A Force For Good: How Data Centers Can Ease The Strain On The U.S. Power Grid

Data centers are intensive energy consumers, expected to represent 12% of total U.S. electricity consumption by 2030 — largely due to constant innovations in the artificial intelligence space.

This could be enough to upend the U.S. power grid, a system already under pressure from the massive push for electrification of vehicles and other decarbonization efforts. 

The U.S. power grid doesn't have infinite supply, and grid upgrades are tremendously expensive. The question, then, becomes whether it is possible for big energy users such as data centers to give back to the grid instead of just consuming. If so, how?

“Data centers will need to be flexible if we're going to keep all of this innovation in the U.S.,” said David Chernis, director of flexible compute platforms at CPower, a Baltimore-based virtual power plant platform. “How to accommodate the seemingly vertical growth we’re witnessing is the industry’s No. 1 problem to solve.”

According to a study from Duke University’s Nicholas Institute for Energy, Environment & Sustainability, the U.S. grid already has the ability to absorb a considerable amount of energy that data centers are now requiring without the immediate need to build new power plants — as long as the demand is flexible. 

The study revealed that 76 gigawatts of new energy load could be integrated into the system if large load users such as data centers curtail their usage just 0.25% of the time. One hundred gigawatts of new energy load could be integrated if they curtailed their usage by 0.5%, and 126 GW could be integrated if usage is curtailed 1% of the time. 

“This paper was like a shot heard around the world,” Chernis said. “If we could find even 0.25% flexibility in our power usage, we’ll have the capacity to build everything we’re talking about.”

Traditionally, data centers operate as real estate companies, not energy companies. They don’t have a lot of legacy dealing with the energy side of the equation in terms of demand response. They are more focused on running and renting out space for server racks with a mission of 100% uptime because servers often host live apps and websites that immediately impact the consumer. There is little room for flexibility here, Chernis said. 

On the other hand, not all servers host live web data, leaving room for some usage flexibility. 

As bitcoin mining has become more popular, it is estimated that cryptocurrency-related energy usage now makes up 2.3% of total U.S. energy consumption. These servers don’t host anything live, so the consumer isn't impacted directly.

“They can shed all of their load in seconds to minutes,” Chernis said. “We see these mining sites as grid assets, because we can take so much load off the grid in real time. There are gigawatts of mining and high-performance computing load in the U.S. that can act like a digital battery by shutting off for a brief period and freeing up power for the grid.”

Flexibility in power usage will be the norm moving forward in this industry, Chernis added. Other ways for data centers to be responsible power consumers, in addition to leveraging computing flexibility, are the implementation of on-site power generation as well as cooling system upgrades and optimization measures. 

States like Texas are at the forefront of this trend, passing new legislation to deal with heavy electricity consumers and their rapidly growing demand, Chernis said. 

With the goal of reducing strain on the grid, Texas Senate Bill 6 establishes that large load users must have a backup power source and be able to reduce load capacity during emergencies. It also mandates that these users pay study and interconnection fees and contribute to grid upgrades to accommodate their increasing consumption. More states are expected to follow in Texas’ footsteps, Chernis said.

“The trend seems to be if you're a data center and you're trying to interconnect, you should bring a plan to participate in demand response programs, where users shift electricity use in exchange for compensation or implement some sort of flexibility measure, such as on-site power generation or cooling improvements,” he said. “It’s about being a partner with the grid instead of being a burden.”

Not only is power usage flexibility beneficial for the grid, but it also stimulates the data center’s local economy. 

“It can stabilize the grid, thus keeping the rest of the local economy growing,” Chernis said. “You’re supporting productivity gains in the local area, and that inherently provides a cost benefit. It helps to keep local energy and utilities affordable for surrounding residents and businesses.”

There are several barriers, however, to data centers becoming true partners to the grid, Chernis said. The biggest? Aversion to risk. 

“There’s so much resistance from traditional data centers to try anything new because there’s already so much risk in their business,” he said. “This is the general industry sentiment, even though we have the tools to get us to a place where we can be flexible energy users.”

The second barrier is the speed of power interconnection. Right now, it is estimated to take an average of five to eight years to get connected to the grid. This long wait hinders a data center’s ability to balance supply and demand, Chernis said.

CPower, as a “virtual power plant” platform, helps clients allocate their energy more efficiently through demand response programs. The firm also supports grid stability by coordinating shifts in power consumption across nearly 23,000 sites.

“This is in our DNA. It's our core strength,” Chernis said. “Think of us as strategic energy advisers: We assess your capabilities and constraints, then tailor monetization opportunities that fit so customers are not leaving money on the table. We also guide customers through regulatory complexities to ensure their optimal performance.” 

Chernis said the data center industry will continue to experience exciting innovations, especially with the increasing use of artificial intelligence. 

“AI is creating a surge in vertical demand, which presents real challenges. But we’ve already solved this by implementing flexible workload strategies that adapt to shifting needs,” he said. “Moreover, AI isn’t just part of the problem. It’s central to the solution. We're using it to optimize every aspect of our work, and it’s helped utilities with everything from predicting telephone poles that are about to snap off to intelligently managing load across the grid.”

Get ahead of Texas’ SB 6: Join CPower’s webinar on Sept. 25.

Texas Senate Bill 6 is reshaping the energy landscape for large-load customers with new rules around interconnection, backup generation and load curtailment. Join the webinar to discover how you can turn compliance into opportunity and gain a competitive advantage by leveraging your flexible energy in demand response programs.

This article was produced in collaboration between Studio B and CPower. Bisnow news staff was not involved in the production of this content.

Studio B is Bisnow’s in-house content and design studio. To learn more about how Studio B can help your team, reach out to studio@bisnow.com.