Compute & Cloud
Niv-AI exits stealth to tackle GPU power throttling
Niv-AI emerged from stealth with $12 million in seed funding to develop power management tools for data centers, which currently throttle GPU usage by up to 30%.
Today, Tel Aviv-based startup Niv-AI emerged from stealth—operating in secret before its public launch—with $12 million in seed funding. The company aims to address the power management challenges that force data center operators to throttle down GPU usage by as much as 30% due to power grid management issues. The seed round was backed by investors including Glilot Capital, Grove Ventures, Arc VC, Encoded VC, Leap Forward, and Aurora. The startup was founded by CEO Tomer Timor and CTO Edward Kizis.
Modern artificial intelligence workloads require rapid, millisecond-scale power surges as processors switch between computation tasks and communication. These surges make it difficult for data centers to manage the power they draw from the electrical grid, leading to inefficiencies. Nvidia CEO Jensen Huang recently highlighted this issue, stating that significant power is squandered in these facilities and that every unused watt represents lost revenue.
Niv-AI CEO Tomer Timor explained that the core issue lies in the relationship between facilities and the electrical grid. According to Timor, “The grid is actually afraid of the data center consuming too much power at a specific time.” He noted that the startup is addressing a problem with two sides of the rope, aiming to help data centers utilize more GPUs and make better use of the power they already pay for, while also creating more responsible power profiles between the facilities and the grid. Lior Handelsman, a partner at Grove Ventures who sits on Niv-AI’s board, added that operators cannot continue building data centers under current methods.
To address these bottlenecks, Niv-AI is developing what its founders describe as an intelligence layer between data centers and the electrical grid. The company is currently deploying rack-level sensors—hardware designed to monitor power usage in data center racks—to detect power usage at the millisecond level. The startup plans to use the collected data to build a model that can predict and synchronize power loads across facilities. Niv-AI expects to have an operational system in a handful of U.S. data centers in the next six to eight months. This solution targets a critical bottleneck for hyperscalers—large-scale cloud computing providers—that face land-use and supply chain constraints when attempting to build new facilities.
Why it matters
Niv-AI is attempting to solve the power management challenges faced by data centers as they struggle with GPU power surges, which currently force operators to throttle usage by up to 30%.