AI is so energy hungry that American utilities companies must upgrade their capacity and networks to meet growing demand.
That's according to consultancy Bain & Company, which warned that demand for power could soon be outstripped by supply in the US, home to many of the biggest AI developers.
The warning follows Amazon investing in a nuclear-powered data center and Google admitting its emissions climbed by 48% over the last five years, largely driven by energy-hungry AI.
Utilities aren't the only ones being told they may have to invest their own funds to keep up with AI, with calls for more data centers too — and those will naturally need more power.
Surging demand
According to a Bain & Company briefing note, meeting the future power demands of data centers will require more than $2 trillion in investment in new energy generation.
In order for the US to meet demand, it will need to boost generation by as much as 26% by 2028 — a far cry above the 5% in annual generation growth seen by US utilities over the last decade or so.
Bain & Company admits that the 26% climb is the "high end" scenario, with 7% at the lower end. However, the rising demand in energy for data centers is exacerbated by other technologies, the consultancy added.
"The late 2022 breakthrough in generative AI and the ensuing data center boom blindsided utilities just as demand was also rising because of repatriated manufacturing, industrial policy, and vehicle electrification," the briefing note says.
Despite these areas of growth, the consultancy predicted that data centers will account for the largest share of new electricity demand growth over the next five years, making up 44% versus 27% for residential. Beyond energy generation, Bain & Company also noted grid infrastructure may need improving.
Not just the US
It's a worldwide concern, the consultancy added. The annual global energy consumption for data centers could more than double by 2027 from 2023 levels, growing at a compound annual rate of 10% to 24% and potentially surpassing 1 million gigawatt hours in 2027.
"These facilities need eye-popping amounts of power. Serving a 1-gigawatt data center requires the capacity of about four natural gas plants or around half of a large nuclear plant."
Indeed, the UK was warned in March that it needed "bold action" to ensure its power networks could keep up with demand, with the National Grid forecasting this would climb six-fold over the next decade because of data center growth, fueled by AI but also the rise of quantum computing.
Paying for it all could prove a problem, with utilities forecast to require up to 19% more in revenue each year, which could lead to higher bills for all customers.
Without investment in new generation sources, utilities risk failing to supply the necessary power but also missing out on the opportunity, Bain & Company added, advising utilities to move quickly to take advantage of the AI boom.
