Here's Why AI Companies Are Looking To The Nuclear Option

The video discusses the escalating electricity demands of AI companies due to the energy-intensive nature of artificial intelligence applications, straining the existing power grid in the US. To address this challenge, companies like New Scale Power are exploring the use of small modular nuclear reactors (SMRs) to provide carbon-free continuous power for data centers, offering a potential solution for meeting the growing energy needs of AI companies.

The video discusses how AI companies are facing a significant challenge in meeting the growing electricity demands of their data centers. With the rise of artificial intelligence applications, there is a surge in data center power needs, with AI queries requiring much more electricity than traditional searches. This surge is straining the existing power grid in the US, which relies heavily on fossil fuel plants. As a result, there is a need for alternative energy sources to meet the escalating demands of AI companies.

One solution being explored is the use of small modular nuclear reactors (SMRs) by companies like New Scale Power. These SMRs can generate carbon-free megawatts continuously, offering a potential solution for data centers to operate independently of the grid. New Scale Power is working with data center developer Standard Power to supply a significant number of SMRs that could collectively generate enough electricity for a midsize city. However, the deployment of these reactors is still pending regulatory approval and is not expected until late this decade.

The video highlights that AI’s energy intensity is driving the need for energy-dense systems like nuclear power. Data centers hosting AI applications consume a significant amount of electricity, with newer AI servers using much more power than previous models. This increase in power consumption also leads to higher heat generation, requiring more water for cooling. The global demand for AI may lead to significant water usage, with estimates showing AI-related water consumption surpassing the total annual water withdrawal of entire countries.

Various companies, including Microsoft and TerraPower (chaired by Bill Gates), are exploring the use of SMRs to power their AI data centers. Additionally, there are startups working on developing fusion-based nuclear reactors as a carbon-free energy source. While fusion-based systems hold promise, commercial viability may not be achieved until the 2030s. Overall, the video underscores the critical need for innovative energy solutions to support the growing demands of AI companies and their data centers.