Intel Demos Chip to Compute With Encrypted Data
View original at spectrum.ieee.orgIEEE Spectrum - Technical Title: Intel Demos Chip to Compute With Encrypted Data Date: 2026-03-10 13:00 Source: https://spectrum.ieee.org/fhe-intel <img src="https://spectrum.ieee.org/media-library/overhead-view-of-intel-s-computing-chip-called-heracles.jpg?id=65174073&width=1200&height=800&coordinates=156%2C0%2C156%2C…
Ce que nous avons tiré de cette source
The claims Via News extracted from this document. We point to the source; we don't replace it.
Optalysys is looking at pushing way past the digital limit for FHE acceleration using photonics
60% confidenceBalancing the movement of data with the crunching of numbers is the key challenge in FHE acceleration
60% confidenceA full 3D-stacked commercial photonic FHE chip could be ready in two or three years
60% confidenceHeracles is the first hardware that works at scale for FHE computing
60% confidenceNiobium's chip will be the world's first commercially viable FHE accelerator designed for real-world cloud and AI infrastructure
60% confidenceIntel's work on Heracles is very good and carries weight when talking about scale
60% confidenceHeracles represents the start of a whole journey in FHE hardware, similar to the first microprocessor
60% confidenceMany smaller AI models will run fine on FHE accelerated hardware even with data expansion
60% confidenceHeracles achieves about the level of speedup expected from an all-digital FHE system
60% confidenceIntel has proven and delivered everything promised for the Heracles chip
60% confidenceSpecialized hardware for FHE is needed for emerging applications around deeper machine learning operations like neural nets, LLMs, or semantic search
60% confidenceAt current scale, there is less need for specialized FHE hardware
60% confidenceFor FHE, the size of cipher text is orders of magnitude larger than plain text
60% confidence