Notebookcheck Logo

Microsoft to build nuclear plants for powering AI

Microsoft's AI ambitions: Nuclear power plants as key to energy transition? (Symbolic image: Bing AI)
Microsoft's AI ambitions: Nuclear power plants as key to energy transition? (Symbolic image: Bing AI)
Microsoft is betting on nuclear power to drive the development of artificial intelligence. Small modular reactors (SMRs) are now set to power the energy-intensive AI. The company is also investing in a project to generate nuclear power through fusion.

AI systems use a lot of energy. Neural networks consume significant electricity when processing internet data, and cooling systems also require power. Companies investing in AI are developing solutions to address this issue. For instance, Microsoft Corp. is currently a leading AI player, working with OpenAI and exploring nuclear power as a solution.

Microsoft's decision could start a trend that speeds up the development of AI. If companies use nuclear power for their AI systems, they could make those systems larger and more effective. In May, Microsoft agreed to purchase power from Helion Energy, a start-up that aims to create nuclear energy from nuclear fusion by 2028. However, this makes up less than 1% of Microsoft's total portfolio of carbon-free power purchase agreements.

Microsoft relies on small modular reactors

To operate the data centres that support Microsoft's cloud and AI, the company is hiring a manager for their nuclear technology programme. Microsoft intends to use small modular nuclear fission reactors (SMRs), which are expected to be safer than older reactors, to combat carbon emissions.

It is well-known that using nuclear energy carries risks as nuclear power plant accidents could be catastrophic. Furthermore, the pairing of AI, often viewed as a danger to mankind, with another intimidating technology is causing mounting concern. Nonetheless, some environmental activists see nuclear power as an essential part of the transition from fossil fuels to a carbon-neutral economy.


Bloomberg | Symbolic image: Bing AI

Read all 13 comments / answer
static version load dynamic
Loading Comments
Comment on this article
Please share our article, every link counts!
Nicole Dominikowski, 2023-11- 1 (Update: 2023-11- 1)