Ethereum's Glamsterdam Hard Fork Explained: How Parallel Execution and ePBS Target 10,000 TPS
Right now, two block builders assemble more than 90% of every Ethereum block. Every transaction waits in a single-file line, no matter how many CPU cores a validator has. And gas prices still reflect benchmarks set years ago on hardware that no longer exists.
Glamsterdam, Ethereum's next hard fork targeting the first half of 2026, is designed to dismantle all three problems at once. With a gas-limit jump from 60 million to 200 million, a new parallel-execution primitive, and proposer-builder separation baked directly into the consensus layer, the upgrade represents the most aggressive structural overhaul since The Merge. If it ships on schedule, Ethereum's Layer 1 could process roughly 10,000 transactions per second — about ten times today's throughput — while cutting gas fees by nearly 79%.
Here is what is actually changing, why it matters, and where the risks hide.
Why Ethereum Needed a Structural Reset
Ethereum has spent 2025 watching competitors eat its lunch on raw speed. Solana processes upwards of 1,000 real-world transactions per second. Monad, an EVM-compatible newcomer, has demonstrated 10,000+ TPS with sub-half-second blocks. Meanwhile, Ethereum's Layer 1 sits between 15 and 20 TPS —a number that has barely budged since the Shanghai upgrade.
The performance gap created a visible migration pattern. Developers building latency-sensitive applications —gaming, high-frequency DeFi, AI agent transactions —increasingly defaulted to faster chains. Ethereum's response until now has been to push throughput onto Layer 2s (Base, Arbitrum, Optimism), but that strategy introduced its own fragmentation problems: liquidity split across 60+ rollups, seven-day withdrawal delays, and user experiences that require bridge literacy.
Glamsterdam does not abandon the L2 roadmap. It makes L1 itself fast enough to reclaim the workloads that leaked to competitors, while simultaneously improving the economics for L2 settlement. Vitalik Buterin outlined eight Ethereum Improvement Proposals in late February 2026 that define the upgrade's scope, with two "headliner" EIPs receiving top priority.
The Two Headliners: Parallel Execution and Protocol-Level Block Building
EIP-7928: Block-Level Access Lists
Today, Ethereum clients discover which accounts and storage slots a block touches only during execution. There is zero advance visibility into a block's state-access pattern. Transactions are processed one by one in strict sequence, even when they involve entirely unrelated parts of the state.
EIP-7928 introduces Block-Level Access Lists (BALs) — structured declarations specifying exactly 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.
The practical impact is transformational. When a validator knows in advance that one transaction is a Uniswap swap and another is an Aave deposit into a separate pool, it can process them simultaneously across multiple CPU cores. Sequential execution becomes a dependency graph, where only genuinely conflicting transactions must wait for each other.
BALs also enable state prefetching: nodes can load the required state data into memory before execution even begins, eliminating I/O bottlenecks that currently slow block processing. The Ethereum Foundation estimates BALs could deliver 10-30x improvements in execution throughput in the near term.
EIP-7732: Enshrined Proposer-Builder Separation (ePBS)
The block-building problem is harder to see but equally corrosive. Today, 80-90% of Ethereum block production depends on off-chain services called relays —primarily Flashbots MEV-Boost. A handful of specialized builders (Flashbots, Titan, BeaverBuild) assemble blocks for validators, extracting Maximal Extractable Value (MEV) through transaction ordering.
This arrangement works, but it creates three structural risks:
- Centralization. Two to three builders control virtually all block production.
- Censorship vulnerability. Roughly 30% of blocks currently comply with OFAC sanctions lists through relay filtering, meaning transactions can be systematically excluded by a small group.
- Trust dependency. Validators must trust that relays will not tamper with blocks or withhold payments.
EIP-7732 moves the entire proposer-builder mechanism on-chain. Under ePBS, builders assemble blocks and cryptographically seal their contents, publishing bids with payload commitments. Proposers choose the highest-paying block without being able to see or tamper with what is inside. Transactions are only revealed after the block is finalized.
A new Payload Timeliness Committee (PTC) handles basic validation checks, and all bids, commitments, and payments flow through protocol-defined messages verifiable on chain. No more off-chain relays. No more trust assumptions.
Flashbots has already anticipated this shift, migrating its operations to BuilderNet in December 2024 — a decentralized building network jointly operated with BeaverBuild and Nethermind. Glamsterdam's ePBS makes the decentralized approach the default for every builder and proposer on the network.
The Supporting Cast: Gas Repricing and State Economics
Beyond the headliners, Glamsterdam includes several EIPs that restructure Ethereum's economic model:
EIP-7904: General Repricing 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 recalibration delivers a 78.6% reduction in gas fees for both simple ETH transfers and complex smart contract interactions.
EIP-8037: State Creation Gas Cost Increase raises the gas cost for creating new accounts and storage slots, discouraging unnecessary state expansion that bloats the chain.
EIP-8038: State-Access Gas Cost Increase raises gas costs for cold account and storage reads, better reflecting the I/O cost of accessing infrequently used state.
The net effect is a gas model that charges more for what is genuinely expensive (creating and accessing cold state) and less for what is cheap (executing common opcodes on modern processors). The gas limit increase from 60 million to 200 million per block provides the raw capacity to absorb these changes while still delivering dramatically lower per-transaction costs.
