The energy needs of data centers continue to grow, seemingly unabated. According to this Utility Dive Article, a report from the National Electrical Manufacturers Association (NEMA) estimates that data center demand (and transportation electrification) will cause U.S. electricity demand to rise two percent year-over-year for the next quarter century. In the near term (the next decade), data center demand is projected to grow 300 percent. The Trump administration has been playing an active role in facilitating the growth of data centers and the global race for AI dominance. Following on President Trump’s Executive Orders from January 2025 on AI and U.S. energy security, the Department of Energy (DOE) recently announced plans to make its DOE-owned land at 16 locations available for data center development. They have since issued a request for information (RFI) to gather feedback from industry experts on how to use this land to further data center development, advocating in favor of a public-private partnership model.
While load growth is projected to be substantial, barriers remain. One barrier that is oft-cited is permitting. At the “Data Centers, AI and Nuclear Event” at the Broadband Breakfast event, panelists repeatedly cited the “sluggish” permitting process as a barrier to meeting the data center demand for power. Panelists predicted that Congress and the current administration would take steps to clear some barriers to permitting the large data center projects and the related permits needed to bring these large projects online, which currently can take 10-15 years. The panelists also discussed how many data center owners are interested in powering their data centers with clean energy. At present, solutions like advanced nuclear reactors simply are not commercially ready and likely will not be before 2030.
During this bridge period before advanced nuclear is commercially available, we frequently see data centers and the utility industry turning to natural gas as one of the only commercially viable baseload resources able to meet the high energy demands of data centers. But, the generation projects being proposed – and the price tags associated with building them – to meet this data center demand are not small, as exemplified by a recent proposal in Pennsylvania to spend $10 billion building an up to 4.5 GW gas-fired generation project at a retired coal plant. The developers of this project (which would be the largest gas-fired power plant in the nation) are proposing to co-locate this plant at a data center campus while also interconnecting the project to the larger transmission grid, which will allow the project to provide power beyond the data center campus. The project is slated to come online in 2027.
Co-location of large loads like data centers near power plants has been a frequent topic of conversation of late. In February 2025, the Federal Energy Regulatory Commission (FERC) launched a review of PJM Interconnection’s co-location rules. According to Utility Dive, FERC is seeking input on whether PJM’s co-location rules should be revised. These rules help to determine what grid costs should be paid by the co-located load. For example, if the generator at the co-located load trips offline and the data center relies upon the grid for power, the question is what costs the data center should pay, not only for making use of grid resources when the co-located generator is offline, but also for having the opportunity to do so. The outcome of this FERC rule-making proceeding will likely have a ripple effect on not just development of co-located load in PJM, but on other regional transmission organizations (RTO) and independent system operations (ISO) around the country. A decision from FERC is expected in the second half of 2025.
As the cited articles make clear, the demand for energy and capacity to power data centers continues, but challenges remain. State regulatory commissions around the country are also grappling with issues of cost allocation, trying to determine how to fairly allocate the costs of generation, distribution, and transmission resources being constructed to serve the massive data center demand. The tension in being first to AI dominance is at odds with who should pay for it. We expect these tensions will remain for the foreseeable future as stakeholders seek to answer the challenging questions associated with the massive energy demand of the data center industry.
