Skip to content
Don't miss

Get the daily Cyber Briefing in your inbox

SIGN UP
The Big Eddy-Knight Transmission Project, seen under construction on June 17, 2015, runs between Wasco County, Ore., and Klickitat County, Wash., to meet increased demand from a hot spot for energy-intensive data centers. (Department of Energy)
Articles |

Artificial intelligence's thirst for power demands greater focus on cybersecurity of the energy sector

By Frank Cilluffo and Kyle D. Klein

Behind the scenes of every new artificial intelligence (AI) app and program is a staggering amount of energy needed to power its data centers. We’re already in the throes of the AI revolution and with unprecedented energy demand comes the increased need for proactively protecting the energy sector from cyber adversaries like the People’s Republic of China (PRC).

The increased demand for power is coming quickly. Today, AI accounts for just 14% of global data center electricity demand, but that will double in the next two years alone, according to Goldman Sachs. By the end of the decade, we will need a whopping 165 times as much energy to support the explosion of AI, growth that comes with an estimated cost of $720 billion in electric utility investment.

Let’s break down how we get to such explosive energy demand. Today, a single query into ChatGPT uses about 10 times as much electricity as a simple, traditional Google search. A server rack at a traditional data center uses about 7 kW of electricity, but an AI server rack consumes 30 kW to 100 kW, anywhere from a three- to fourteen-fold increase. And by 2035, AI data centers are estimated to be using the same amount of electricity as all of the households of New York, California, Texas, and Florida combined.

Related Content