From 22c034afb8a968273627cdd7f46db02718c88d06 Mon Sep 17 00:00:00 2001 From: Himess Date: Thu, 23 Jul 2026 17:16:20 +0300 Subject: [PATCH 1/4] fix(vendor): update Zama to mainnet, correct architecture to handle/coprocessor model --- patterns/pattern-private-shared-state-fhe.md | 4 +- vendors/zama.md | 71 ++++++++++++-------- 2 files changed, 46 insertions(+), 29 deletions(-) diff --git a/patterns/pattern-private-shared-state-fhe.md b/patterns/pattern-private-shared-state-fhe.md index b08afa1..b6c4fac 100644 --- a/patterns/pattern-private-shared-state-fhe.md +++ b/patterns/pattern-private-shared-state-fhe.md @@ -46,7 +46,7 @@ related_patterns: open_source_implementations: - url: https://github.com/zama-ai/fhevm - description: "Zama fhEVM, Fully Homomorphic Encryption runtime for EVM contracts (testnet)" + description: "Zama fhEVM, Fully Homomorphic Encryption runtime for EVM contracts (mainnet)" language: "Solidity, Rust" - url: https://github.com/zama-ai/tfhe-rs description: "TFHE-rs, pure-Rust implementation of TFHE primitives used by coprocessors" @@ -109,4 +109,4 @@ Threat model: - [Zama](../vendors/zama.md) - [Fhenix](../vendors/fhenix.md) -- [Zama fhEVM documentation](https://docs.zama.ai/fhevm) +- [Zama Protocol documentation (FHE on blockchain)](https://docs.zama.org/protocol/protocol/overview) diff --git a/vendors/zama.md b/vendors/zama.md index ceeba7a..9e207c9 100644 --- a/vendors/zama.md +++ b/vendors/zama.md @@ -1,62 +1,79 @@ --- title: "Vendor: Zama" status: ready -maturity: testnet +maturity: mainnet --- -# Zama – FHE SDK & Confidential Smart Contracts (Fully Homomorphic Encryption for EVM) +# Zama – Confidential Blockchain Protocol (Fully Homomorphic Encryption on existing chains) ## What it is -Zama develops **open-source FHE (Fully Homomorphic Encryption)** tools that enable confidential smart contracts and encrypted data processing directly on EVM-compatible networks. +Zama develops open-source Fully Homomorphic Encryption (FHE) tooling and operates the Zama Confidential Blockchain Protocol, which lets smart contracts compute on encrypted data on existing public chains. Zama describes it as "not a new L1 or L2, but rather a cross-chain confidentiality layer sitting on top of existing chains", so applications stay on their host chain instead of migrating to a dedicated network. Ethereum mainnet is live; further EVM chains and Solana are on the published roadmap. ## Fits with patterns -- [Private Stablecoin Shielded Payments](../patterns/pattern-private-stablecoin-shielded-payments.md) - FHE enables encrypted balances/amounts for confidential transfers -- [Private Intent-Based Vaults](../patterns/pattern-private-vaults.md) - FHE-enabled chains allow private strategy execution while keeping assets auditable -- [Shielded ERC-20 Transfers](../patterns/pattern-shielding.md) - Native FHE implementation for confidential token transfers -- [Private L2s](../patterns/pattern-privacy-l2s.md) - Zama's fhEVM provides FHE-based rollup capabilities -- [Shielding](../patterns/pattern-shielding.md) - FHE can complement ZK approaches for confidential balance management +- [Private Shared State (FHE)](../patterns/pattern-private-shared-state-fhe.md) - encrypted state shared between parties, computed off-chain by coprocessors +- [Private Stablecoin Shielded Payments](../patterns/pattern-private-stablecoin-shielded-payments.md) - encrypted balances and transfer amounts +- [Private Intent-Based Vaults](../patterns/pattern-private-vaults.md) - strategy parameters stay encrypted during execution +- [Shielding](../patterns/pattern-shielding.md) - partial fit: ERC-7984 hides amounts and balances, but sender and receiver stay public ## Not a substitute for -- ZK-based L2 privacy (e.g., Aztec, Scroll) -- MPC or TEEs for off-chain privacy computation +- Graph privacy: sender and recipient addresses remain public, so unlinkability needs composition with stealth addresses +- Privacy rollups that move execution and state off the host chain entirely +- Multi-party computation or Trusted Execution Environments as general-purpose off-chain compute ## Architecture -Smart contracts are compiled with Zama’s FHE libraries (e.g., `concrete`, `fhEVM`) allowing arithmetic on encrypted values. Execution