Data centres in the United States are projected to consume one-fifth of the country’s electricity by 2035, quadrupling current usage, according to a recent BloombergNEF report. This surge is largely attributed to the rapid expansion of artificial intelligence (AI) computing needs.
AI Compute Driving Capacity Growth
The report estimates data centre capacity will near 200 gigawatts within the next decade, with nearly half dedicated to AI training and inference workloads. The US is expected to dominate this demand, hosting 64% of AI chip power consumption by 2033. Some experts believe these figures may be conservative given the accelerating pace of data centre development.
Challenges for Power Grids
Many new data centres will connect to already strained electrical grids. For example, the PJM Interconnection, spanning Virginia to Illinois, could see 34% of its electricity consumed by data centres, while Texas’s ERCOT grid may allocate 22% of its generating capacity to these facilities.
PJM has faced difficulties balancing supply and demand, having paused new connection applications for four years before reopening in April. The imbalance has driven electricity prices up by 76% over the past year. Despite these challenges, data centres remain significant consumers, accounting for 38% of charges in PJM’s latest capacity auction.
Global Impact and Future Outlook
While the US leads in AI compute demand, data centre growth is global. If AI adoption continues aggressively, worldwide electricity demand from data centres could reach 1,935 terawatt-hours by 2033—comparable to India’s current annual consumption.
Businesses and operators must consider these energy demands when planning AI infrastructure. Solutions that improve efficiency and optimise workflows will be crucial. For practical AI adoption strategies and workflow automation, visit https://jasonjuul.com to explore relevant products and insights.
Disclaimer: This article provides an overview of projected energy trends related to data centres and AI compute. It does not offer medical or technical advice. Readers should consult specific experts for detailed implementation guidance.