Hyperscale and wholesale data center construction is increasing the average data center size.
Data centers are typically built on one story to more easily manage heat, water, and weight distribution.
Most data centers are built in large metropolitan areas for faster connection to users, but AI models can be trained in rural facilities.
Data centers are facilities that contain computer servers and provide storage and online services (Lawson et al. 2026). They are categorized by their operations and who they serve. Operations typically correlate with facility size, but exact definitions and thresholds vary (Ferreira et al. 2026, Shehabi et al. 2024). The average size of new data centers in the U.S. has increased, nearly doubling from about 100,000 square feet (ft2) in 2016 to almost 200,000 ft2 in 2025, and is expected to continue to increase through 2030 (Ferreira et al. 2026). This growth is driven by more construction of large data centers.
Small data centers are used to support telecommunications networks, provide service to a single enterprise, or lease server space to retail customers. They can range in size from a small closet up to several thousand ft2.
There are two main types of large data centers (Ferreira et al. 2026, Shehabi et al. 2024). Wholesale data centers sell large portions of server space to a few customers. Hyperscale data centers are typically owned and operated by a single large technology company for cloud or Artificial Intelligence (AI) services. One common definition for hyperscale data centers is a minimum of 5,000 servers and 10,000 ft2 of physical space (Powell & Smalley n.d.). However, wholesale and hyperscale data centers can be hundreds of thousands of ft2.
As of July 2026, the largest data center projects in progress in MO are in:
Data centers are typically built on a single story (Barroso et al. 2019). Multi-story construction makes heat and water distribution more difficult, requires managing heavy equipment on each floor, and reduces rooftop space for cooling equipment (Johnson 2023).
However, some companies have built multi-story data centers that can fit into dense urban areas. An eleven-story data center in Singapore contains a total of 1.8 million ft2, and a 30-story data center in Hong Kong contains 350,000 ft2 (Huang 2018, Varidata n.d.).
To build data centers, developers prefer land that is easy to clear or already developed (Arzumanyan et al. 2025). Areas with lower risk of natural disasters are ideal to prevent service interruption. Data centers also require access to electricity, internet, and skilled labor to operate.
Data centers are mostly located in or near urban areas due to more reliable electricity (grid) and internet access (Fang & Greenstein 2025). They also are closer to their end users, which lowers the latency period, or users’ wait time for a response. Metropolitan areas with more than 1 million people contain over 80% of data centers in the Great Lakes region (Ferreira et al. 2026, Figure 1). That share is decreasing, as smaller metros and rural areas make up more than a quarter of data centers currently in development.
AI operations have different location requirements. Data centers run AI models using two different operations: training, which internally develops the model, and inference, which responds to user requests (Torell & Torres Arango 2026). Training requires larger amounts of computation and power than inference but does not connect to users. Because of this, large data centers in remote locations with less expensive land are used to run AI training (Iron Mountain 2026a). As the number of users grows, inference will require more power and data centers than training (Iron Mountain 2026b, Ord 2025). Inference’s short user response time requirements will need to operate in urban and suburban data centers.
Figure 1. Distribution of operating data centers in the Great Lakes region by location compared to data centers planned and under construction. Most data centers are in large metropolitan areas (dark blue) to provide short response times for users. However, smaller cities (gold) and rural areas (light blue) comprise a larger share of data centers currently in development. Data centers in remote locations can be used for AI model training. Figure adapted from Ferreira et al. 2026.
References
Arzumanyan M, Rodriguez Calzado E, Lin N, Bahadur V, Das J, et al. (2025) Geospatial suitability analysis for data center placement: A case study in Texas, USA. Sustainable Cities and Society. 131(1): 106687. https://doi.org/10.1016/j.scs.2025.106687
Barroso LA, Hölzle U, Ranganathan P (2019) The Datacenter as a Computer: Designing Warehouse-Scale Machines. 3rd Edition. https://link.springer.com/book/10.1007/978-3-031-01761-2
City of Independence (n.d.) Data Center FAQs. https://www.independencemo.gov/data-center-faqs
Fang TP, Greenstein S (2025) Where the Cloud Rests: The Economic Geography of Data Centers. Strategy Science. 10(4): 281-444. https://doi.org/10.1287/stsc.2024.0225
Ferreira J, Shobe W, Rephann T, Scheffel M (2026) Economic, Fiscal, and Energy-related Impacts of Data Centers in the Great Lakes Region. Weldon Cooper Center for Public Service, University of Virginia. https://www.coopercenter.org/research/GLDC
Huang E (2018) Facebook’s US$1b data centre in Singapore to open in 2022. Singapore Economic Development Board. https://www.edb.gov.sg/en/business-insights/insights/facebook-s-us-1b-data-centre-in-singapore-to-open-in-2022.html
Iron Mountain (2026a) Building at scale: Where we deliver accelerated infrastructure. https://resources.ironmountain.com/whitepapers/d/data-centers-building-at-scale-where-we-deliver-infrastructure
Iron Mountain (2026b) Will AI Eat the Cloud? https://resources.ironmountain.com/whitepapers/d/data-centers-will-ai-eat-the-cloud
Johnson K (2023) Going Up: The High-Rise Data Center. Corgan. https://www.corgan.com/news-insights/2023/going-up-the-high-rise-data-center
Lawson AJ, Offutt MC, Ortiz NR, Ling Z (2026) Data centers and their energy consumption: frequently asked questions. Congressional Research Service. R48646. https://www.congress.gov/crs-product/R48646
Montgomery County, MO (n.d.(a)) Montgomery County, MO Data Center – Project Spade (Google). http://mcmo.us/google/
Montgomery County, MO (n.d.(b)) Montgomery County, MO Data Center – Project Green (Amazon Web Services). http://mcmo.us/amazon/
Ord T (2025) Inference Scaling and AI Governance. GovAI. https://cdn.governance.ai/Inference_Scaling_and_AI_Governance_Technical_Report.pdf
Powell P, Smalley I (n.d.) What is a hyperscale data center? IBM. https://www.ibm.com/think/topics/hyperscale-data-center
Shehabi A, Smith SJ, Hubbard A, Newkirk A, Lei Nuoa, et al. (2024) 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory. LBNL-200163. https://doi.org/10.71468/P1WC7Q
Torell T, Torres Arango MA (2026) How 6 AI Attributes Change Data Center Design. Schneider Electric. https://www.se.com/us/en/download/document/SPD_WP110_EN/
Varidata (n.d.) MEGA-I Data Center in Hong Kong, China. https://www.varidata.com/data-center/hk/mega-i/
