@banlamduocma

Gm @zama 🟨 Zama's Fully Homomorphic Encryption (FHE) allows performing calculations (addition, subtraction, multiplication, comparison) on encrypted data without decryption, ensuring optimal security, especially suitable for blockchain and AI. 🟨 Technology based on TFHE (Fast Fully Homomorphic Encryption over the Torus), optimized for integer arithmetic with high precision, outperforming schemes like CKKS. Main components include: • TFHE-rs: Open source Rust library, supporting functions like TFHE.le, TFHE.add/sub, with bootstrap under 1ms on GPU. • ​​Concrete: Includes Concrete Compiler (based on MLIR for program compilation) and Concrete ML (for cryptographic machine learning), version v2.11 integrates well with Rust. • fhEVM: Ethereum virtual machine supporting cryptographic smart contracts with euint data type, reducing gas fees with coprocessors. • Performance: Currently 20 TPS/chain, target 1,000 TPS by 2026 with GPU/ASIC, and 100,000 TPS long term. HPU (FPGA, 13,000 ops/sec) is an open source hardware accelerator, expected to expand to servers in 2026. • Security: Surpassing TEE with cryptographic proofs, independent of vulnerable hardware. Public testnet running, Mainnet expected Q4 2025, EVM and Solana support from 2026. 🟨 Real-world application examples • Blockchain/DeFi: Encrypted ERC20 for anonymous transfers, preventing front-running; confidential swaps/lending on Arbitrum via Conduit. • AI: Concrete ML analyzes encrypted medical data (patient data) or predicts credit; encrypted DNA testing protects identity. • Other: Sealed-bid auctions, anonymous proof of humanity, crypto games like Rock-Paper-Scissors (PlayAscend). ❇️ Zama leads FHE with over 5,000 developers in the community, promising to change the way data is handled on the blockchain. Image created by #ZamaCreatorProgram

14 replies

10 recasts

36 reactions