How Glamsterdam Compares to the Competition
The parallel-execution race is not happening in isolation. Here is how Glamsterdam stacks up:
| Feature | Ethereum (Glamsterdam) | Solana | Monad |
|---|---|---|---|
| Parallelism approach | Pre-declared access lists (BALs) | Native Sealevel scheduler | Optimistic parallel execution |
| Target TPS | ~10,000 | ~1,000-1,500 (real-world) | ~10,000+ |
| EVM compatibility | Native | No (requires Rust rewrites) | Full EVM compatible |
| Block time | ~12 seconds | ~400ms | ~400ms |
| MEV management | Protocol-level ePBS | Jito tip marketplace | Inherited from EVM ecosystem |
| Validator set | ~500,000+ | ~1,500 | Emerging |
Ethereum's approach is distinctly conservative: transactions must declare their state access upfront, which guarantees conflict-free parallel execution but requires tooling changes from developers. Solana's Sealevel runtime handles parallelism natively at the VM level. Monad uses optimistic execution —processing transactions in parallel and re-executing on conflict —which preserves full Ethereum compatibility without developer changes.
The critical difference is ecosystem gravity. Ethereum has 500,000+ validators, the deepest DeFi liquidity, and the largest developer community. Even if Monad delivers equivalent throughput, it must bootstrap everything else from scratch. Glamsterdam's bet is that making the incumbent fast enough removes the migration incentive.
The L2 Impact: Better Economics for Rollups
Glamsterdam's improvements do not just benefit L1 users. The gas limit increase and potential expansion to 72+ data blobs per block dramatically improve the economics for Layer 2 rollups.
Rollups like Arbitrum, Optimism, and Base currently pay Ethereum L1 gas to post transaction data. With lower gas costs and higher blob capacity, L2 settlement becomes cheaper, which translates to lower fees for end users of these chains. The Ethereum Foundation has projected that combined L1 + L2 throughput could eventually reach hundreds of thousands of TPS as blob scaling compounds with parallel execution.
This creates a virtuous cycle: cheaper L2 settlement attracts more L2 users, more L2 usage generates more L1 settlement fees, and higher L1 throughput ensures the settlement layer does not become a bottleneck.
What Could Go Wrong
Glamsterdam is ambitious, and ambition carries risks.
BAL adoption requires tooling changes. Developers and wallets must update to generate block-level access lists. Until the ecosystem tooling matures, the parallel-execution benefits may remain theoretical for some transaction types.
The timeline is aspirational. Community documentation references June 2026 as a target, but the Ethereum Foundation's DevOps team has only tested three of the proposed EIPs on Devnet-4 as of early 2026, with Devnet-5 in progress. Ethereum hard forks have historically slipped by months.
ePBS disrupts existing MEV infrastructure. Block builders and searchers have built entire businesses around the current relay-based system. The transition to ePBS will reshape the MEV supply chain, and not all participants will benefit.
Gas repricing could break assumptions. Applications that hardcoded gas estimates may behave unexpectedly when opcode costs change. While the repricing reduces most costs, the increases for state creation and cold reads could affect contracts that perform many cold storage operations.
Over 30 proposals were declined for Glamsterdam, including reduced slot times, multidimensional gas metering, and post-quantum signature verification. These features are expected to land in Hegota, the second 2026 fork scheduled for the second half of the year.
What Comes After: The Four-Fork Strategy
Glamsterdam is the first move in a coordinated four-fork performance strategy extending through 2027:
- Fast Confirmation Rule —13-second deposit confirmations without a hard fork (already live or imminent)
- Glamsterdam (H1 2026) —Parallel execution + ePBS + gas repricing
- Fusaka —PeerDAS for 1 GB/s data availability with 30,000x capacity increase
- Hegota (H2 2026) —Verkle Trees for 10x smaller proofs + post-quantum cryptography migration
Together, these upgrades transform Ethereum from a single-lane road into a multi-lane highway with on-ramps optimized for every type of traffic. The question is no longer whether Ethereum can match competitor speed, but whether it can ship these upgrades before the performance-sensitive use cases permanently settle elsewhere.
The Bottom Line
Glamsterdam is not a marketing exercise. It is a structural response to the existential threat that Ethereum's Layer 1 becomes irrelevant as a general-purpose execution environment while competitors capture the workloads that matter most: gaming, payments, AI agent transactions, and real-time DeFi.
The upgrade introduces genuine architectural innovations —BALs for parallel execution, ePBS for decentralized block building, and empirical gas repricing —that address problems Ethereum has tolerated for years. If it ships on schedule, the network moves from 15 TPS to 10,000 TPS while becoming more decentralized, not less.
The catch is that "on schedule" is doing heavy lifting in that sentence. Ethereum's history suggests caution about timelines. But the engineering work is real, the devnets are running, and the competitive pressure has never been higher.
For builders, validators, and users watching from the sidelines: Glamsterdam is worth watching closely. It is either the upgrade that secures Ethereum's position for the next decade —or the one where the gap between ambition and execution finally catches up.
BlockEden.xyz provides enterprise-grade Ethereum RPC endpoints and multi-chain API infrastructure for developers building on Ethereum L1 and the rollups that settle to it. As Glamsterdam reshapes the execution layer, reliable node infrastructure becomes more critical than ever. Explore our Ethereum API services to build on foundations designed for the next generation of on-chain throughput.