Ethereum's next hard fork, Glamsterdam, targets a throughput increase from 15–20 TPS to approximately 10,000 TPS through parallel execution and a gas limit raise from 60 million to 200 million per block. Two headliner EIPs — EIP-7732 (enshrined Proposer-Builder Separation) and EIP-7928 (Block-Lev...
"If a transaction is encrypted until it's included, no one gets the opportunity to wrap it in a hostile way." — Vitalik Buterin, Co-Founder, Ethereum
Ethereum's next hard fork, Glamsterdam, targets a throughput increase from 15–20 TPS to approximately 10,000 TPS through parallel execution and a gas limit raise from 60 million to 200 million per block. Two headliner EIPs — EIP-7732 (enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists) — form the core of the upgrade, which is nominally scheduled for H1 2026 but showing signs of slipping toward Q3.
As of the Ethereum Foundation's Checkpoint #9 blog post on April 10, 2026, only three of the proposed EIPs have been tested on Devnet-4. Devnet-5 is in progress, with developers aiming for the first generalized Glamsterdam devnet in the coming weeks — contingent on stabilizing the ePBS devnet. Meanwhile, Solana's Firedancer validator client, live on mainnet since late 2025, has reached roughly 25% validator adoption and sustains 1,000–1,500 real-world TPS with a theoretical ceiling north of one million TPS in controlled tests. The two networks are converging on comparable real-world throughput from opposite engineering philosophies: Ethereum retrofitting parallelism onto a sequential architecture, Solana hardening a natively parallel one.
The economic stakes are material. ETH trades at approximately $2,320 with a market capitalization of $280 billion. Two to three block builders currently assemble roughly 90% of all Ethereum blocks via off-chain MEV-Boost relays — a centralization vector that Glamsterdam aims to dismantle by moving the builder market on-chain. Whether ePBS actually reduces builder concentration or merely enshrines it is an open question, with a January 2026 academic paper suggesting the latter.
Ethereum's Layer 1 processes 15–20 transactions per second. Solana's mainnet, now running a mix of Agave (the legacy client) and Jump Crypto's Firedancer, sustains 1,000–1,500 TPS in production, with a peak of 6,284 TPS recorded by Chainspect. On January 30, 2026, Solana logged 148 million non-vote transactions in a single day, according to Solana Compass.
The gap is roughly two orders of magnitude. Glamsterdam's stated target of 10,000 TPS, if achieved, would put Ethereum L1 on par with or ahead of Solana's sustained real-world throughput — though well below Solana's theoretical maximum.
Block time remains a structural difference. Ethereum produces blocks every ~12 seconds; Solana's slot time is approximately 400 milliseconds. No EIP in Glamsterdam addresses block time reduction. Ethereum's approach relies on packing more computation into each block rather than producing blocks faster.
| Metric | Ethereum (Current) | Ethereum (Post-Glamsterdam Target) | Solana (Current) | |---|---|---|---| | Real-world TPS | 15–20 | ~10,000 | 1,000–1,500 | | Block time | ~12 sec | ~12 sec | ~400 ms | | Gas limit / block capacity | 60M gas | 200M gas | N/A (compute units) | | Validator count | ~500,000 | ~500,000 | ~1,500 | | Block builder concentration | 90% by 2–3 entities | TBD | Jito marketplace |
EIP-7732: Enshrined Proposer-Builder Separation (ePBS)
Currently, 80–90% of Ethereum block production depends on off-chain MEV-Boost relays operated by Flashbots and similar entities. EIP-7732 moves this mechanism into the protocol itself. Under ePBS, builders assemble blocks, cryptographically seal their contents, and publish bids with payload commitments. Validators select the highest bid without viewing transaction contents before locking in. A newly introduced Payload Timeliness Committee (PTC) validates that builders deliver blocks on time.
The fundamental engineering challenge, as described in the Ethereum Foundation's Checkpoint #9: ePBS "splits block production into two parties acting in sequence inside consensus instead of the way it currently happens outside of the protocol, so the protocol now has to handle disagreement or failure between them." Every part of the stack must reason about "partial blocks" and two-party coordination.
EIP-7928: Block-Level Access Lists (BALs)
BALs are structured declarations specifying which storage slots and accounts each transaction will read or write. A hash of this access list — the "BAL root" — is embedded directly in the block header. This enables nodes to identify non-interfering transactions and process them simultaneously across multiple CPU cores.
Estimated execution throughput improvement: 10–30x, according to BlockEden.xyz analysis. The practical benefit depends heavily on tooling adoption. Wallets, dApps, and smart contracts must generate accurate access declarations for the parallelism to function. Incorrectly declared lists would force sequential fallback processing.
Glamsterdam includes multiple gas repricing EIPs beyond the headliners:
Tomasz Stańczak, co-director of the Ethereum Foundation, stated at the Bankless Summit that the gas limit would increase to 100 million in the first phase and double to 200 million once ePBS is fully operational. He indicated the limit could reach 300 million by year-end, contingent on network stability.
The combined effect of higher gas limits and repriced opcodes targets a 78.6% reduction in gas fees for complex smart contract interactions, according to analysis by Bitfinex. Simple transfers would see smaller but still material reductions.
