The increasing power requirements of AI data centers have given rise to the phenomenon of power bursting – unforeseen spikes in energy usage due to the intensive computation needs of GPU servers powering AI workloads. Subsynchronous oscillations (SSO) can cause severe, costly damage to both data center infrastructure and the grid, which may not be equipped to handle these large load swings, making it essential for operators to identify and mitigate SSO bursts before they lead to costly outages due to transformer overheating, ferro resonant damage and other impacts to equipment.

Power management company Eaton today announced that it has developed a solution for identifying large fluctuations in energy demanded by artificial intelligence (AI) computing infrastructure, known as AI power bursts. Eaton’s new edge-based solution, available via a firmware update for its Eaton Power Xpert quality (PXQ) event analysis system, helps detect potential SSO in data centers, enabling operators to more effectively protect critical infrastructure and enhance resiliency as AI energy needs grow.

Eaton’s PXQ is a power quality meter for switchgear, switchboards and power distribution units (PDUs) that integrates sophisticated edge analytics to streamline the identification and resolution of power quality events such as sags, swells, transients and harmonics. Now, through a PXQ meter remote firmware upgrade, customers can leverage their system to reliably detect SSO and allow operators to take preventive actions to resolve issues before they cause lasting damage to data centers and the grid.

“The energy demands of AI workloads surpass anything data centers and the grid have encountered before, with load fluctuations that can exceed the limits of existing infrastructure,” said JP Buzzell, vice president and chief data center architect at Eaton.