Niv-AI Exits Stealth With $12M Seed: High-Frequency Sensors + AI Models Tame Millisecond GPU Power Surges in Data Centers

TechCrunch·medium signal

Tel Aviv startup Niv-AI, backed by Glilot Capital, Grove Ventures, and Arc VC, emerged from stealth to solve a foundational AI infrastructure problem: millisecond-scale GPU power demand spikes that occur as processors switch between computation and inter-GPU communication make data center power management nearly impossible at scale. The company developed high-frequency power sensors feeding AI prediction models that dynamically balance workloads against available grid power, targeting the 'stranded power' problem where data centers can't safely utilize full electrical capacity. With $1T in GPU orders forecasted by Jensen Huang, power management infrastructure is a structural bottleneck.