Harvard John A. Paulson School of Engineering and Applied Sciences
The Intersection of Power System and AI Computing
Research and education at the intersection of power systems, computing, and artificial intelligence.
The Power and AI Initiative (PAI) at Harvard School of Engineering and Applied Sciences (SEAS), led by Professor Le Xie and Professor Minlan Yu, is dedicated to advancing research, solutions, and educational opportunities at the intersection of power systems and AI. Our expertise in power systems, computing and AI, coupled with Harvard’s interdisciplinary strengths, ensures that our initiative is positioned to drive impactful innovation. By leveraging AI for grid management, enabling predictive maintenance, optimizing energy distribution, and integrating renewable resources seamlessly, we are reimagining the grid as an adaptive, efficient foundation for a sustainable energy future. This synergy between electrification and AI empowers us to accelerate progress toward a carbon-neutral world while helping AI data centers reduce operational costs and minimize their carbon impact.
Featured research
Two directions and five subtopics organize the current PAI research agenda
PAI’s research portfolio spans 14 projects across two broader directions: AI for Power, exemplified by PowerAgent, and Power for AI, exemplified by data center-power grid co-design.
AI for Power
Power for AI
- Efficient LLM Scheduling Latency, memory, and orchestration policies for modern large-model serving. Open
- Energy-Efficient LLM Training and Inference Resilient runtimes and power-aware inference systems aligned with grid conditions. Open
- Adaptive Modeling of Data Center Dynamics Control, resonance, and stability models for data center and grid interaction. Open
- Electricity and Data Centers Policy Brief and Strategy Planning and regulatory strategy for AI data centers and the U.S. electric grid. Open