Data center

Data center is covered here in 6 connected parts: Data center, Colossus (data center), AI data center, Hyperion Data Center, Google data centers, Utah Data Center.

Data center

A data center is a facility for storing, managing, and disseminating digital information, including computer systems and associated components, IT infrastructure, equipment to maintain the proper conditions for the operation of electronic equipment, and training artificial intelligence. According to IBM, data centers date back to the 1940s, with the United States military's Electrical Numerical Integrator and Computer (ENIAC) as an early example. Data centers have different models depending on their workloads.^[1]

Types

Data centers are usually classified according to their ownership model, scale, and intended use.^[1]

Enterprise

Enterprise data centers are owned and managed by a single organization to meet its internal information technology (IT) needs rather than to host data for other companies commercially. They used to be essential infrastructures; as in 2014, enterprise data centers accounted for over 60% of the U.S. server energy consumption, but this share fell sharply to about 10% by 2023 as workloads migrated to hyperscale and colocation facilities.^[1]

Colocation

Colocation centers provide shared physical infrastructure, such as power, cooling, space, etc., to multiple tenants who lease space rather than building their own facilities. Colocation data centers reduce expenditure for organizations while enabling operators to achieve economies of scale in power procurement and cooling efficiency. Some of the largest data centers in the world are colocation centers, and this category accounted for about 22% of worldwide data center capacity as of 2024, making them the second-largest category by capacity after hyperscale facilities.^[1]

Hyperscale

Hyperscale data centers are large-scale facilities, typically exceeding 100 megawatts of power capacity, designed to support massive, scalable computing workloads for cloud services, AI training, and large-scale data processing. By the end of 2024, there were 1,136 operational hyperscale data centers globally, a figure that doubled over the previous five years, with the U.S. accounting for approximately 54% of total hyperscale capacity. In particular, the three largest operators, Amazon Web Services, Microsoft Azure, and Google Cloud, collectively account for approximately 59% of all hyperscale data center capacity globally.^[1]

Edge

Edge data centers are smaller facilities, typically ranging from 1 to 10 megawatts. They are positioned closer to end users or data sources to reduce network latency and support real-time applications, such as autonomous vehicles, industrial automation, and content delivery. Unlike hyperscale data centers, edge data centers are intentionally distributed across many locations rather than consolidated, with deployments occurring in urban areas, cell tower sites, as well as industrial locations.^[1]

History

Data centers have their roots in the huge computer rooms of the 1940s, typified by the Electronic Numerical Integrator and Computer (ENIAC), one of the earliest examples of what has become known as a data center. Early computer systems were complex to operate and maintain, and required a special environment in which to operate. Many cables were necessary to connect all of the components, and methods to accommodate and organize these were devised, such as standard racks to mount equipment, raised floors, and cable trays (installed overhead or under the elevated floor).^[1]

In the United States

The U.S. hosted 5,381 data centers in March 2024, the highest number of any country worldwide, accounting for 40% of the global market in 2023. As of 2025, roughly one third of America's data centers were in three states: Virginia (643), Texas (395), and California (319).^[1]

Design and architecture

Data centers house critical computing resources in a controlled environment and must generally operate with very high availability. Key design elements include providing power for the equipment, temperature and humidity control, cabling, fire safety, and security. Information security is also a concern.^[1]

Obsolescence and modernization

Industry research company International Data Corporation (IDC) put the average age of a data center at nine years As of August 2011. As of December 2024, the mean age of data center facilities had risen to 11.3 years, according to research analyst Jacqueline Davis of the Uptime Institute. Gartner, another research company, argued in 2010 that data centers older than seven years are obsolete.^[1]

Related: Colossus (data center) · AI data center · Hyperion Data Center · Data mesh

External references: ^[1][2][3]

Colossus (data center)

Colossus is the name used for a cluster of data centers developed by xAI (later SpaceX Super Intelligence) in the Memphis, Tennessee–Southaven, Mississippi area. The original facility, Colossus 1, began construction in 2024 at a former Electrolux plant in South Memphis and became operational in July–September 2024. Its primary purpose was to train the company's chatbot Grok A second major facility, Colossus 2 (also referred to as Macrohard), was developed beginning in 2025 in the Whitehaven area of Memphis, with associated power generation across the state line in Southaven, Mississippi.^[4]

Colossus 1

Colossus 1 was built in a former Electrolux factory in South Memphis. Construction of the initial cluster of approximately 100,000 Nvidia H100 GPUs was completed in 122 days; the system went live in mid-to-late 2024. Capacity was later expanded; Epoch AI estimates the site's chips now include about 150,000 H100s, 50,000 H200s and 30,000 B200s.^[4]

Colossus 2 and cluster expansion

In February 2025 an xAI affiliate purchased a large warehouse and adjacent parcels in Whitehaven, Memphis, for roughly $80 million. This became the core of Colossus 2 (branded Macrohard on the rooftop). The first major GPU cluster (~110,000 Nvidia GB200-class processors, ~210 MW) came online around January 2026 after a reported build of about 91 days; a subsequent cluster of GB300-class processors followed.^[4]

