enshrined Proposer-Builder Separation — is Glamsterdam's headline feature. It does what the name suggests: it moves the proposer-builder relationship from an off-chain relay marketplace into the Ethereum consensus protocol itself.
How it works today (MEV-Boost): Proposers → trust Flashbots relay → relay connects to builders → builder assembles block → relay verifies → proposer signs blind
How it works after ePBS: Proposers → protocol-native auction → builder commits cryptographically sealed block → proposer selects highest bid without seeing contents → transactions revealed only after finalization
The critical difference: no more relays. Currently, the entire system depends on relay operators (Flashbots, bloXroute, Ultrasound, etc.) to honestly mediate between proposers and builders. These relays are uncompensated public goods with no protocol-level incentive, which is why they've been consolidating and disappearing. ePBS eliminates this dependency entirely.
Under ePBS, builder commitments are enforced at the consensus layer. A builder who wins the auction but fails to deliver a valid block is slashed. A proposer who attempts to peek inside a sealed block gains nothing. The trust assumptions collapse from "trust the relay operator" to "trust the protocol math."
Ethereum Foundation researcher Justin Drake has estimated that approximately 10% of validators will switch to ZK verification after ePBS implementation, further decentralizing the verification pipeline.
EIP-7928 is the less discussed but arguably more transformative of the two headliners. Block-level Access Lists (BALs) introduce a new field in the block header — block_access_list_hash — containing the Keccak-256 hash of every account touched, storage key modified, balance changed, and nonce updated during block execution.
Why this matters: Ethereum's EVM currently processes transactions sequentially. One transaction finishes, the next begins. This is the computational equivalent of a single-lane highway. BALs provide a complete map of every state access in advance, enabling client software to parallelize execution across multiple CPU cores.
As Trintinalia explained, this removes "the biggest bottleneck we have" — sequential disk reads that force validators to process transactions one at a time regardless of available computing power.
Performance implications:
The BALs devnet (bals-devnet-2) is already under stress testing as of Q1 2026, while the more complex ePBS devnet (epbs-devnet-0) is in early stages.
The parallel execution capability unlocked by BALs feeds directly into Ethereum's gas limit expansion plans. Gary Schulte, Senior Staff Blockchain Protocol Engineer on the Besu client, projected: "I think in 2026, I would expect to see 100 million fairly soon. Anything beyond that is probably just too speculative to consider."
Others are more ambitious. Tomasz Stańczak, Co-director of the Ethereum Foundation, has predicted the gas limit will increase to 100 million in H1 2026, then double to 200 million after ePBS is stable, with 300 million possible before year-end.
The math is straightforward:
This would represent Ethereum's largest single capacity expansion since the Dencun upgrade introduced blob transactions in March 2024.
Glamsterdam's blob expansion — increasing the target from 3 to 6 blobs per block with a maximum of 9 — directly impacts the cost structure of every rollup settling on Ethereum. Research projections suggest a 60–95% reduction in average L2 transaction fees as rollups gain more data availability space and avoid the bidding wars that spike blob fees during congestion.
For Coinbase's Base, which extracts all L2 revenue and is already profitable, cheaper blob costs translate directly to higher margins or lower user fees — competitive ammunition against Solana's sub-cent transactions. For Optimism's Superchain architecture, which captures fee portions from 40+ L2s, the calculus is similar: more blob space means more throughput at lower per-unit cost across the entire chain family.
The economic logic is clear: Ethereum's L1 has never been the user-facing product. Its role is to be the cheapest, most secure settlement and data availability layer for L2s. Glamsterdam makes that value proposition 2–3x more compelling.
On March 2, 2026, Vitalik Buterin published a detailed roadmap for what comes after Glamsterdam. His central argument: ePBS solves the proposer-builder trust problem but does not solve block builder centralization itself. A world where two builders still assemble 90% of blocks remains vulnerable — just to different attacks.
His proposed solutions, targeted for the Hegota upgrade in late 2026:
FOCIL (Fork-Choice Enforced Inclusion Lists): Sixteen validators per slot, chosen at random, each specify transactions that must appear in the next block. Blocks missing these transactions are rejected by the fork-choice rule. Even if 100% of block building were captured by a single hostile actor, they could not exclude valid transactions because FOCIL validators enforce inclusion independently.
Big FOCIL: An expansion where FOCIL lists grow large enough to include all transactions in a block, reducing the builder's role to mere ordering of MEV-relevant transactions rather than full block construction.
Encrypted Mempools: Transactions are encrypted before entering the mempool, preventing builders and searchers from seeing transaction contents until inclusion. This directly targets "toxic MEV" — sandwich attacks and frontrunning — which constituted $289.76 million (51.56%) of the $561.92 million in measured MEV volume during a 2025 study period.
Network-layer Privacy: Anonymization of the "transaction ingress layer" — the path between a user sending a transaction and that transaction reaching the mempool — using tools like Tor and Ethereum-focused mixnets like Flashnet.
Through the lens of webthreepedia's economic value framework, Glamsterdam presents a rare case where protocol-level changes directly address the subsidy dependency that defines most blockchain networks.
Ethereum's L1 currently generates approximately $65 million in annual fee revenue against $4–5 billion in staking inflation subsidies — a ratio where user-generated revenue covers barely 1.5% of network security costs. Glamsterdam does not fix this ratio overnight, but it attacks the problem from two angles:
The upgrade also reduces the hidden "MEV tax" that users pay on every DeFi transaction. With $3–7 billion in annual MEV extraction flowing to searchers, builders, and relay operators, Glamsterdam's ePBS represents the first protocol-level attempt to restructure these flows — making them transparent, competitive, and eventually redistributable via mechanisms like MEV-Share.
Glamsterdam is Ethereum's most important upgrade since the Merge because it addresses the network's structural contradiction: a protocol designed for decentralization that has allowed its block production pipeline to centralize into a duopoly. ePBS doesn't just fix a technical problem — it rewires the economic incentives that allowed centralization to happen.
The parallel execution unlock via BALs is equally significant. For three years, Ethereum's L1 throughput has been artificially constrained by sequential processing. BALs remove that constraint, and the cascading effect — higher gas limits, cheaper L2 settlement, more transactions generating more fees — represents the most direct pathway to narrowing Ethereum's sustainability gap between $65 million in fee revenue and $4–5 billion in inflation subsidies.
The risk, as always, is execution. ePBS is the most complex consensus-layer change since the Merge. The ePBS devnet is in its earliest stages (epbs-devnet-0), and the community's aspirational June timeline leaves little room for the kind of delays that have historically plagued Ethereum upgrades. If Glamsterdam slips to Q3 or Q4, the competitive window narrows as Solana, Monad, and other high-throughput chains continue capturing mindshare.
But if it ships on time, Glamsterdam marks the moment Ethereum stops outsourcing its most critical infrastructure to off-chain intermediaries — and starts building the economic plumbing worthy of a $233 billion network.