The AI boom has given way to a rush to build more 24/7 baseload power, regardless of whether it’s gas, nuclear, or geothermal. But Eric Gimon, a consultant with the think tank Energy Innovation, thinks this impulse to build is based on flawed assumptions.
“The idea that a new data center is a 24/7 block, and that I’m going to need a power plant that’s also 24/7 matching up to it, doesn’t make sense to us,” he told Latitude Media, while discussing his new report on how to move past what he calls the industry’s “firm fixation.”
For one, the concept of baseload is complicated by the fact that power plants and data centers are part of the larger interconnected system. Individual plants frequently go offline for maintenance or outages, and it’s the grid as a whole that ensures customers receive reliable, around-the-clock power.
Additionally, thinking that only new firm resources and costly transmission upgrades can provide data centers with their required capacity doesn’t take into account the complex way they’re built and operated.
Data center consumption profiles are choppy, with swings of hundreds of megawatts over short intervals. “And these massive steam turbines that run off nuclear plants don’t want to go up and down in seconds or even half hours,” Gimon said. Data centers are built in phases, a process that aligns “poorly with ‘lumpy’ firm large one-time investments in dispatchable power plants and infrastructure upgrades,” as the report notes.
Data center developments also tend to be built in clusters, which risks amplifying grid stress even after dedicated firm power sources and transmission upgrades have been implemented. Their energy strategy is further complicated by the involvement of many actors, including developers, operators, and tenants.
According to Gimon, single sources of firm power are not suited for this level of complexity. “You don’t want a single power plant — it is not the right solution for meeting your needs. You want to solve this with portfolios,” he said, referring to clean energy assets, advanced transmission technologies, and demand-side flexibility.
The importance of storage
Battery energy storage is the lynchpin of these portfolios: “the thread that snuck through all the things,” according to Gimon.
As the report notes, “data centers are not ‘a perfect baseload’ fit to directly couple with large mechanical generators or even the grid, and they… need significant electrical equipment to buffer this connection and prevent extra wear and tear on co-located generation or nearby grid users.” Incorporating battery energy storage is therefore likely to become “a key element in managing data center impacts on the grid,” and an essential support for flexibility solutions such as curtailment and baseload shifting.
This is especially true for co-located data centers with multiple tenants, which — unlike hyperscalers — have limited control over their customers’ willingness to be flexible, and might refuse to curtail their loads for fear of scaring tenants off.
These potential benefits add to the fact that when the grid is powered by wind and solar, front-of-the-meter storage helps balance those fluctuations, easing the integration of data centers.
“Because batteries are increasingly essential for buffering, backup, and power quality, they also provide a built-in solution for integrating variable renewables… [and] can take advantage of existing surplus interconnection to more quickly connect data centers to the grid in ‘power couples,’” the report says. (Data centers are also beginning to embrace behind-the-meter, onsite batteries for data centers, as panelists discussed at the DERVOS conference last month.)
“[Storage] is a two-sided resource,” Gimon said. “On one hand, it helps you build better data centers, and on the other hand, it makes it easier to integrate with the cheapest, best resources out there.”
Ideally, these portfolios would be located in energy parks that co-locate multiple renewable energy sources and storage solutions with data centers — all connected to the grid at a single point — something Gimon outlined in a dedicated report last year. But that would require active participation from data centers, which aside from a few hyperscalers are mostly reluctant to get into the energy business. “Data centers are not excited about being tied at the hip to an energy asset that they have to manage… they don’t want to be dealing with this stuff and want somebody else to deal with it,” Gimon said.
Interest, however, has been increasing. And the fact that data centers are “willing to pay and are in a hurry” to get power means that utilities have some leverage, concentrated at the moment of interconnection.
“It is the moment to ensure consumption tariffs reflect cost causation, encourage flexibility, and align incentives without imposing unworkable burdens later,” the report says. “Interconnection is the moment of maximum leverage.”
A version of this story was published in the AI-Energy Nexus newsletter on November 5. Subscribe to get pieces like this — plus expert analysis, original reporting, and curated resources — in your inbox every Wednesday.
