Australia's government is reportedly the latest nation that experts predict will suffer from the growing demand for power from artificial intelligence data centers, straining supply and increasing prices, according to analysts at investment bank UBS.
According to Tom Allen, an analyst at UBS, the demand, partly fueled by the expansion of energy-hungry artificial intelligence services, might require between 3.3 and 5 gigawatts of new capacity by 2030, or approximately 15% of the total load.
Data centers account for over 25% of Australia's major industrial electricity demand. UBS predicts that the load will climb by 16% annually until 2030.
(Photo : JOSEP LAGO/AFP via Getty Images)
Evening peak demand could cause wholesale electricity price margins to increase by up to 70% by 2030. Price gaps between 3.30 pm and 8.30 pm in the national energy market have averaged $312 per megawatt-hour this year, with UBS estimating they will reach $525/MWh by the end of the decade.
The Australian Energy Market Operator has mostly missed the surge in data center demand in its estimates, including last month's biannual integrated system plan (ISP) release. Allen predicts that the next edition of the ISP will see higher load growth from this burgeoning and major customer segment.
Read Also: Can Hardware Development Keep Up with AI's Skyrocketing Capabilities?
Underestimated AI Power Demand
Australia is not the only country that underestimates AI's power needs. CoreWeave, a worldwide cloud computing provider, recently said that AI's demand for additional computing power, data centers, and electricity is "underestimated."
This miscalculation will create stress and limit the world's electricity grid infrastructure.
Brian Venturo, the co-founder of CoreWeave, made the forecast. He believes that the data center demands received by CoreWeave daily are "absurd," with some seeking whole campuses.
According to the co-founder, the supply chains that historically supported highly physical firms are not intended to accommodate the market's rapid growth.
He predicts that when there are more "megacampuses," electrical networks will be pressured and political tensions will rise.
The emergence of generative AI is boosting the need for data centers, placing Roseland, New Jersey-based CoreWeave among the most popular IT firms.
Venturo voiced concern about the infrastructure required to satisfy demand, adding, "It worries me." According to Venturo, AI's power consumption is comparable to a sprint requiring all the world's money.
He also mentions that the difficulties of rehabilitating old data centers are pushing demand for new ones. Given the present infrastructure that powers the data centers, such additions are not conceivable.
Natural Gas for AI Energy
Experts believe that natural gas producers will need to step up to meet this demand. The CEO of pipeline operator Kinder Morgan, Richard Kinder, claims that big tech powers these data centers and that businesses are beginning to recognize the advantages of nuclear and natural gas.
With 40% of the market, Kinder Morgan is the US's biggest natural gas pipeline operator.
Related Article: Google's Climate Change Goals Fall as Emissions Blow Up Due to AI