A recent report from 404 Media reveals that a small research team is meticulously mapping the expansion of artificial intelligence datacenters across the United States. Using publicly available information and satellite imagery, Epoch AI, a non-profit research institute, is tracking facilities that often evade public discourse. Their work aims to provide insights into the scale and speed of AI development, shedding light on an industry that is growing rapidly.
Mapping Infrastructure and Community Impact
Datacenter construction has emerged as a contentious issue nationwide. These facilities require significant amounts of electricity and water, and many communities only become aware of their existence after construction has commenced. To address this gap, Epoch AI has developed an interactive map that visually marks known sites. Each marker connects to satellite views and detailed project information. For instance, a marker over New Albany, Ohio, identifies Meta’s “Prometheus” datacenter complex, which Epoch AI estimates has cost approximately $18 billion and consumes 691 megawatts of power.
Epoch AI describes this complex as a combination of weatherproof tents, colocation facilities, and traditional datacenter buildings, reflecting Meta’s strategic shift towards AI development. The map allows users to scroll through a timeline, observing the growth of the complex through satellite images that display the addition of new buildings and cooling systems over time.
Understanding Energy Consumption
A significant portion of Epoch AI’s analysis concentrates on cooling infrastructure, which is critical for modern AI systems that generate substantial heat. The researchers note that cooling units are often installed outside buildings or on rooftops. According to Epoch AI, “Modern AI datacenters generate so much heat that the cooling equipment extends outside the buildings.”
The team meticulously counts fans, measures their sizes, and analyzes their placements. These details are then incorporated into a custom model to estimate energy usage, which in turn helps infer compute capacity and construction costs. “We focus on cooling because it’s a very useful clue for figuring out the power consumption,” stated Jean-Stanislas Denain, a senior researcher at Epoch AI. Nevertheless, the model carries some uncertainty; variations in fan speed and configuration mean actual cooling capacity could deviate significantly from estimates.
Despite its thoroughness, the map is not exhaustive. Variations in state and local disclosure laws mean that some projects remain unreported, and smaller facilities frequently escape detection. Epoch AI estimates that the current dataset represents roughly 15 percent of global AI compute provided by chipmakers as of November 2025.
Markers on the map indicate various projects across the country, including one near Memphis, Tennessee, which points to xAI’s Colossus 2 project. The organization has reportedly installed natural gas turbines across the Mississippi border to expedite approval processes. Epoch AI notes that around 110,000 NVIDIA GB200 GPUs are now operational at this site.
The mapping project seeks to illuminate infrastructure that is shaping the economic landscape, often without adequate public awareness. Epoch AI acknowledges that even with detailed mapping, gaps remain. “Even if we have a perfect analysis of a datacenter, we may still be in the dark about who uses it, and how much they use,” the organization remarked.
As Epoch AI plans to extend its research globally, the initiative aims to provide greater transparency in an industry that is pivotal to the future of technology and the economy.
