AI Power Consumption Nears that Of Small Countries, Warns Study

Artificial intelligence (AI) is rapidly becoming a significant consumer of global energy, with figures from Schneider Electric, a French energy management company, indicating that AI now consumes approximately 4.3GW of power worldwide. This energy consumption is roughly equivalent to that of some small countries. As AI technology continues to see widespread adoption, its power usage is expected to rise significantly.

Schneider Electric predicts that by 2028, AI could consume between 13.5GW and 20GW of power, marking a substantial increase with a compound annual growth rate of 26-36%. This increase in energy consumption is raising concerns about the environmental impact and sustainability of AI applications.

The study also highlights the broader issue of data center power consumption. Currently, AI accounts for only 8% of a typical data center’s energy usage, which totals 54GW. However, by 2028, data center energy consumption is projected to reach 90GW, with AI contributing around 15-20% of this demand. The study notes that AI’s power requirements may shift from being primarily used for training (the current 20%) to being more inference-heavy in the coming years.

Cooling data centers is an essential but energy-intensive process, and it can also lead to high water usage. Data centers have faced criticism for their environmental impact, as they often require substantial natural resources. Schneider Electric suggests that as AI workloads continue to grow, accurately predicting energy usage will become more challenging.

To address these energy challenges, Schneider Electric advises data center operators to transition from the conventional 120/208V power distribution to 240/415V, allowing them to accommodate the high power densities associated with AI workloads. This transition must be coupled with infrastructure upgrades and efficiency improvements to manage and reduce power usage while sustaining the growth of cloud computing and AI technologies. The findings underscore the importance of sustainable energy solutions and increased efficiency in the development and deployment of AI technologies.

Paulo Montenegro
Categories: Robots
Tags: AI (Artificial Intelligence)