Jump Crypto's Firedancer validator client, which launched on Solana mainnet after three years of development, represents the competitive context against which Glamsterdam must be measured.
As of late March 2026, approximately 25% of Solana validators by stake weight run Firedancer or its hybrid variant (Frankendancer), up from ~8% in June 2025 and ~21% in October 2025, per OpenPR data. The client uses a tile-based modular architecture where consensus, transaction processing, and networking run in parallel across CPU cores — an approach fundamentally different from Ethereum's single-threaded EVM with bolt-on parallelism.
Kevin Bowers, Chief Scientist at Jump Trading Group, demonstrated over one million TPS on commodity hardware at Breakpoint 2024. Production performance remains well below that ceiling. The practical gap between testnet demonstrations and mainnet reality is substantial across both ecosystems.
The architectural differences are worth noting: Solana's Sealevel runtime was designed for parallel execution from inception. Ethereum is retrofitting parallelism via BALs onto a fundamentally sequential virtual machine. The latter approach carries higher implementation risk but affects a network with over 500,000 validators versus Solana's approximately 1,500.
Glamsterdam's ePBS addresses a problem that has intensified since the Merge. Currently, two to three specialized builders — including Flashbots, Titan, and BeaverBuild — assemble approximately 90% of all Ethereum blocks. They extract Maximal Extractable Value (MEV) through transaction ordering, with the proceeds shared partially with validators via MEV-Boost bids.
A January 2026 academic paper modeling ePBS behavior in MEV-present environments found that while ePBS reduces validator-side concentration, it "significantly amplifies profit and content centralisation" among builders. Access to private order flow confers a structural bidding edge that compounds over time, the paper concluded.
Ethereum developers have proposed complementary measures:
FOCIL (Forward Obligatory Commitment to Inclusion Lists): Sixteen randomly selected attesters mandate that transactions — including potentially controversial ones — make it into blocks. Vitalik Buterin has argued that FOCIL could expand to handle all transactions in a block, relegating builders to MEV-relevant ordering only. FOCIL was initially considered for Glamsterdam but has been moved to Hegotá, the subsequent upgrade.
Encrypted mempools: Encrypting the transaction pool until inclusion would prevent sandwich attacks and frontrunning — the most extractive forms of MEV. Buterin noted this approach eliminates the ability to "wrap" a transaction in hostile orders.
BuilderNet: Flashbots' proposed multi-operator network where builders, searchers, and dApps jointly assemble blocks and share rewards proportionally to contributed value.
The "free option" problem compounds the centralization risk. According to Bitfinex analysis, builders can delay or withdraw bids in approximately 0.82% of blocks under normal conditions, rising to ~6% during high volatility — effectively gaining optionality at the validator's expense.
Glamsterdam's throughput improvements arrive against a backdrop of declining L1 revenue. L2 payments to Ethereum L1 have collapsed more than 90% year-over-year following the introduction of blob transactions in the Dencun upgrade (March 2024), which reduced L2 data posting costs by orders of magnitude.
The economic tension is structural. Ethereum's L2 ecosystem generates an estimated $161 million per year in sequencer revenue, according to analysis cited by Blockworks. None of that flows to L1 validators. If Glamsterdam successfully raises L1 throughput to 10,000 TPS, more activity may remain on L1 rather than migrating to L2s — potentially reversing the revenue leakage, but also competing directly with the rollup ecosystem that Ethereum spent years cultivating.
The Ethereum Foundation's March 2026 blog post, "How L1 and L2s Can Build the Strongest Possible Ethereum," acknowledged the tension directly, proposing based rollups and native rollups as alignment mechanisms. The forthcoming Hegotá upgrade (H2 2026) includes Account Abstraction work that could further reshape the L1–L2 economic boundary.
The Ethereum Foundation's Checkpoint #9, published April 10, 2026, characterizes Glamsterdam progress as "slow but steady." Specific status indicators:
The June 2026 target date is described as "aspirational" in community documentation. The Base engineering team publicly warned that adding FOCIL alongside ePBS could delay the upgrade beyond 2026 — a risk mitigated by FOCIL's subsequent move to Hegotá. Nonetheless, the interplay between ePBS and BALs introduces complexity untested at mainnet scale.
Ethereum hard fork timelines have historically slipped by months. A Q3 2026 mainnet activation appears more realistic than June, based on current devnet progress.
Glamsterdam is the most structurally ambitious Ethereum upgrade since the Merge. It attempts to solve three problems simultaneously: throughput (via BALs and gas limit increases), MEV centralization (via ePBS), and fee economics (via opcode repricing). The scope is defensible given the competitive pressure from Solana and emerging chains like Monad, but the engineering complexity is reflected in the devnet pace.
The economic implications extend beyond raw performance metrics. If Ethereum L1 becomes fast and cheap enough to compete with its own L2s, the value distribution across the ecosystem shifts. Validators, L2 sequencers, block builders, and MEV searchers all face altered incentive structures. The foundational question — who captures the economic value generated by on-chain activity — remains the central tension, and Glamsterdam repositions the pieces without fully resolving the game.
Market participants should track three signals: devnet stability milestones, builder concentration metrics post-ePBS testnet, and L2 fee revenue trends. These will determine whether Glamsterdam delivers a structural shift or an incremental one.