Nethermind Quarterly Update - Q1 - 2026

Aztec
1

Summary

The Nethermind team has focused on a number of workstreams over the last quarter, including Alpha work, shipping the multi-asset FPC, kicking off the Aztec Wallet, a broad working of research (GPU proving, L3 & TEE multi-proving, prover economics, PIR, personal-data marketplace), product maintenance on Gov Dashboard and Aztec nodes, devrel tooling and events!

The major focuses this quarter were:

  • Aztec Alpha Development - Worked on hardening core protocol infrastructure, security, and mainnet-alpha readiness across networking, state management, and operational reliability.
  • Research - Focused on advancing protocol research across scaling, proving infrastructure, economics, and privacy primitives.
  • Aztec Wallet kickoff - Kicked off a new desktop wallet and are racing towards an MVP.
  • FPC development - Developed a multi-asset fee payment contract as a deployable infrastructure.
  • Devrel & Tooling - Built a number of developer tools like Testnet Faucet, Aztec-lint, and Aztec-watcher.

Deliverables for Q2:

  • Engineering
    • GPU Proving - GPU proving generates Aztec proofs ~10Ă— faster than CPUs. Faster transactions for users, on cheaper hardware for provers.
      • Definition of done: Barretenberg Icicle MSM integration merged with reproducible benchmarks and documented integration path.
  • Research
    • Multi-Prover Architecture and Research - Enables additional safety and trust for adoption, and transacting with assets on Aztec.
      • Definition of done for Q2: Multiple components and long timeline to completion, depending on initial research results.
        • Architecture Design and consensus finding
        • TEE Implementation Research
        • Start coordination around implementation
  • Product
    • Wallet - General purpose Aztec desktop wallet to support adoption.
      • Definition of done: Can open wallet, and interact and transact on the Aztec Network
  • Devrel
    • Aztec Remix Plugin - A browser based Aztec development IDE that removes the toolchain setup barrier for first time builders.
      • Definition of done: Able to interact with Aztec network via Remix IDE.
    • Documentation - Efforts based on team feedback to support and update documentation.
      • Definition of Done: Ongoing!

Workstreams

Engineering

Alpha/Protocol

Major components:

  • Mempool Improvements
    • Improved the mempool with a sorted collection implementation, codified reorg rules in world state sync, and added low-priority tx eviction after pruning.
    • Implemented price-bump replacement so users can supersede stuck transactions, added rejection of txs that exceed DA or mana limits at admission, and fixed a bug where evicted txs were on node restart.
  • P2P / Networking
    • Integrated BatchTxRequester with FastTxCollection over dynamic peer sets and resolved a race in fast_txs_collection. Added smart peer-scoring penalties for failed req-resp responses.
  • Audit hardening - 34 issues
    • Covered wire-format versioning and P2P hardening, bounded deserialization for DoS resistance (15 of 22 P2P call sites were missing maxSize), PXE error-handling robustness, and crypto/timing safety (decryption timing oracle, non-constant-time comparisons, missing length prefix collision risk).
    • Also fixed gas-math precision (Number() on bigint silently truncating above 2^53), resource lifecycle (executeBB timeouts, zombie children), L1 integration (silent Multicall3 reverts), and URL / JSON-RPC / BigInt / blob / beacon validation.
  • World state fixes to be reorg safe
    • Shipped COMMIT_FORK, which fixes an archiveRoot to forkId race that was producing false reorg detection. Contract data now soft-deletes on prune and only hard-deletes at finalization.
    • Fixed a checkpoint-building race and refactored config to be type-safe.
  • Public-setup validation & allowlist
    • Re-enabled function selectors in txPublicSetupAllowList, added arg-shape validation, and tightened set_authorized and only_self checks.
  • Operational robustness
    • Fixed blob-tx simulation, added publisher rotation on failure, enabled multi-EOA publisher funding, added keystore hot reload and auto-stop signaling, and improved BlobFilestore download handling.
  • Test Clearance
    • Cleared 10 flaky tests around epochs, P2P, and consensus slashing.
  • Tooling fixes
    • Fixed a zig cc cross-compilation issue where targeting native on Graviton CI emitted SVE instructions that failed at runtime on M3.
  • Post-launch incident response
    • Supported recovery through the April 1 - 3 block-header-mismatch outages, the April 20 reorg, and the late-April MessageStoreError.

Research

  • Multi-proving Exploration
    • Ongoing exploration of different solutions to introduce a 2nd prover “redundancy layer” for Aztec’s private-state ZK.
    • We see this as a critical security upgrade for the system for driving adoption among any token-security/-recoverability conscious users, especially for any use case bridging assets into L2.
  • GPU proving
    • Explored GPU proving, to enable hardware-accelerated proving in Barretenberg, with early benchmarks and designs. Finishing results, and will be sharing soon!
  • Prover economics
    • Contributed to the community conversation on more consistent prover rewards - LINK
  • Pool FPC v2
    • Generated a spec for a self-balancing on-chain AMM model that broadens who can provide fee-payment liquidity, lowers operational overhead, and returns overpayments to users automatically.
  • Proving optimizations
    • WASM 5 - 7% mont mul speed increase (merged) - LINK
    • WASM 20% e2e proving speed increase (PR proposed). Refracted the code to compute two field multiplications together at the cost of only 1.2Ă— single multiplication, instead of 2x. - LINK
    • WebGPU WASM client-side proving - Quick investigation with of the shelf libraries, indicating as infeasible for now (~40Ă— slower than CPU for 2^18 MSM).
  • L3 Exploration
    • Building on the initial L3 sketch, we explored moving more complex functionality into the L3 while still leveraging the compression techniques enabled by proof verification using the L2 private execution.
    • Given the customizability of L3s, this path had the potential to enable a recoverability and throughput optimized instance of Aztec.
    • Although initial results were promising, we paused this exploration to focus on improving the L2.
  • Intern Exploration - Personal Data Marketplace
    • Generate a proof-of-concept personal data marketplace where users keep their data, buyers receive the answers they paid for, and Aztec guarantees the fairness of the exchange.

Product

  • Fee Payment Contracts (FPC)
    • Multi-Asset FPC to testnet with Docs! - LINK
  • Aztec Wallet
  • Aztec Oracle contracts (semi research)
    • Explored Oracle design on Aztec for Push and Pull functions (write to contract, and consumer fetch). Navigating a trilemma of privacy / freshness / contention for private oracle reads.
  • AztecNodes Maintenance
    • Added support for Mainnet and fixed a number of issues with AztecNodes.xyz to properly present mainnet information.
  • Governance dashboard
    • Built the Aztec Governance Dashboard - LINK

DevRel

  • Aztec Faucet
    • Built the Aztec Testnet Faucet - LINK
    • Handles L2 Fee Juice bridging, L1 Sepolia ETH drip, keypair gen and deploy and claim for new accounts.
  • Aztec Skills
    • Built an installer for Aztec-focused agent skills - LINK
      • Six skills across: aztec-contracts, aztec-deployment, aztec-js, aztec-testing, aztec-wallet-sdk, aztec-pxe
    • Install: npx install-aztec-skills@latest
  • Aztec-lint v0.1.0
    • Released aztec-lint v0.1.0, a deterministic linter for Noir projects with Aztec-specific static analysis. - LINK
  • Aztec Watcher
    • Built a npm release monitor for aztec developers - LINK

Events

6 Likes
2

Incredible work.

The community is excited for the wallet!

I am hopeful you will bulid a bridge as well :slight_smile:

1 Like