iExec – Confidential Computing & Off-Chain Execution (TEE-Based Verifiable Compute for EVM)

Fits with patterns

Not a substitute for

  • Fully Homomorphic Encryption (FHE)-based on-chain privacy (e.g., fhEVM)
  • ZK-based shielded L2s (e.g., Aztec)
  • Pure MPC-based decentralized compute networks

Architecture

Smart contracts emit task requests that are matched via the iExec marketplace. A worker node executes the task inside a Trusted Execution Environment. A cryptographic attestation proves the expected code was executed inside a genuine secure enclave; the verified result is returned on-chain.

Core components include PoCo (Proof-of-Contribution), TEE-enabled worker nodes (Intel SGX), a decentralized marketplace for compute, data, and applications, on-chain verification.

Privacy domains

  • TEE-based confidentiality: data is decrypted inside secure enclaves and not exposed in plaintext to infrastructure operators.
  • Code confidentiality: application logic can be protected and executed privately within enclaves.
  • Hybrid models combining TEE execution with ZK proofs for verification, on-chain settlement, off-chain AI pipelines, and enterprise compliance.

Enterprise demand and use cases

  • Financial institutions and DeFi applications requiring verifiable off-chain confidentiality with attested settlement.
  • Confidential portfolio computation, risk analysis, private liquidations, and sealed-bid auctions.
  • AI and data marketplaces enabling secure dataset monetization and confidential AI inference.
  • Web3 and Web2 integration through secure SaaS automation, enterprise data bridges, and confidential API orchestration.

Technical details

  • Ethereum-compatible, live on mainnet
  • Intel SGX-based TEE infrastructure
  • JavaScript / TypeScript SDKs and smart contract integration tooling
  • Decentralized task marketplace

Strengths

  • Infrastructure designed for production deployments
  • Hardware-backed TEE security model
  • Verifiable off-chain execution
  • Integration possibilities across AI, data, and Web3 applications
  • Enterprise and Web2 interoperability capabilities

Risks and open questions

  • Reliance on hardware security assumptions (Intel SGX trust model)
  • Centralization considerations around TEE hardware supply
  • Workloads requiring GPUs or specialized hardware accelerators cannot run inside SGX enclaves
  • Very large memory workloads are constrained by enclave memory limits

Links