Summary
The 2026 Ethereum protocol roadmap organizes work into three tracks (Scale, Improve UX, Harden the L1) and two hard forks: Glamsterdam (H1 2026) focused on throughput, and Hegotá (H2 2026) focused on privacy, account abstraction, and censorship resistance. The EF’s Strawmap document (Feb 2026) projects 7 hard forks through 2029 with five North Stars: fast L1, gigagas L1, teragas L2, post-quantum L1, and private L1.
Three Protocol Tracks
| Track | Leads | Focus |
|---|---|---|
| Scale | Ansgar Dietrichs, Marius van der Wijden, Raúl Jordan | Throughput, gas limits, DA |
| Improve UX | Barnabé Monnot, Matt Garnett | Account abstraction, wallets, fast confirmations |
| Harden the L1 | Thomas Thiery (censorship), Fredrik Svantes (security), Parithosh Jayanthi (infrastructure) | Censorship resistance, privacy, security, PQ readiness |
The Hardness track (2026-03-18) was added explicitly to ensure scaling and UX improvements don’t sacrifice Ethereum’s core properties: open source, censorship resistance, privacy, security, permissionlessness, trust minimization.
Glamsterdam (H1 2026)
Theme: Scale and performance.
Key EIPs:
- EIP-7732 — ePBS (Execution Payload Separation): builder-proposer separation at the protocol level; relays become optional.
- EIP-7928 — Block-Level Access Lists (BALs): headliner for parallel transaction execution and I/O prefetching. Builder attaches read/write sets for all transactions; verifiers can execute in parallel.
- Parallel execution support
- Higher gas limits (targeting ~60M+ gas, up from current levels)
- Blob parameter increases (BPO continuation from Fusaka)
BAL mechanics: The block builder executes transactions sequentially, records all state reads/writes, and attaches the BAL to the block. Validators use the BAL to execute transactions in parallel (up to 4 batches for a typical 161-tx block). I/O prefetching eliminates disk reads during execution. BALs are stored for ~2 weeks then prunable; the BAL hash is permanent in the block header.
Hegotá (H2 2026)
Theme: Privacy, account abstraction, censorship resistance.
Key EIPs:
- EIP-7805 — FOCIL (Fork-Choice Enforced Inclusion Lists): validators randomly selected for 17-member inclusion list committee; builders must include their transactions or blocks fail attestation.
- EIP-8141 — Frame Transactions: native AA with VERIFY/SENDER/DEFAULT/PAY frame model; default account for EOA compatibility.
- LUCID encrypted mempool: threshold-encrypted transactions in the public mempool, decrypted N+1 slot (or same-slot with economic commitment).
- PQ readiness groundwork (Falcon/Dilithium signatures, leanMultisig for finality layer).
Strawmap (Feb 2026)
The EF Protocol team’s 7-fork roadmap through 2029. Five North Stars:
| North Star | Description | Key dependency |
|---|---|---|
| Fast L1 | Slot time reduction; sub-second finality | Goldfish/Minimmit consensus |
| Gigagas L1 | 1 gigagas/sec (~10,000 TPS) via zkEVM + real-time proving | ZK stack maturity |
| Teragas L2 | 1GB/sec (~10M TPS) via full DAS | PeerDAS → Full DAS |
| Post-Quantum L1 | Hash-based signatures throughout | leanMultisig; Falcon aggregation |
| Private L1 | Privacy transfers at L1 level | Encrypted mempool; anonymous broadcast |
Slot time reduction path: 12s → 8s → 6s → 4s → 3s → 2s. Each reduction increases pressure on non-Western validators (shorter propagation windows). Concurrent zkAttester transition reduces hardware requirements to partially offset this.
Consensus Evolution
Beyond the immediate roadmap:
- Decoupled consensus (Goldfish + Majorum): dynamically available heartbeat (~256 validators/slot) + trailing finality gadget. Eliminates Gasper’s complexity; enables post-quantum heartbeat without full-validator-set aggregation.
- Dynamic availability (Luca Zanolini, Mar 2026): the heartbeat layer must be dynamically available (safe and live as long as a majority of currently awake stake is honest). Required because Ethereum has never gone offline even during client bugs.
- Leaderless BFT (longer term): eliminates the leader/proposer concept entirely; removes PBS motivation at the consensus layer.
CROPS Mandate
The Ethereum Foundation’s mission for 2026 is defined by CROPS:
- Censorship Resistance
- Responsibility (open source)
- Open Source
- Privacy
- Security
CROPS is explicitly non-negotiable as performance and UX improvements ship — each improvement must preserve all five properties.
EF Platform Team (Parallel Track)
Separate from protocol R&D, the EF Platform team was announced to work on:
- Native rollup architecture (L2s directly using L1 execution)
- L1 as settlement hub; L2 for differentiation
- Cross-chain composability (EEZ framework with Gnosis/Zisk)
- Reducing the current tooling/coordination friction between L1 and L2
Geographic Implications
The roadmap has significant geographic infrastructure implications:
- FOCIL’s censorship resistance depends on 17-member committees being geographically distributed across jurisdictions.
- PeerDAS (Fusaka, Dec 2025) requires 128 data column subnets distributed globally.
- Slot time reduction amplifies the GossipSub mesh latency penalty for non-Western validators.
- ePBS (Glamsterdam) embeds relay functionality in the protocol, potentially reducing relay-location advantages.
- Europe + North America currently host ~78.7% of validators (Lido Q4/2025 data).
Related Pages
- ePBS: Enshrined Proposer-Builder Separation (EIP-7732) — ePBS (EIP-7732) detailed
- Frame Transactions and Native Account Abstraction (EIP-8141) — Frame Transactions / EIP-8141 detailed
- FOCIL: Fork-Choice Enforced Inclusion Lists (EIP-7805) — FOCIL / EIP-7805 detailed
- Encrypted Mempools — LUCID encrypted mempool
- Block-Level Access Lists (BALs) and Parallel Execution (EIP-7928) — BALs / EIP-7928
- Decoupled Consensus: Goldfish, Majorum, and Dynamic Availability — Goldfish/Majorum/dynamic availability
Key Sources
- Protocol Priorities Update for 2026 (EF, 2026) — three tracks; EIP list per fork
- A Deeper Look at Priority: Hardness (Fredrik, Thomas, Parithosh, Mar 2026) — CROPS mandate; hardness track scope
- Ethereum’s Geographic Blind Spot (Four Pillars, Mar 2026) — Strawmap summary; node distribution data; geographic implications