OpenAI CEO Sam Altman has pointed to major technological breakthroughs as necessary for the development of superintelligent AI. Speaking at a Bloomberg event during the World Economic Forum's annual meeting in Davos, he discussed the energy requirements of expanding AI infrastructure, according to TechRadar.
The International Energy Agency estimates that data centers worldwide could use around 945 terawatt-hours of electricity annually by 2030, equivalent to 3% of global consumption. AI is expected to be a key driver of this growth.
Current energy infrastructure may struggle to meet the projected demand. Altman is looking to advances such as fusion energy to help address the challenge, while limited electricity availability is already restricting companies' ability to expand computing capacity and improve AI models.
Energy is not the only obstacle. TechRadar also identifies connections between systems, optical components and memory as increasingly important infrastructure requirements. Without sufficient power and suitable sites for new facilities, further expansion of AI capabilities could face significant limits.