The scale of energy consumption by U.S. data centers is set to reach a fundamentally new level in the coming decades. According to my analysis of BloombergNEF data, by 2035 these facilities will consume approximately 20% of all electricity generated in the United States. This represents a fourfold increase compared to current levels — a true energy boom driven by the explosive growth of artificial intelligence infrastructure.

The key driver of this growth is AI computing. Over the next decade, the total capacity of data centers could approach 200 GW. For comparison, today this figure stands at around 50 GW. Such dynamics pose unprecedented challenges for the U.S. power grid — from the need to modernize networks to finding new sources of generation.

Particular attention should be paid to the geographical distribution of demand. By 2033, the United States will account for 64% of global demand for AI chips by capacity. This underscores the dominant role of the American market in the global race for computing resources, but simultaneously creates risks of overloading local power grids, especially in hubs such as Northern Virginia.

From my professional perspective, this trend will inevitably lead to increased investment in renewable energy and nuclear power, as well as stricter regulation of data center energy efficiency. Investors should pay attention to companies involved in energy infrastructure and cooling — these sectors will become beneficiaries of the upcoming energy boom.