happens deterministically, with ciphertexts persisted on-chain. +Contracts import the FHEVM Solidity library and use encrypted types (`euint8` to `euint256`, `ebool`, `eaddress`). Each encrypted value is referenced on-chain by a 32-byte handle. Zama's documentation places the ciphertext itself "off-chain, with the coprocessor" and the handle on-chain as the identifier pointing to it. FHE operations run symbolically on the host chain, generating new handles and emitting events, while a network of coprocessors performs the computation using TFHE-rs. + +The protocol has five components: the FHEVM Solidity library; host contracts that enforce access control and emit events; coprocessors that verify inputs, execute FHE operations and store ciphertexts; the Gateway, a dedicated Arbitrum rollup that orchestrates input validation, decryption and bridging; and a threshold multi-party computation Key Management Service. ## Privacy domains -- FHE encryption at the contract and variable level. -- Supports hybrid models (FHE + ZK for verification). +- Fully Homomorphic Encryption at the contract and variable level: amounts and balances stay encrypted end to end +- Programmable access per ciphertext through an Access Control List (`allow`, `allowThis`, `allowTransient`, `makePubliclyDecryptable`), including delegation of user decryption +- Hybrid by design: zero-knowledge proofs validate encrypted inputs, multi-party computation performs threshold decryption ## Enterprise demand and use cases -- Financial institutions seeking **on-chain confidentiality** with deterministic settlement. -- Private vaults, confidential lending, or yield strategies. +- Confidential payments and stablecoin transfers where amounts and balances stay hidden from other participants +- Tokenization and real-world assets where holdings and allocation sizes are confidential but settlement stays on a public chain +- Sealed-bid auctions, confidential distributions and governance, where inputs stay private until a defined reveal ## Technical details -- EVM-compatible FHE runtime (`fhEVM`) -- SDKs for Solidity and TypeScript -- Ongoing collaborations with Fhenix and Aleph Zero +- FHEVM Solidity library (`@fhevm/solidity`), with a relayer SDK and a Hardhat plugin +- Confidential tokens follow ERC-7984, implemented in OpenZeppelin's confidential-contracts library +- Encrypted inputs carry a zero-knowledge proof of knowledge, validated through `FHE.fromExternal` +- Conditional logic uses `FHE.select` instead of plaintext branching, so the taken branch is not revealed +- Protocol fees apply to input verification, decryption and bridging. They are priced in USD and can be paid by the end user, the application or a relayer ## Strengths -- Native EVM integration -- Strong cryptographic research pedigree -- Open-source and actively maintained +- Runs on existing chains, so users do not bridge to a separate network +- Composable between confidential contracts and with non-confidential ones +- Compliance rules are defined by each application in its own contracts rather than by the protocol +- The FHE scheme is post-quantum ## Risks and open questions -- High compute cost and latency -- Limited developer tooling maturity -- Interoperability between FHE networks still emerging +- Decryption depends on the threshold Key Management Service: 13 multi-party computation nodes, with a documented threshold example of 9 of 13, giving a correct output with up to one third malicious nodes +- Those nodes run inside AWS Nitro Enclaves, so verifiability rests partly on hardware assumptions. Zama states the goal is to add zero-knowledge proofs to the multi-party computation protocol to remove that dependency +- Coprocessor correctness rests on a majority-honest assumption, backed by staking and slashing +- The zero-knowledge proof used for input verification is not yet post-quantum, unlike the FHE and multi-party computation layers +- Operators can pause the protocol and blacklist addresses, and the Access Control List carries an account deny list +- Interoperability between Fully Homomorphic Encryption networks is still emerging +- Throughput and cost stay above plaintext execution, and protocol support beyond Ethereum mainnet is roadmap rather