Payne, Jonathan EdgarMagnano, Chase N.2025-07-292025-07-292025-05-10https://theses-dissertations.princeton.edu/handle/88435/dsp01rf55zc13rThe rapid development of artificial intelligence has ushered in a new era of energy demand growth in the United States. This paper seeks to establish a quantitative link between the increase of artificial intelligence computing and energy consumption Data Center Alley - the highest concentration of data centers in the world - in northern Virginia. Energy data are obtained directly from the Pennsylvania-New Jersey-Maryland (PJM) Interconnection Regional Transmission Organization (RTO). Geocoded and standard differences-in-differences analyses are performed to examine this link. There is a significant positive connection between the release of new models of ChatGPT and congestion energy prices surrounding high concentrations of data centers. The clearest pricing results coincide with the release of GPT-o1. Carbon emissions were also studied, but the results were inconclusive. There is also an increase in energy consumption found following the release of new GPT models, but this is not significant at the 5% level. This increase is likely connected to the surge in prices seen within Dominion Energy. These results are used to inform PyPSA-USA, an open-source grid modeling software, to obtain estimates for needed future capacity expansion in 2030, 2040, and 2050. The results from this model suggest that Dominion Energy needs 85 GW of new capacity expansion by 2040 but currently only plans to build 33 GW. These results have implications for energy prices in areas with high concentrations of data centers and the future ability of the electricity grid to meet demand in PJM and beyond.en-USPrinceton University Senior Theses