Ethereum's Glamsterdam upgrade — the most structurally ambitious hard fork since the Merge — reached its first generalized devnet the week of April 14, 2026, according to the Ethereum Foundation's Checkpoint #9 published April 10. The upgrade enshrines proposer-builder separation (ePBS) into the ...
"ePBS is proving trickier than anticipated. Every part of the stack has to reason about partial blocks and two-party coordination — a change that touches practically everything." — Nixo, Ethereum Foundation Protocol Support, Checkpoint #9 (April 10, 2026)
Ethereum's Glamsterdam upgrade — the most structurally ambitious hard fork since the Merge — reached its first generalized devnet the week of April 14, 2026, according to the Ethereum Foundation's Checkpoint #9 published April 10. The upgrade enshrines proposer-builder separation (ePBS) into the protocol, raises the gas limit from 60 million to 200 million, introduces block-level access lists (BALs), and targets a 78.6% reduction in gas fees alongside a tenfold throughput increase to approximately 10,000 transactions per second.
However, the Ethereum Foundation now describes a Q2 2026 mainnet activation as "unlikely." The ePBS devnet required stabilization before the first generalized devnet could proceed, and developers acknowledge the complexity of coordinating "partial blocks" across the entire protocol stack. A slip into Q3 or Q4 2026 is the working assumption among core developers. The stakes are high: Ethereum's market dominance fell to 10.4% in March 2026, its lowest since mid-2021, while L1 gas sits at 0.07 gwei — so low that the protocol is generating negligible fee revenue. Glamsterdam is not just a technical upgrade. It is Ethereum's answer to a structural economic problem.
Glamsterdam combines a consensus-layer overhaul (Gloas) with an execution-layer upgrade (Amsterdam). The two headliner EIPs are:
Beyond the headliners, the Ethereum Foundation's Checkpoint #9 lists over 25 non-headliner EIPs under consideration, including EIP-7954 (increased maximum contract size, prioritized due to ecosystem demand) and EIP-8007 (bundled gas repricings).
Today, 85-90% of Ethereum blocks are produced through external MEV-Boost relays operated by a handful of entities. According to Flashbots data, the top five builders construct approximately 56% of all MEV-Boost blocks and are responsible for a proportional share of execution-layer rewards. Three of the four largest builders also operate their own relays — Flashbots, Bloxroute (three relays), and builder0x69 (Relayooor) — creating a vertically integrated supply chain that sits outside the protocol.
This architecture introduces several risks:
ePBS eliminates relays by encoding the builder-proposer auction directly into the consensus layer. Builders submit sealed bids on-chain; proposers commit to the highest bid without seeing block contents; builders then reveal the full block. The protocol handles disputes and fallbacks natively.
Between December 2025 and January 2026, searchers extracted approximately $24 million in MEV profit over a 30-day period, according to Flashbots data. By March 2026, Flashbots Protect had saved users over 4,600 ETH in MEV costs and 2,200 ETH in gas fees. ePBS does not eliminate MEV — extraction will continue — but it removes the trusted intermediary layer and subjects the auction to protocol-level enforcement, reducing the surface area for opaque value capture.
The challenge, as Checkpoint #9 documents, is that "every part of the stack has to reason about partial blocks and two-party coordination." The protocol must now handle scenarios where a builder submits a bid but fails to reveal a valid block, requiring fallback logic that did not previously exist. This is the primary source of the timeline uncertainty.
EIP-7928 introduces a requirement that each transaction pre-declare its state access patterns. This is a prerequisite for parallel execution: if the EVM knows in advance that Transaction A reads storage slot X and Transaction B reads storage slot Y, it can process both simultaneously.
Current Ethereum execution is strictly sequential. Every transaction in a block is processed one after another, regardless of whether they touch overlapping state. This limits throughput to roughly 1,000 TPS under optimal conditions with the current 60 million gas limit.
BALs, combined with the gas limit increase to 200 million, create the conditions for the targeted 10,000 TPS. The BAL devnets are described as "making predictable progress" in Checkpoint #9, though the Ethereum Foundation characterizes the work as a "fundamental rethinking of how gas and state access work."
EIP-7904 recalibrates gas costs for EVM opcodes using empirical benchmarks from modern hardware. Many current gas prices were set years ago and no longer reflect actual computational cost. The repricing delivers a projected 78.6% reduction in gas fees for both simple ETH transfers and complex smart contract interactions.
Current L1 gas prices tell the story of Ethereum's fee-revenue problem. As of April 10, 2026, Etherscan's gas tracker shows an average gas price of 0.071 gwei. At ETH trading near $1,800, a simple transfer costs approximately $0.006. Ethereum's L1 is already cheap. The question Glamsterdam answers is not "how do we make it cheaper" but "how do we make it capable of handling enough volume to generate meaningful fee revenue at these prices."
