From Energy Hogs to Climate Assets: Teaching Buildings to Think With Artificial Intelligence-Driven Heating, Ventilation, and Air-Conditioning Control

Abstract

Commercial buildings are among the largest and least intelligent energy users in the modern economy, with heating, ventilation, and air-conditioning (HVAC) systems alone often accounting for more than half of their total energy consumption. For decades, the control of HVAC systems has relied primarily on static building management systems (BMSs) and manual adjustments, using fixed setpoints and schedules and periodic readjustment. Inefficient control has led to significant energy waste. As cities decarbonize and grids integrate more variable renewable energy, buildings must transform from passive energy consumers into intelligent, flexible, and grid-supporting assets. Recent advances in sensing, optimization, and artificial intelligence (AI) pave the way for this future. This article traces the evolution from well-tuned BMSs and supervisory analytics to optimized control using model predictive control (MPC) and advanced learning and looks ahead to the next generation of HVAC control systems capable of sensing occupancy, operating across spaces, interacting with the grid, and scaling to different buildings. We illustrate how real-time AI control can dynamically manage HVAC systems across zones and buildings, adjusting setpoints based on occupancy, weather, and energy prices to reduce waste, lower costs, and create new revenue streams through services, such as demand response (DR). As a case study, we highlight an AI-based automation system developed by the Massachusetts Institute of Technology (MIT) for its campus buildings. This system utilizes graph learning and reinforcement learning to capture limited room occupancy, local weather, and thermal interactions across multiple zones. In ongoing pilot projects, it has demonstrated significant energy savings by adjusting zone-level heating and cooling setpoints. Linking such academic innovations with a deployable and BMS-compatible software layer is key to transforming commercial buildings into intelligent and sustainable players in a clean energy future.