In response to the significant energy demands of running AI models like ChatGPT, Microsoft is exploring the use of nuclear energy to power its AI data centres. A recent job posting indicates that Microsoft is actively recruiting a “principal program manager for nuclear technology” to assess the feasibility of integrating nuclear energy into the infrastructure of data centres hosting AI models.
Microsoft‘s focus appears to be on the use of microreactors and Small Modular Reactors, which are more cost-effective and manageable compared to larger nuclear reactors. The goal is to sustain the data centres supporting the Microsoft Cloud and AI infrastructure, which require immense computational power to operate AI models effectively.
AI models, such as ChatGPT, impose substantial server expenses, potentially costing up to $700,000 per day to run. Microsoft sees nuclear energy as a viable solution to address this escalating energy demand.
Microsoft’s commitment to environmental sustainability is evident in its $10 billion agreement with OpenAI, the creators of ChatGPT. Microsoft provides cloud services to support OpenAI’s models and has integrated OpenAI’s technology into its Bing search engine.
The environmental concerns surrounding the energy consumption of generative AI are growing. Studies estimate that training models like GPT-3 result in significant carbon emissions and water usage.
Microsoft has recently pledged to accelerate its efforts to power data centres with renewable energy sources and reduce emissions. The company aims to become carbon negative, water positive, and achieve zero waste status by 2030, aligning its goals with a more sustainable future.