At 10,000 TPS and 200 million gas per block, the protocol can process dramatically more transactions per block. Even at sub-cent fees per transaction, aggregate revenue scales with volume. The gas repricing is therefore less about user cost reduction (fees are already negligible) and more about removing computational bottlenecks that currently cap throughput.
According to Checkpoint #9, published April 10, 2026:
The Ethereum Foundation's process overview references All Core Developers calls ACDT 67-76, ACDC 173-176, and ACDE 229-234 spanning January 21 to April 9, 2026 — a 12-week period of intensive coordination.
Two factors create delay risk beyond the ePBS complexity:
FOCIL has since been deferred. It will serve as the consensus-layer headliner for the subsequent Hegotá upgrade, reducing Glamsterdam's scope.
Ethereum's economic position in April 2026:
| Metric | Value | Source | |---|---|---| | Market dominance | 10.4% (three-year low) | CoinGecko, March 2026 | | ETH/BTC ratio | ~0.030 (five-year low) | TradingView | | ETH price | ~$1,800 | CoinGecko, April 2026 | | Average L1 gas | 0.071 gwei | Etherscan, April 10, 2026 | | Simple transfer cost | ~$0.006 | Calculated | | Total staked ETH | 35.86M ETH (~28.9% of supply) | Datawallet | | Active validators | ~919,000 | Beaconcha.in | | DeFi TVL (Ethereum) | $54.1B | DefiLlama | | BTC dominance | 57.2% | CoinGecko, April 11, 2026 |
Ethereum's fee revenue problem is structural. The Dencun upgrade (March 2024) and Fusaka (December 2025) reduced L2 data posting costs so effectively that L1 gas demand collapsed. Blobs are frequently underpriced. The protocol's economic model — which relies on fee burn (EIP-1559) to create deflationary pressure on ETH supply — requires transaction volume that the current L1 throughput ceiling cannot support at sub-cent gas prices.
Glamsterdam's throughput expansion is the intended fix. By processing 10x more transactions per block, the protocol can generate aggregate fee revenue even at low per-transaction costs. Whether sufficient demand materializes to fill 200 million gas blocks is an open question. If L2s have permanently absorbed the marginal user, L1 throughput expansion may produce capacity without utilization.
Glamsterdam creates a dual dynamic for L2 ecosystems:
Cost reduction for rollups. The gas limit increase and repricing reduce the cost of posting calldata and blobs to L1. Projections suggest a 60-95% reduction in average L2 settlement costs, depending on blob market conditions. This directly improves margins for rollup operators and reduces fees for end users on Base, Arbitrum, Optimism, and others.
Competitive pressure from L1. If Glamsterdam delivers sub-cent L1 transactions at 10,000 TPS, the value proposition of L2s narrows. Activity that migrated to rollups specifically for cost reasons may partially return to L1, which offers a higher security guarantee. The extent of this reversion depends on whether L1 can match L2 user experience — particularly confirmation times and application-specific customization.
The Ethereum Foundation's economic framework positions L1 as the settlement and security layer, with L2s providing execution scalability. Glamsterdam's throughput increase complicates this division of labor.
Checkpoint #9 confirms that the Hegotá upgrade (following Glamsterdam, targeted late 2026) has selected its consensus-layer headliner: FOCIL (EIP-7805) — Fork-Choice Inclusion Lists. FOCIL requires proposers to include a minimum set of pending transactions, making censorship-resistant transaction inclusion a protocol-level guarantee.
Non-headlining features for Hegotá include:
Non-headlining EIP submissions for Hegotá opened on April 9, 2026. The closing date will be announced with a minimum two-week notice period.
Glamsterdam is Ethereum's attempt to resolve a contradiction: a network with $54 billion in DeFi TVL and 919,000 validators that generates negligible L1 fee revenue. The upgrade's two-pronged approach — eliminating relay centralization through ePBS and unlocking parallel execution through BALs — addresses both the structural and throughput dimensions of this problem.
The April 10 Checkpoint #9 confirms that development is progressing but slower than the post-Fusaka optimism suggested. The ePBS implementation complexity, the number of non-headliner EIPs, and the need for rigorous security review across what the Foundation describes as changes that "touch practically everything" all point toward a second-half 2026 activation.
Whether the upgrade reverses Ethereum's declining market share depends on a question the protocol cannot answer alone: will demand follow capacity? At 10,000 TPS and sub-cent fees, Ethereum L1 would be cost-competitive with most L2s. But if applications and users have structurally migrated to rollups, the additional capacity may sit unused — a faster highway with the same traffic. The market will settle this debate. The protocol's job is to build the road.