Electrification of Everything

Answer: The amount of electricity required to run massive AI programs to compute, train, and operate these models is staggering. The increased demand for electricity will continue to grow as more non-techies like me explore ChatGPT, CoPilot, Perplexity, or Grok. AI is becoming the new buzzword as people give this technology a whirl. 

The introduction and adoption of AI reminds me of when handheld calculators came on the scene in the late 1960s. As a young girl at that time, my Grandpa would take me to work on Saturday mornings. We’d always stop at the bakery for donuts before going to the plant where I’d “help” him prepare the weekly bank deposit using an old-style adding machine. During that time, Grandpa gave me my first handheld calculator. The life-changing gift was an expensive ($149.95 in 1960’s dollars!), clunky and heavy, performing only four mathematical functions. 

In the 1970s, calculators entered classrooms, sparking debates whether students would lose their ability to perform mathematics by using this radical new device. The consensus was that students would use calculators regardless of what policies were in place. 

Fast forward to now, as my husband Pete and I discuss the pros and cons of AI use in the high school and college classrooms with our own grandchildren. Although AI use has already gone mainstream, the philosophical questions are ongoing. I  sympathize with our grandchildren and their educators as to the use of AI in educational settings. The grandkids do appreciate that they’ll need to have command of concepts to formulate their own thoughts as critical thinkers prior to engaging with AI. Fortunately, they see AI as a time-saving tool rather than a substitute for knowledge. 

AI is here to stay, and there’s no argument that it will increase electricity demand. A Department of Energy (DOE) report states that US data center power use will nearly triple by 2028, accounting for 12% of all U.S. electricity consumption and putting pressure on power grids and infrastructure. The DOE further states that data centers collectively use more electricity than some countries. For the first time in three decades, the demand for power in the U.S. is growing, driven by AI, the return of manufacturing via reshoring, and growth in electric vehicle use. 

The DOE forecasts that AI buildout and data center demand will cause growth in energy consumption of 2% to 3% per year through 2030. For comparison’s sake, energy demand has only increased on average by 0.4% between 2020 and 2023. Proof that these assumptions are accurate is shown by an increase in energy use of 3% alone in 2024. Shahar Levi, Co-Founder and CEO of LocusView, states that the 2050 estimate of annual global spending on power grid expansion and improvement will be $6 trillion. The DOE indicates that 30% of large, high voltage power lines and 60% of smaller, neighborhood distribution lines are due very soon for replacement.

We’ll need to consider incorporating the use of natural gas, wind, solar, and nuclear to meet growing needs while addressing the importance of modernizing our electric grid. Stanford University research shows that the U.S. leads the way in AI, with Washington, DC, as home to the most data centers of any metro region in the world. Beijing is number two, with Europe and emerging markets participating as well. 

Beyond philosophical debates and higher energy consumption, AI will also increase the need for a range of strategic materials. According to MIT, these include copper for electric cables, lithium, nickel, manganese, and cobalt for lithium-ion batteries, and steel for data centers. These industrial metals aren’t scarce, but supplies are concentrated geographically. A December 18, 2024, U.S. Geological Study shows that nearly a quarter of the world’s copper comes from Chile. Lithium is primarily found in Australia, and approximately 77% of the world’s supply of lithium-ion batteries are manufactured in China, where 54% of the world’s steel was produced in 2023. Kazakhstan produces nearly half the world’s uranium, and 5% comes from Russia. China dominates rare earth minerals, providing two-thirds of the global supply. 

The key message here is the importance of identifying price dynamics and resources for energy consumption. Demand for electric power is intensifying more rapidly with growing use of electric vehicles, the internet, and data centers powering AI. As we “electrify” everything, there’s corresponding growth in demand for key industrial metals such as steel, copper, and lithium, the majority of which are found outside the US. We’ve come a long way from that 1960s vintage handheld calculator Grandpa gave me! Stay focused and plan accordingly.