House to investigate Puerto Rico’s energy capacity for AI

Rep. José Aponte Hernández

By THE STAR STAFF

Rep. José Aponte Hernández has joined Rep. Ángel Morey Noble to spearhead an investigation into Puerto Rico’s power generation capacity and the measures being taken to address soaring electricity demand driven by artificial intelligence (AI) technologies.

“Since 2021, we have worked on the issue of AI algorithms, their potential, and the challenges associated with their use. By 2025, the AI industry generated over $173.56 billion in economic activity in the United States. In fact, 14.1% of the nation’s Gross National Product ($254.4 billion) for the last quarter of 2025 is closely tied to AI products,” Aponte said Thursday. “Alongside this historic surge in AI adoption, electricity consumption has increased to levels never seen before with any technological innovation. Our goal is to evaluate whether we have the generation capacity to meet the demand that is already on the horizon.”

According to U.S. Department of Energy data, electricity consumption by AI platforms, such as data centers, surpassed 183 terawatt-hours in 2024, representing 4.4% of total U.S. electricity use. That figure exceeds the entire energy consumption of countries such as Pakistan, Poland, Norway, the Philippines, Colombia and Argentina.

Morey emphasized the urgency of the issue, noting that he filed House Resolution 488, which directs the House Government Committee to investigate the capacity of each power generation plant on the island and assess projected electricity demand over the next five years, factoring in potential increases from AI-driven data centers.

“We are fully aware of this matter and the need to examine whether our infrastructure can handle the load,” Morey said. Aponte signed on as co-author of the measure earlier Thursday.

AI technologies -- particularly large language models, or LLMs -- require massive computing resources. Training those models involves thousands of graphics processing units, or GPUs, running continuously for months, consuming enormous amounts of electricity. Experts warn that by 2030-2035, data centers could account for 20.4% of global electricity consumption, an unprecedented figure that would place immense pressure on power grids worldwide.

Several U.S. states, including Virginia, Oregon, Washington and California, are already assessing their electrical systems to identify strategies for reducing energy use by AI computing clusters. California, for example, is exploring ways to maximize transmission and distribution networks and prioritize generation infrastructure in high-demand areas.