The US Energy Department published a set of recommendations for meeting new electricity demand from datacenters, noting that existing federal incentives can help power system operators address challenges posed by the rise of artificial intelligence.
The recommendations, published Aug. 12, come ahead of a highly anticipated report by the DOE's Lawrence Berkeley National Laboratory on current and near-future datacenter energy consumption and water use, due at the end of 2024.
Lawrence Berkeley's most recent congressionally mandated report on datacenters, submitted in 2016, found that they accounted for approximately 1.8% of US electricity consumption in 2014. That figure is now significantly out of date.
A report released in May by the Electric Power Research Institute found that datacenters could account for 9.1% of the nation's electricity demand by the end of the decade, up from 4% of total US load in 2023. The report estimated that datacenters consumed about 25% of Virginia's electricity production in 2023 and accounted for more than 10% of demand in North Dakota, Iowa, Nebraska and Oregon.
On its second-quarter earnings call earlier in August, PPL Corp. executives noted that the company's Pennsylvania service territory has nearly 5 GW of electricity demand from datacenters in advanced development, with 17 GW of interconnection requests from datacenter customers in the interconnection queue for 2026 through 2033. A number of other utilities during their second-quarter calls remarked on expected sharp increases in demand triggered by datacenters and other large-load customers.
The Energy Act of 2020 directed Lawrence Berkeley to update its 2016 analysis, specifically requiring the lab to evaluate the impact of cloud platforms and "big data" on energy usage.
With the Lawrence Berkeley update pending, the DOE on Aug. 12 highlighted multiple tools to address growing electricity demand from datacenters.
The Inflation Reduction Act's new and modified clean energy production and investment tax credits can facilitate cost-effective investments in clean power generation to meet demand from large datacenter customers, the DOE noted.
The DOE also highlighted $10.5 billion in grid resilience and innovation funding available through the 2021 bipartisan infrastructure law for grid-enhancing technologies enabled by artificial intelligence, such as dynamic line ratings and grid topology optimization.
Maximizing energy efficiency will also be crucial, the DOE said. The DOE noted that it is leading a scaling initiative that aims to increase the energy efficiency of the microelectronics needed for computation at datacenters by a factor of 1,000 over the next two decades. Datacenter owners can also qualify for the Inflation Reduction Act's expanded 179D tax credit for energy-efficient commercial buildings, the DOE said.
On the demand side, the DOE noted that datacenters can operate as controllable load resources during periods of peak demand.
In January, the DOE provided feedback on 18 concept papers submitted in response to $50 million in funding for grid integration projects that demonstrate coordinated control of industrial buildings. The DOE plans to select two to four projects that, among other requirements, utilize distributed energy resources with an aggregated capacity of 25% of the grid system's peak load.
Finally, the DOE listed a range of "key enablers," including generator interconnection queue reform and innovative tariffs to support datacenter development.
"Near-term data center driven electricity demand growth is an opportunity to accelerate the build out of clean energy solutions, improve demand flexibility, and modernize the grid while maintaining affordability," the DOE said.