than shipped ## Links -- [https://www.zama.ai](https://www.zama.ai) -- [https://github.com/zama-ai/fhevm](https://github.com/zama-ai/fhevm) -- [Confidential ERC-20](https://www.zama.ai/post/confidential-erc-20-tokens-using-homomorphic-encryption) +- [Zama Protocol documentation](https://docs.zama.org/protocol) +- [Litepaper](https://docs.zama.org/protocol/zama-protocol-litepaper) +- [Handles: what lives on-chain and off-chain](https://docs.zama.org/protocol/solidity-guides/smart-contract/handles) +- [Coprocessor](https://docs.zama.org/protocol/protocol/overview/coprocessor) +- [Gateway](https://docs.zama.org/protocol/protocol/overview/gateway) +- [Key Management Service](https://docs.zama.org/protocol/protocol/overview/kms) +- [Access Control List](https://docs.zama.org/protocol/solidity-guides/smart-contract/acl) +- [zama-ai/fhevm](https://github.com/zama-ai/fhevm) +- [TFHE-rs](https://github.com/zama-ai/tfhe-rs) From 723abc3ead4032094410470f24d1e24066ffdd8d Mon Sep 17 00:00:00 2001 From: Himess Date: Thu, 23 Jul 2026 18:14:19 +0300 Subject: [PATCH 2/4] docs: changelog entry for #180 --- CHANGELOG.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/CHANGELOG.md b/CHANGELOG.md index 3f29a37..1fcf085 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -4,6 +4,8 @@ All notable changes to the EthSystems Map are documented here. ## [Unreleased] +- fix(vendor): [Zama](vendors/zama.md) — mainnet maturity, symbolic-execution architecture, TFHE-rs, ERC-7984; sync fhEVM status and docs link in [Private Shared State (FHE)](patterns/pattern-private-shared-state-fhe.md) ([#180](https://github.com/ethsystems/map/pull/180)) + ## [0.4.0] - 2026-07-02 23 commits, 166 files changed since [v0.3.0](https://github.com/ethsystems/map/releases/tag/v0.3.0) (Apr 2026). Major additions: resilience use cases (civic participation, disbursement rails, identity continuity), I2U protection patterns, pattern/approach schema v2 (strict flip) with CROPS and post-quantum analysis, domain reframing beyond FIs, and a Q2 2026 content QA audit. From 2b3b3c571f08334a453196ee1d5688469ac6245a Mon Sep 17 00:00:00 2001 From: Himess Date: Thu, 23 Jul 2026 18:56:53 +0300 Subject: [PATCH 3/4] fix(vendor): note dual licence and latency in Zama risks --- vendors/zama.md | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) diff --git a/vendors/zama.md b/vendors/zama.md index 9e207c9..0c40b49 100644 --- a/vendors/zama.md +++ b/vendors/zama.md @@ -27,7 +27,7 @@ Zama develops open-source Fully Homomorphic Encryption (FHE) tooling and operate Contracts import the FHEVM Solidity library and use encrypted types (`euint8` to `euint256`, `ebool`, `eaddress`). Each encrypted value is referenced on-chain by a 32-byte handle. Zama's documentation places the ciphertext itself "off-chain, with the coprocessor" and the handle on-chain as the identifier pointing to it. FHE operations run symbolically on the host chain, generating new handles and emitting events, while a network of coprocessors performs the computation using TFHE-rs. -The protocol has five components: the FHEVM Solidity library; host contracts that enforce access control and emit events; coprocessors that verify inputs, execute FHE operations and store ciphertexts; the Gateway, a dedicated Arbitrum rollup that orchestrates input validation, decryption and bridging; and a threshold multi-party computation Key Management Service. +The protocol combines the FHEVM Solidity library; host contracts that enforce access control and emit events; coprocessors that verify inputs, execute FHE operations and store ciphertexts; the Gateway, a dedicated Arbitrum rollup that orchestrates input validation, decryption and bridging; and a threshold multi-party computation Key Management Service. Relayer and oracle services connect users and contracts to the Gateway and are not part of the trusted base. ## Privacy domains @@ -51,20 +51,21 @@ The protocol has five components: the FHEVM Solidity library; host contracts tha ## Strengths -- Runs on existing chains, so users do not bridge to a separate network +- Host contracts deploy onto existing chains, EVM or non-EVM, so users do not move to a dedicated network - Composable between confidential contracts and with non-confidential ones +- Every FHE computation and input verification carries a commitment and a signature, so anyone can recompute the result and check it independently +- No single party ever holds the decryption key: it is secret-shared across the threshold network, and each decryption is authorized by the Access Control List and logged through the Gateway +- Confidential contracts are written in plain Solidity, without a new language or cryptography expertise - Compliance rules are defined by each application in its own contracts rather than by the protocol -- The FHE scheme is post-quantum +- Audited by Trail of Bits and Zenith +- The FHE and multi-party computation layers are post-quantum ## Risks and open questions -- Decryption depends on the threshold Key Management Service: 13 multi-party computation nodes, with a documented threshold example of 9 of 13, giving a correct output with up to one third malicious nodes -- Those nodes run inside AWS Nitro Enclaves, so verifiability rests partly on hardware assumptions. Zama states the goal is to add zero-knowledge proofs to the multi-party computation protocol to remove that dependency -- Coprocessor correctness rests on a majority-honest assumption, backed by staking and slashing -- The zero-knowledge proof used for input verification is not yet post-quantum, unlike the FHE and multi-party computation layers +- The trust model is layered: decryption runs through a threshold Key Management Service of 13 multi-party computation nodes, with a documented threshold example of 9 of 13 and tolerance up to one third malicious; those nodes run inside AWS Nitro Enclaves, so verifiability rests partly on hardware assumptions; and coprocessor correctness rests on a majority-honest assumption backed by staking and slashing. Zama states the goal is to add zero-knowledge proofs to the multi-party computation protocol to remove the hardware dependency - Operators can pause the protocol and blacklist addresses, and the Access Control List carries an account deny list -- Interoperability between Fully Homomorphic Encryption networks is still emerging -- Throughput and cost stay above plaintext execution, and protocol support beyond Ethereum mainnet is roadmap rather than shipped +- Open source under a dual license: free for non-commercial use, with a commercial license required for use outside the Zama Protocol +- Cost and latency stay above plaintext execution, and protocol support beyond Ethereum mainnet is roadmap rather than shipped ## Links From b40b951af8302dc6445174403036a34396b59bb6 Mon Sep 17 00:00:00 2001 From: Himess Date: Thu, 23 Jul 2026 19:20:41 +0300 Subject: [PATCH 4/4] fix(vendor): tighten Zama strengths section --- vendors/zama.md | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/vendors/zama.md b/vendors/zama.md index 0c40b49..802d700 100644 --- a/vendors/zama.md +++ b/vendors/zama.md @@ -52,13 +52,11 @@ The protocol combines the FHEVM Solidity library; host contracts that enforce ac ## Strengths - Host contracts deploy onto existing chains, EVM or non-EVM, so users do not move to a dedicated network -- Composable between confidential contracts and with non-confidential ones +- Composable between confidential contracts and with non-confidential ones, and contracts are written in plain Solidity rather than a new language - Every FHE computation and input verification carries a commitment and a signature, so anyone can recompute the result and check it independently -- No single party ever holds the decryption key: it is secret-shared across the threshold network, and each decryption is authorized by the Access Control List and logged through the Gateway -- Confidential contracts are written in plain Solidity, without a new language or cryptography expertise +- No single party ever holds the decryption key: it is secret-shared across the threshold network, and each decryption is authorized by the Access Control List - Compliance rules are defined by each application in its own contracts rather than by the protocol -- Audited by Trail of Bits and Zenith -- The FHE and multi-party computation layers are post-quantum +- Audited by Trail of Bits and Zenith, and the FHE and multi-party computation layers are post-quantum ## Risks and open questions