Sandia National Laboratories requires precise double-precision floating point computation to simulate complex physics problems and maintain national security systems.
The leading tech manufacturers face intense pressure, which limits their ability to deliver specialized hardware to government researchers.
This shift opens a massive market for smaller players and ambitious startups like Israel’s NextSilicon to compete.
Sandia scientists currently test these alternative architectures to ensure a stable supply of high-performance computing power.
The new chips utilize an innovative data flow architecture that reprograms itself on the fly to maximize efficiency.
This method successfully reduces electricity usage by minimizing the time spent shuffling data back and forth to memory.
Recently, the startup’s systems passed a key technical milestone during a battery of rigorous government supercomputing tests.
Sandia officials will decide this fall whether to deploy the new hardware for demanding nuclear security work.
