Guy Itzhaki is CEO of Fhenix, a fully homomorphic encryption (FHE) company whose CoFHE coprocessor lets smart contracts on existing EVM chains compute over encrypted data without users moving to a new chain. Itzhaki spent much of his career at Intel, where he took its SGX and TDX trusted execution environments to market and later led its FHE and blockchain initiatives, with a stint as a technology advisor for the Tokyo and Paris Olympic Games, before joining founder Guy Zyskind to launch Fhenix three years ago. Why you should listen For as long as most engineers have worked with encryption, encrypted data has been unusable until it is decrypted, whether in the open or inside secure hardware. Itzhaki, who spent years selling that secure-hardware approach at Intel, says the moment of decryption is where the risk lies, which is why FHE has long been called the holy grail of cryptography. The math has existed since Craig Gentry's breakthrough more than 15 years ago, but FHE was dismissed as too complex for anyone but specialist cryptographers and too slow for production. Itzhaki argues both objections are now largely answered. Developers building on Fhenix don't need to understand the cryptography at all, and faster algorithms running on GPUs, and eventually dedicated ASICs, keep improving performance. CoFHE went live on Ethereum and Arbitrum mainnet in October. Itzhaki sees Web3 as only the first market, and expects private queries to AI models to become feasible within two to three years. The more interesting part of the conversation is how Fhenix found its market. The company started as a privacy-focused Layer 2, then concluded the world didn't need another public chain and rebuilt itself as a coprocessor for chains developers already use. It spent its early days testing gaming, NFTs and meme coins before settling on the area where privacy matters most: financial assets moving between parties, especially stablecoins and tokenized assets. Itzhaki separates confidentiality from anonymity. Fhenix keeps senders and receivers visible so KYC and AML still work, and hides only what is in the transaction. A permit system lets an asset owner grant decryption rights to an auditor, a compliance officer or a regulator, which puts onchain data on a need-to-know basis. A dual token standard lets a single ERC-20 switch between a public mode for DeFi and a private mode for balances and payments. Early design partners show the range of uses: AlphaEngine lets traders keep strategies encrypted while liquidity providers see only a performance score, Obolos puts company cap tables and compliance rules onchain, and Fluton is building confidential payments, swaps and trading. Itzhaki is candid about the competition. He credits Zama with proving there is real demand for onchain privacy, and notes that Fhenix absorbed Sunscreen, the third early FHE team, in July. He says Fhenix differs in three ways: a second FHE scheme planned for next year, its dual token SDK, and support for EVM chains generally rather than a single chain. He also previews fUSD, a confidential token due in about a month, whose staked version, fUSDY, is meant to earn yield through partners, so private balances can still be put to work. The hot take round covers his firmest conviction that privacy matters and institutions are coming, a ten-year outlook in which AI agents run most onchain activity, and a warning that quantum attacks may arrive within roughly two years, much sooner than most people assume. He closes by admitting he is a Lord of the Rings maximalist. Supporting links Stabull Finance Fhenix Fhenix on Twitter Andy on Twitter Brave New Coin on Twitter Brave New Coin If you enjoyed the show please subscribe to the Crypto Conversation and give us a 5-star rating and a positive review in whatever podcast app you are using.
