This article was published by The Energy Mix on Nov. 22, 2024.
By Christopher Bonasia
Energy analysts and policymakers concerned about the growing electricity demand driven by data centres may find a solution in transforming the facilities into assets that support the grid, suggests a recent policy brief.
The promise lies in flexible demand powered by carbon-free electricity, states a report from the American Council for an Energy-Efficient Economy (ACEEE). Data centres can reduce their energy consumption during periods of high grid demand, ramp up usage during renewable energy surpluses, and store excess electricity to feed back into the grid.
Future electricity demand was already projected to rise in an energy transition where fossil fuel-powered technologies—like cars and furnaces—will be replaced by electrified alternatives like electric vehicles and heat pumps. But the projected demand has become more significant as a growing number of data centres in North America become more energy-intensive to support developing artificial intelligence technologies. According to the Canada Energy Regulator, global data centre power use is expected to increase 160% by 2030 because of AI.
While a traditional data centre consumes around 7.5 kilowatts per rack of servers, just one new server for high-performance AI tasks requires over 10 kW. “The power and heat density of a GenAI centre is at least four times that of a comparable cloud-computing facility,” ACEEE writes.
This is affecting regions seeking to encourage data centre development for economic gain. For instance in Ontario—which forecasts that 16 new data centres will be in service by 2035—new projections have electricity demand rising 70% by 2050, compared to a 60% forecast last year.
And in Virginia, which began attracting data centres in 2010, peak electricity usage of the state’s largest Utility—Dominion Energy—is expected to rise from almost 2.8 gigawatts in 2022 to roughly 13.3 gigawatts by 2038, reports Inside Climate News.
But, given the right incentives, those centres can help cut emissions and lower energy costs, ACEEE says. Instead of seeing AI data centres as a threat, “it is possible to transform them into valuable grid and regional assets” that can avoid renewable energy curtailment and defer the cost of grid expansions.
To accomplish this, regulators will have to encourage data centre developers to employ smart meters and other technologies to forecast and manage energy supply, or to invest in battery storage. Rate structures can also be adjusted to encourage data centres to reduce their energy consumption when grids are strained.
ACEEE adds that some companies are already developing technologies to do this. Enel X, for example, has developed uninterruptible power supplies, backup generators, and battery storage that can allow data centres to manage energy supply while participating in grid services.
“While new technologies like these show promise, using data centres as grid assets is still an emerging field,” writes ACEEE.
Be the first to comment