Google is funding a new virtual power plant (VPP) in the US East Coast's largest power grid, PJM, through a deal with Voltus. The VPP will aggregate distributed energy resources like EVs and smart thermostats, paying participants to reduce power use during peak demand. This initiative aims to provide extra capacity to power Google's regional data centers, addressing the growing energy needs of AI infrastructure.
Google has entered into an agreement with Voltus, a virtual power plant (VPP) and distributed energy resources platform, to fund a new VPP in the PJM grid, which serves a significant portion of the US East Coast. This initiative aims to address the substantial energy demands of data centers, particularly those supporting AI technologies. The VPP will aggregate devices such as electric vehicles and smart thermostats from participating customers, who will be compensated for reducing their electricity usage or utilizing stored energy during peak demand periods when the grid is stressed. Google will finance the setup of this VPP, with the intention of utilizing the generated capacity to power its own data centers in the region. Voltus plans to aggregate up to 100 megawatts of distributed energy resources annually through this program, which is slated to become operational in 2027. This move highlights a growing trend of tech companies seeking flexible energy solutions to manage their significant power consumption. The article also touches upon regulatory approaches, such as proposals to expedite data center approvals if they agree to demand curtailment, and a new Texas law requiring large energy users to switch to backup power or reduce demand in emergencies. It references a Duke University study suggesting that data centers could reduce their energy demand for about 40 hours per year without necessitating new power plants or transmission infrastructure. However, questions remain regarding the incentives needed to encourage participation from both consumers and data centers, especially given the immediate revenue implications for data centers that might delay computing tasks. A California study on EV charging programs indicated low enrollment rates even with financial incentives, suggesting potential challenges in securing widespread participation for VPPs. Gallup polling also noted significant public opposition to AI data centers in local areas.
This initiative represents a concrete step by a major tech company to address the escalating energy demands of data centers, particularly for AI, by leveraging virtual power plants and distributed energy resources. It highlights potential solutions for grid stability and capacity expansion, while also underscoring challenges related to participant incentives and public acceptance.