If AI data centers were more flexible in the timing of their power consumption, electricity prices could be lower, according to a new MIT study. The researchers wrote about their study and findings in the peer-reviewed journal iScience (citation below).
Artificial intelligence (AI) is driving accelerating growth in US data centers, which raises concerns about electricity demand and emissions. In this context, “emissions” does not refer to what data centers themselves emit, but rather to greenhouse gases such as CO₂, produced by electricity generation. Coal-fired and natural gas-fired power stations emit greenhouse gases, which are the main cause of global warming.
Effect on the energy grid
Their rapidly increasing energy consumption also has raised concerns about their impact on the energy grid. What would happen to the country’s energy grid if hundreds of new data centers came online?
MIT researchers carried out a study on data centers and the country’s electricity supply and prices. They found that their impact could vary significantly, depending on how their energy consumption is structured.
Data centers might help reduce average energy costs if they consumed more electricity during non-peak hours. The environmental effect would depend on where the data center is. In some areas, more energy would be produced from renewable sources such as solar or wind power, while others would burn fossil fuels. “Fossil fuels” include natural gas, coal, and oil, which contribute to global warming when burned.
Study researcher and co-author, Christopher Knittel, an economist in the MIT Sloan School of Management, said:
“The key with data centers is: How can we add them to the network without adding a lot to our peak usage? One way for data centers to do that — to add to average usage but not the peak usage — is if they provide some grid flexibility during those high-cost periods. And that’s what we’ve been interested in understanding.”
Consume more energy during off-peak times
The researchers explained that a more flexible arrangement for energy consumption would produce the following cost savings:
- Texas: Up to 5%
- Mid-Atlantic region: Up to 4%
- Western U.S. states: Up to 2%
These savings would only be possible if data centers shifted over 20% of their electricity consumption to off-peak hours. In some cases, there would need to be a 50% shift.
Citation:
Senga, J.R.L., Wang, S., and Knittel, C.R. (2026). Flexible Data Centers Reduce Power System Costs but Can Increase Emissions. iScience. DOI: 10.1016/j.isci.2026.116497.
