A recent report from 404 Media reveals that a small research team is diligently mapping the swift growth of datacenter infrastructure in the United States. Utilizing publicly available information and satellite imagery, the team from Epoch AI, a non-profit research institute, is shedding light on facilities that often escape public debate. Their efforts aim to provide a clearer picture of how artificial intelligence (AI) is expanding across the nation.
Epoch AI employs open-source intelligence to identify, analyze, and document the surge of datacenters. By examining satellite images, construction permits, and local regulatory documents, the researchers have developed an interactive map that estimates costs, ownership, and power consumption of these facilities. This project offers crucial insights into an industry that is growing faster than public scrutiny can keep pace with.
Mapping the Infrastructure
The construction of datacenters has emerged as a contentious issue nationwide. These facilities are notorious for their massive demands for electricity and water. Often, communities discover their existence only after construction has commenced. The interactive map created by Epoch AI features visual markers over identified sites, linking to satellite images and project specifics. For instance, a green circle on the map denotes Meta’s “Prometheus” datacenter complex located in New Albany, Ohio. Epoch AI estimates that this project has cost approximately $18 billion and draws 691 megawatts of power.
Epoch AI describes the complex as a combination of weatherproof tents, colocation facilities, and traditional datacenter buildings, highlighting Meta’s shift towards AI. Users can explore a timeline showcasing the site’s development, observing the addition of new structures and cooling systems over time.
Estimating Power Consumption
Much of Epoch AI’s analysis centers on cooling infrastructure, which is critical due to the intense heat generated by modern AI systems. Datacenters often position cooling units outside buildings or on rooftops to manage this heat. As noted on their website, “Modern AI datacenters generate so much heat that the cooling equipment extends outside the buildings.”
To estimate energy usage, the Epoch AI team counts fans, measures their dimensions, and assesses their placement. These details are fed into a custom model to approximate energy consumption, which in turn helps to infer computing capacity and construction costs. Jean-Stanislas Denain, a senior researcher at Epoch AI, explained, “We focus on cooling because it’s a very useful clue for figuring out the power consumption.” However, the model does carry some uncertainty, as fan speed and configuration can vary significantly. The team acknowledges that real cooling capacity could be much higher or lower than their estimates.
Despite their advancements, the mapping project is not exhaustive. Variations in state and local disclosure laws mean that some projects may remain undisclosed or receive little publicity. Smaller facilities frequently go unnoticed. Epoch AI estimates that their current dataset represents about 15 percent of the global AI computing power provided by chipmakers as of November 2025.
The map reveals various markers across the country, including one near Memphis, Tennessee, which points to xAI’s Colossus 2 project. Epoch AI notes that the company has installed natural gas turbines across the Mississippi border, likely to expedite approval processes. Based on this information and earlier tweets from Elon Musk, it is estimated that 110,000 NVIDIA GB200 GPUs are operational.
Even with detailed mapping, blind spots persist. Epoch AI acknowledges, “Even if we have a perfect analysis of a data center, we may still be in the dark about who uses it and how much they use.” Looking ahead, Epoch AI plans to extend its mapping efforts globally. The project aspires to illuminate the infrastructure that is shaping the future economy, often without public visibility.
