Key Highlights:
- New research from the MIT Future Energy Systems Center suggests that data centers that adjust their workloads can reduce pressure on the electric grid and save money for consumers. However, this may lead to higher emissions from power plants in some areas.
- The study noted that load shifting in data centers can help with the incorporation of renewable energy and better use of existing energy sources. The emissions impact varies based on which factor is more dominant. Aligning data center operations with renewable energy deployment is crucial.
- As data center electricity demand is expected to grow, utilities and grid operators are seeking resources and strategies to manage this new load. Predictions indicate that AI data centers could consume around 34 GW, which is roughly 3% of the U.S. generating capacity by 2030.
Insights:
Experts point out that not every proposed data center will be built, making future power demands uncertain. Still, they acknowledge that significant growth is expected, and adjustable power profiles can help mitigate stress on the grid.
A RAND Corporation forecast for 2024 estimates a demand of 130 GW from data centers in the U.S. by 2030, while the U.S. Department of Energy stated recently that an additional 100 GW of peak capacity will be required in the same timeframe.
Christopher Knittel, MIT’s associate dean for climate and sustainability, emphasized the urgent need to rethink grid management and adopt more flexible data center operations, especially for AI training.
According to the research, data centers typically operate at around 80% capacity, leaving 20% available for load shifting to periods when renewable energy supply is high or electricity prices are lower. This can lead to cost savings for data centers and enhance reliability for the power system while advancing climate goals.
The study analyzed data center demand flexibility across three U.S. power markets: the Mid-Atlantic, Texas, and the Western Electricity Coordinating Council (WECC), covering 14 states and part of Canada and Mexico. The findings suggested that flexibility in data centers could lower total system costs by an average of 3.7% across these regions.
To navigate the potential challenges posed by data centers, policymakers are encouraged to incentivize or enforce flexible operations through dynamic pricing, demand response programs, or load-shifting incentives.
In regions with abundant renewable energy and limited coal usage, flexible data centers can significantly reduce emissions. In Texas, modeling showed emissions could decrease by up to 40%. However, this reduction was not observed in areas with higher coal and gas usage.
In the Mid-Atlantic and WECC, increased reliance on coal generation was noted, diminishing the need for flexible natural gas capacities.
MIL’s analysis indicated that if data centers fully utilized their flexibility, coal utilization in the Mid-Atlantic could rise from 50% to 59%, impacting overall emissions. The effect of data center adaptability is not solely about flexibility but also depends on the local energy resources and investment environment.