@0xvendetta
FHE is not cheap.
Operations on ciphertexts are thousands of times slower than on plaintexts although @zama's optimizations are cutting that gap fast.
What's the trade-off?🤔
You get deterministic privacy.
With zero-knowledge proofs or MPC, you rely on distributed protocols and assumptions about the participant's honesty.
With FHE, you rely solely on math. Specifically, the hardness of lattice problems.
Zama’s roadmap focuses on:
• Reducing bootstrapping costs
• Leveraging GPU acceleration
• Compiling high-level code to optimized FHE circuits automatically
It’s not yet for every app, but for privacy-critical workloads, like encrypted AI inference or private blockchain contracts. It is the cleanest guarantee we have.
0 reply
0 recast
2 reactions