Physics / mechanism

Homomorphic encryption (HE) allows computation on ciphertext such that decrypting the result equals applying the same function to plaintext. Fully homomorphic encryption (FHE) supports arbitrary operations via lattice-based constructions (CKKS, BFV, TFHE schemes). The core mechanism relies on learning-with-errors (LWE) hardness; noise accumulates with each operation and must be managed via bootstrapping. Current SOTA: TFHE bootstrapping ~1–10 ms per gate on CPU; GPU-accelerated libraries (OpenFHE, HEIR) cut this 10–100×. Practical deployments remain 3–6 orders of magnitude slower than plaintext compute. Key parameters: polynomial modulus degree (n = 2¹⁴–2¹⁶), noise budget, and multiplicative depth.

Competitive landscape

The primary competitive tension is between FHE, secure multi-party computation (MPC), and trusted execution environments (TEEs like Intel TDX, AMD SEV). Each occupies a different point on the trust-vs-performance curve.

Companies using

Connected ideas

Sources

Frontier (open questions)

• To be added.

Merged from root duplicate (homomorphic-encryption.md at concepts/ root, 2026-06-10)

Homomorphic Encryption (FHE)

Encryption that allows computation directly on ciphertext, so data stays encrypted in use. The blocker is compute cost (orders of magnitude over plaintext), which drives the Fhe Acceleration Silicon thesis: purpose-built accelerators for the polynomial/NTT bottleneck. A pillar of confidential computing alongside TEEs.