Environmental impact and community response

Rapid deployment relied on temporary gas turbines while grid and permanent generation were incomplete. In the Memphis area this raised concerns about air quality (NOx, formaldehyde, smog), water demand (millions of gallons per day), and strain on the local grid. Shelby County initially treated many units as portable/temporary and therefore outside certain permitting rules; later federal guidance and state actions tightened requirements.^[4]

Current status (as of mid-2026)

Colossus 1: Largely leased to Anthropic; older H100/H200-heavy configuration.^[4]

Related: Data center · AI data center · Hyperion Data Center · Data mesh

External references: ^[4][5][6]

AI data center

An AI data center is a specialized data center facility designed for the computationally intensive tasks of training and running artificial intelligence (AI) and machine learning models. Unlike general-purpose data centers, they are often optimized for the parallel processing demands of AI workloads, typically using hardware such as AI accelerators (GPUs, TPUs) and high-speed interconnects. The global push to construct these specialized facilities accelerated dramatically during the AI boom of the 2020s.^[7]

Operators

As of August 2025, The Information tracked 18 planned or existing AI data centers in the United States, operated by Amazon Web Services, CoreWeave, Crusoe, Meta, Microsoft/OpenAI, Oracle, Tesla, and xAI. Other AI data center operators include Digital Realty and Alibaba. Data centers are also being built in China, India, Europe, Saudi Arabia, and Canada.^[7]

Finances

Between January and August 2024, Microsoft, Meta, Google and Amazon collectively spent $125 billion on AI data centers. Citigroup forecasted that $2.8 trillion would be spent on AI data centers by 2030, while McKinsey and Company estimated that almost $7 trillion would be spent globally by that time. According to S&P Global, $61 billion has been spent on the data center market as a whole in 2025, while debt issuance for data centers was $182 billion during the same year.^[7]

Environmental footprint

Average AI data centers have an electricity footprint equivalent to 100,000 households, and use billions of gallons of water for cooling their hardware. In 2025, the International Energy Agency estimated that the larger AI data centers currently under construction could consume as much electricity as 2 million households. A 2024 report from the United States Department of Energy stated that data centers overall used 17 billion gallons of water per year in the United States, primarily due to "rapid proliferation of AI servers", and that this usage was forecasted to grow to nearly 80 billion gallons by 2028.^[7]

Global memory supply shortage

The construction of AI data centers during the AI boom starting in 2025 has exhausted the global supply of the computer memory. To meet demand from AI servers, the three major DRAM manufacturers, Samsung, SK Hynix, and Micron, prioritized production of High Bandwidth Memory (HBM) and high-end server DRAM, contributing to a global memory supply shortage and intensifying competition for advanced memory chips. Because HBM consumes far more fabrication capacity than conventional memory—producing a single bit of HBM requires roughly three times the wafer capacity of DDR5.^[7]

In the United States

In the United States, both the Biden administration and second Trump administration supported the construction of AI data centers. In January 2025, then-president Joe Biden signed an executive order for federal government agencies to support AI data centers on federal sites built by private companies, study their effect on energy prices, and encourage their use of renewable energy. In April 2025, the United States Department of Energy suggested 16 possible sites, including Los Alamos National Laboratory, Sandia National Laboratories and Oak Ridge National Laboratory.^[7]

Concerns and opposition

Some analysts have expressed concerns about overbuilding of AI data centers, warning that their infrastructure risks obsolescence due to changes in demand and technology. Estimates for the lifespan of the GPUs used to power data centers range from one to eight years. By contrast, CPUs in more traditional data centers have a shelf life of about 5–7 years.^[7]

In space

Several tech companies, including Starcloud, Google, Nvidia, Blue Origin and SpaceX, have announced projects for or otherwise expressed interest in building data centers in outer space. Google announced Project Suncatcher, which plans to test whether its Tensor Processing Unit (TPU) chips (which are specifically designed for machine learning) can function in the environment of space. The company aims to deploy satellites with these chips in a data center by 2027.^[7]

Related: Data center · Colossus (data center) · Hyperion Data Center · Data mesh

External references: ^[7][8][9]

Hyperion Data Center

The Richland Parish Data Center, nicknamed "Hyperion", is a planned artificial intelligence data center by Meta Platforms under-construction along Louisiana Highway 183 in Richland Parish, Louisiana, just outside of Holly Ridge. It is one of a number of "titan clusters" being built in preparation for the emergence of AI superintelligence. Modern technological researchers disagree as to whether or not superintelligence will ever exist, though Meta CEO Mark Zuckerberg has expressed belief that its creation is inevitable.^[10]

History

Meta was considering potential locations for their flagship data center in early 2024. Before being announced later in December, the plan was completely secret; meetings held between involved organizations and even government officials could only refer to it by the codename "Project Sucre" to protect it from potential corporate espionage. The data center was first announced on December 4, 2024, though its full scale was yet to be revealed.^[10]

Operation

The facility is expected to consume up to 5 gigawatts (GW) of computational power, a figure Meta confirmed in July 2026 with roughly 2 GW expected online by 2030 and the full campus completed around 2032. As part of their deal made with Meta, Entergy plans to be able to produce at least 3.8 GW of electricity for the operation.^[10]

