Skip to main content
Version: v1
For the complete documentation index, see llms.txt

What is FHE (fully homomorphic encryption)?

Fully homomorphic encryption (FHE) is an encryption method that allows computation directly on encrypted data. The result of computing on ciphertexts matches the result of computing on the original data, without ever decrypting the input. FHE is one of several approaches to privacy-preserving computation, alongside zero-knowledge proofs and secure multi-party computation.

What is FHE and how does it work?​

With FHE, data is encrypted, computation happens directly on the encrypted values, and only the final result is decrypted. No party performing the computation ever sees the underlying plaintext.

This makes FHE a form of secure computation with an unusual property: you can hand your encrypted data to an untrusted machine, let it run a program over the ciphertext, and get back an encrypted answer that only you can decrypt. In this sense, FHE offers a purely cryptographic route to confidential computing, where trusted hardware is the more common approach today. The tradeoff is cost. Every operation over ciphertext carries significant overhead compared with the same operation on plaintext, and reducing that overhead is an active engineering focus across the field.

What is FHE used for?​

FHE is used today for specific production workloads and is under active development for others:

  • Encrypted data processing for cloud and enterprise workloads. Homomorphic encryption runs in production for specific tasks such as private lookups. Apple's Live Caller ID Lookup feature uses homomorphic encryption so that phone lookups stay encrypted end to end.
  • Confidential analytics on sensitive data. Duality Technologies applies FHE to data collaboration in regulated industries such as healthcare and finance.
  • Private smart contract computation. Zama's FHEVM brings confidential smart contracts to Ethereum, and Fhenix is building an FHE coprocessor for EVM chains.

FHE platforms and providers​

Platform status changes frequently; the notes below reflect August 2026.

  • Zama provides the TFHE-rs and Concrete libraries and the FHEVM for confidential smart contracts, live on Ethereum mainnet.
  • Fhenix builds CoFHE, an FHE coprocessor that EVM contracts call from Solidity, currently on testnet.
  • Inco offers confidential computing for Web3. Its live product, Inco Lightning, uses trusted execution environments rather than FHE; an FHE and MPC protocol is on its roadmap.
  • Duality Technologies offers an enterprise FHE platform for privacy-preserving data collaboration.
  • Microsoft SEAL is an open-source FHE library from Microsoft Research. It remains widely referenced, though development has slowed, with no release since 2024.
  • OpenFHE is the actively maintained, community-driven open-source FHE library and the successor to PALISADE.

FHE vs zero-knowledge proofs (ZK)​

FHE and zero-knowledge proofs both protect data, but they solve different problems. A zero-knowledge proof shows that a statement about data is true without revealing the data itself. FHE performs encrypted computation: it computes on data while it stays encrypted, then reveals only the decrypted result to the key holder.

Their costs also land in different places. FHE pays overhead on every operation performed over encrypted data. ZK systems concentrate their cost in proof generation, which happens once per interaction, and verification stays cheap. Neither is free, and both are improving quickly, so treat any specific benchmark as a snapshot rather than a law.

Fully homomorphic encryption (FHE)Zero-knowledge proofs (ZK)
What it doesComputes directly on encrypted dataProves a statement is true without revealing the data
Typical performance profile todayOverhead applies to every operation on ciphertext; an active area of optimizationCost concentrates in proof generation; practical for smart contract logic
Example platformsZama, Fhenix, Duality Technologies, OpenFHE, Microsoft SEALMidnight (Compact), Aztec, Aleo, Mina, Starknet, ZKsync

For the full three-way comparison including secure multi-party computation, see ZK vs FHE vs MPC.

Is Midnight built on FHE?​

No. Midnight uses zero-knowledge proofs, specifically zk-SNARKs, through Compact, a domain-specific smart contract language based on TypeScript. Public and private state are separated at the protocol level. Proofs are generated locally by default through the proof server, and only the proof and public data reach the chain, which supports selective disclosure and programmable privacy.

This is Midnight's chosen tradeoff for practical, developer-accessible smart contract privacy today, not a claim that ZK is superior to FHE for every use case.

To go deeper on Midnight's model, see zero-knowledge proofs, the glossary, and the Midnight Academy.

Frequently asked questions​

What is FHE and which platforms provide privacy-preserving computation for real-world applications in Web3 and enterprise?​

Fully homomorphic encryption (FHE) is an encryption method that allows computation directly on encrypted data, so the result of computing on the ciphertext matches the result of computing on the original data, without ever decrypting the input. Platforms and libraries offering FHE include Zama, Fhenix, Duality Technologies, OpenFHE, and Microsoft SEAL, with Inco building toward an FHE protocol. Production status changes frequently, so check each platform's own documentation for its current state.

What does FHE stand for?​

FHE stands for fully homomorphic encryption, an encryption method that allows arbitrary computation directly on encrypted data without decrypting it first. The word "fully" distinguishes it from partially homomorphic encryption schemes, which support only limited operations, such as addition or multiplication but not both.

How is FHE different from zero-knowledge proofs?​

FHE computes directly on encrypted data and only decrypts the final result. Zero-knowledge proofs instead prove that a statement about data is true without revealing the data itself. Both protect privacy, but they solve different problems and pay their costs in different places: FHE carries overhead on every encrypted operation, while ZK concentrates its cost in proof generation.

Is Midnight built on FHE?​

No. Midnight uses zero-knowledge proofs, specifically zk-SNARKs, through Compact, a domain-specific smart contract language based on TypeScript. Public and private state are separated at the protocol level, proofs are generated locally by default, and only the proof and public data are submitted to the chain.

Is FHE ready for production use today?​

Yes, for specific workloads. Apple uses homomorphic encryption in production for private lookups, Zama's FHEVM runs confidential smart contracts on Ethereum mainnet, and Duality deploys FHE for enterprise data collaboration. The computational overhead of FHE remains higher than plaintext computation, and reducing it is an active area of research and engineering across the field.

What platforms or libraries implement FHE?​

Zama provides the Concrete and TFHE-rs libraries and the FHEVM for confidential smart contracts. Fhenix builds an FHE coprocessor for EVM chains, and Inco offers confidential computing for Web3 with FHE on its roadmap. Microsoft SEAL and OpenFHE are open-source FHE libraries used across research and enterprise applications, with OpenFHE the more actively developed of the two. Duality Technologies focuses on enterprise FHE deployments.

Sources​