As artificial intelligence (AI) expands rapidly, large-scale data centers are becoming one of the largest and most dynamic classes of new electric loads. Unlike regular data centers with relatively constant demands for power, AI training centers use thousands of specialized computer chips that work together in tightly coordinated cycles.
May 28, 2026
Estimated Read Time min
Sandra Jenkins
Sandra Jenkins is a program manager in the Grid Controls and Communications Division in the Office of Electricity at the U.S. Department of Energy. The programs she manages oversee transmission reliability, renewable integration, sensors, and data analytics.
She has also worked with the U.S. Department of State as a global sustainability advisor. Before coming to Office of Electricity, she worked on the DOE Quadrennial Energy Review in the Energy Policy and Systems Analysis office, focusing on interdependencies with electricity and natural gas.
Sandra earned a master’s degree in technology and policy from MIT and a bachelor’s degree in electrical engineering from the University of Massachusetts-Amherst.
Key Measurement Needs for a Changing Grid
As artificial intelligence (AI) expands rapidly, large-scale data centers are becoming one of the largest and most dynamic classes of new electric loads. Unlike regular data centers with relatively constant demands for power, AI training centers use thousands of specialized computer chips that work together in tightly coordinated cycles. This synchronized activity causes repetitive fluctuations—called oscillations—of the electrical load over time. These oscillations can span a wide range of frequencies, some of which may adversely interfere with equipment at nearby power plants, affecting the reliability of the electric grid.
Recognizing these challenges, Novel Applications for Synchronized Power Instrumentation (NASPI) released a technical report, Measurement Adequacy for Monitoring Data Center Oscillations. The report examines how well today’s measurement systems can detect oscillations from large AI-driven loads and highlights where additional capabilities may be needed.
Report Findings
PMUs remain essential, but have bandwidth limitations
Phasor measurement units (PMUs) are widely used to monitor voltage, current, and frequency at key locations, and they excel at detecting low-frequency oscillations from large data center loads. However, phasor estimation involves filtering that causes PMUs to miss or misrepresent high-frequency oscillations generated by AI workloads, even when configured to report at higher rates.
POW has a full frequency range, but requires more data storage
Point-on-wave (POW) measurements are high-resolution waveform recordings that can capture fast, high-frequency dynamics with precision. This process preserves the full frequency range of the waveforms, which is necessary for detecting high-frequency oscillations. However, continuously streaming these waveforms produces large amounts of data, resulting in communication and storage challenges for utilities.
A hybrid approach offers a practical path forward
The report concludes that hybrid monitoring, integrating both PMU and POW measurements, offers the most reliable pathway for assessing oscillations from large loads and supporting utility compliance. This approach allows operators to continuously monitor oscillations without overwhelming communication networks.
NASPI’s Role in Advancing Grid Measurement
Funded by the Department of Energy’s Office of Electricity with support from Pacific Northwest National Laboratory and the Electric Power Research Institute, NASPI has provided national leadership since 2007 in accelerating the adoption of advanced grid measurement technologies.
The deployment of PMUs fostered by NASPI has dramatically improved the ability of transmission operators to detect and respond to oscillations to protect grid equipment and prevent power outages. An example of NASPI delivering on support to industry is through their routine webinar series. The PMU Adequacy for Monitoring Data Center Oscillations session presented the strengths and limitations of PMUs in detecting high-frequency oscillations from data centers and highlighted the need for complementary high-resolution POW measurements to improve monitoring.
Drawing on its leadership and technical expertise, NASPI remains positioned to support the electric utility industry in effectively monitoring large loads using PMU and POW measurements, enabling rapid integration of data centers in the U.S. electric grid.
Looking Ahead
As AI-driven data centers continue to scale, understanding and monitoring their electrical behavior is critical to ensuring reliable, affordable power for American communities. The Office of Electricity will continue working with utilities, researchers, and industry through NASPI and other initiatives to strengthen the grid’s ability to safely and efficiently accommodate these emerging large loads.