Response to the project

Louisiana Governor Jeff Landry thanked Meta for their decision to build Hyperion in Louisiana, stating that it would "create opportunities for Louisiana workers to fill high-paying jobs of the future." and calling it "A New Chapter" for the state. The Louisiana Economic Development (LED) state agency further praised the project, citing Meta's estimate that it would create 1,500 jobs. Additionally, Richland Parish Supervisor Joey Evans stated that he was excited about the project.^[10]

Related: Data center · Colossus (data center) · AI data center · Data mesh

External references: ^[10][11][12]

Google data centers

Google uses large data center facilities to provide their services, which combine large drives, computer nodes organized in aisles of racks, internal and external networking, environmental controls (mainly cooling and humidification control), and operations software (especially as concerns load balancing and fault tolerance). There is no official data on how many servers are in Google data centers, but Gartner estimated in a July 2016 report that Google at the time had 2.5 million servers. This number is changing as the company expands capacity and refreshes its hardware.^[13]

Locations

Google's infrastructure comprises both company-owned hyperscale data center campuses (which support consumer products such as Search, YouTube, and Gemini) and Google Cloud Platform (GCP) regions (which include both Google-owned campuses and leased colocation facilities).^[13]

Original hardware

The original hardware (circa 1998) that was used by Google when it was located at Stanford University included:^[13]

Production hardware

As of 2014, Google has used a heavily customized version of Debian Linux. They migrated from a Red Hat-based system incrementally in 2013.^[13]

Network topology

Details of the Google worldwide private networks are not publicly available, but Google publications make references to the "Atlas Top 10" report that ranks Google as the third largest ISP behind Level 3.^[13]

Data center network topology

To support fault tolerance, increase the scale of data centers and accommodate low-radix switches, Google has adopted various modified Clos topologies in the past. Known internally as Jupiter, this centralized software-defined networking (SDN) fabric initially relied on a hierarchical electrical packet-switched spine to deliver over 1 Pbit/s of total bisection bandwidth across tens of thousands of machines. In 2022, Google disclosed that it had replaced the electrical spine layer in its Jupiter fabric with MEMS-based optical circuit switches (OCS) that dynamically steer light paths between aggregation blocks without optical-to-electrical-to-optical conversion.^[13]

Project 02

One of the largest Google data centers is located in the town of The Dalles, Oregon, on the Columbia River, approximately 80 miles (129 km) from Portland. Codenamed "Project 02", the complex was built in 2006 and is approximately the size of two American football fields, with cooling towers four stories high. The site was chosen to take advantage of inexpensive hydroelectric power, and to tap into the region's large surplus of fiber optic cable, a remnant of the dot-com boom.^[13]

Summa papermill

In February 2009, Stora Enso announced that they had sold the Summa paper mill in Hamina, Finland to Google for 40 million Euros. Google invested 200 million euros on the site to build a data center and announced additional 150 million euro investment in 2012. Google chose this location due to the availability and proximity of renewable energy sources.^[13]

Floating data centers

In 2013, the press revealed the existence of Google's floating data centers along the coasts of the states of California (Treasure Island's Building 3) and Maine. The development project was maintained under tight secrecy. The data centers are 250 feet long, 72 feet wide, 16 feet deep.^[13]

Software

Most of the software stack that Google uses on their servers was developed in-house. According to a well-known former Google employee in 2006, C++, Java, Python and (more recently) Go are favored over other programming languages. For example, the back end of Gmail is written in Java and the back end of Google Search is written in C++.^[13]

Related: Data center · Colossus (data center) · AI data center · Data mesh

External references: ^[13][14][15]

Utah Data Center

The Utah Data Center (UDC), also known as the Intelligence Community Comprehensive National Cybersecurity Initiative Data Center, is a data storage facility for the United States Intelligence Community that is designed to store data estimated to be on the order of exabytes or larger. Its purpose is to support the Comprehensive National Cybersecurity Initiative (CNCI), though its precise mission is classified. The National Security Agency (NSA) leads operations at the facility as the executive agent for the Director of National Intelligence.^[16]

Purpose

Critics believe that the data center has the capability to process "all forms of communication, including the complete contents of private emails, cell phone calls, and Internet searches, as well as all types of personal data trails—parking receipts, travel itineraries, bookstore purchases, and other digital 'pocket litter'." In response to claims that the data center would be used to illegally monitor email of U.S. citizens, in April 2013 an NSA spokesperson said, "Many unfounded allegations have been made about the planned activities of the Utah Data Center, ... one of the biggest misconceptions about NSA is that we are unlawfully listening in on, or reading emails of, U.S. citizens.^[16]

Structure

The structure provides 1 to 1.5 million sq ft (93,000 to 139,000 m2), with 100,000 sq ft (9,300 m2) of data center space and more than 900,000 sq ft (84,000 m2) of technical support and administrative space.^[16]

Related: Data center · Colossus (data center) · AI data center · Data mesh

External references: ^[16][17][18]