Nov. 21, 2025 — A new study found that Texas hosted more AI cluster sites than any other state in 2025, reflecting the rapid buildout of large computing facilities across the South.
Texas Royalty Brokers, a mineral-rights brokerage firm, analyzed 214 AI data centers and organized them into 83 GPU cluster locations. Each cluster is a concentrated site of graphics processors used to train and run artificial intelligence systems. The study measured chip counts, electricity needs, and computing capability using H100 equivalents, a benchmark that converts various processors into the performance level of Nvidia’s H100, one of the industry’s standard high-end AI chips.
Texas led the country with 17 clusters. The sites contained more than 811,000 AI chips with computing capacity equal to 6.6 million H100 equivalents. Texas generated 547 million megawatt hours of electricity last year, the highest total in the nation. The study said AI operations could consume about 15 percent of that output if fully utilized.
California ranked second with eight data centers and roughly 216,000 chips. Virginia placed third with five and about 163,000 chips. Illinois also had five clusters but with lower total computing capacity.
Tennessee held four clusters and had the highest chip count in the country at more than 1.27 million units. Ohio operated four clusters with roughly 500,000 chips and computing power equal to 1.26 million H100 equivalents, the third-highest total nationally.
Indiana, Iowa, Washington, and New Mexico each hosted three clusters. Indiana’s sites could consume nearly half of the state’s electricity output if running at full capacity, the highest proportion among the top ten.
The study said lower energy costs, available land, and reduced regulatory requirements helped make southern states the largest concentration of AI clusters in the United States, home to 26.