Yes, AI uses energy. So does every click you make

While public attention largely focuses on the massive energy demands of artificial intelligence, everyday digital activities like checking out on a shopping site, searching the web, or opening an app also consume electricity. Behind the scenes, these actions trigger database queries that require servers to draw power. As digitalization deepens, researcher Song Wu is working to address the hidden energy costs of this underlying software, moving past the traditional focus on execution speed to prioritize power consumption.

Wu is involved in the GreenSQL project, a collaboration involving several academic institutions and the software company Software Improvement Group. Historically, database research focused almost entirely on runtime performance to make query processing as fast as possible. GreenSQL introduces an entirely new dimension by shifting toward energy awareness and helping organizations understand the environmental cost of the instructions they send to databases.

The project aims to give businesses the ability to predict and optimize for energy consumption in the same way they currently track performance. By developing a catalogue of energy-efficient query patterns, the initiative allows companies to continue operating while reducing their electricity use and overall environmental impact.