Microsoft requires large amounts of energy to run its BingAI artificial intelligence model. In search of a clean and sustainable power source to provide this power, the company set its sights on nuclear energy. For this reason, AI is exploring plans to use nuclear energy to power data centers.
Nuclear energy could be a clean, reliable power source for artificial intelligence
The company will hire a “nuclear technology program manager” to evaluate how nuclear energy could be used to power data centers that host artificial intelligence models, according to a job posting posted Thursday.
The tech giant will focus on using microreactors and small modular reactors, which are much cheaper to build and run than large nuclear reactors, to “power cloud and AI data centers.”
AI models require large amounts of computing power to run, and analysis reveals that AIs like ChatGPT can cost up to $700,000 per day to run due to massive server costs. The job posting suggests that Microsoft sees nuclear power as the way to meet its growing demand for power.
Nuclear energy can be a clean and reliable power source for artificial intelligence. Nuclear reactors can produce large amounts of power without producing pollution and greenhouse gas emissions. Additionally, this type of energy can guarantee the uninterrupted operation of AI models by providing an uninterrupted power supply.
There are growing concerns about the environmental impacts of generative AI’s massive energy use. Academic analysis estimates that GPT-3 training produced more than 550 tonnes of carbon dioxide and required 3.5 million liters of water.
Microsoft recently pledged to power its data centers on renewable energy and accelerate efforts to reduce emissions, and plans to make its business carbon negative, water positive and zero waste by 2030.