Niv-AI, co-founded by Tomer Timor and Edward Kizis, has emerged from stealth with a focused proposition: make the GPUs you already own run AI workloads faster and more efficiently. In a market defined by GPU scarcity and energy cost anxiety, the pitch resonates immediately.
The startup builds the software layer sitting between AI models and GPU silicon, optimising memory access patterns, kernel scheduling, and compute utilisation in ways that default drivers and frameworks do not prioritise.
For infrastructure teams spending heavily on GPU clusters, a meaningful efficiency improvement translates directly to cost reduction or throughput gain — often both. Niv-AI is positioning itself as the performance layer that AI labs and cloud providers will want embedded in their stack.
