Data Center
Overview
A data center is a facility that integrates numerous servers, storage devices, and network equipment in one place to stably operate and provide computing resources and data. Equipped with power, cooling, security, and network infrastructure, it forms the foundation of the modern digital economy, including web services, cloud, and artificial intelligence training. Recently, with the spread of AI, the power demand and investment scale of ultra-large data centers have grown rapidly, emerging as a key agenda item in industry and policy.
Key Details
Definition and Components
A data center is broadly divided into IT equipment (servers, storage, switches, routers) and the physical infrastructure that supports it. The physical infrastructure includes power supply (UPS, generators, substation facilities), cooling (air conditioning, liquid cooling), racks and cabling, physical security, and fire response equipment. The combination of these elements determines availability (uptime), scalability, and security.
Classification
- Enterprise Data Center (On-Premises): Directly owned and operated by the enterprise.
- Colocation: The facility is leased while the customer operates the equipment.
- Hyperscale Data Center: Ultra-large facilities operated by AWS, Microsoft, Google, and others.
- Edge Data Center: Small-scale distributed facilities designed to reduce latency near users.
Design and Operational Principles
Availability and Redundancy
Power, cooling, and networks are duplicated, and reliability is classified using Tier ratings. To ensure uninterrupted operation, spare capacity such as N+1 and 2N is secured.
Power and Cooling Efficiency
PUE (Power Usage Effectiveness) is the representative metric, and the closer to 1, the more efficient. The latest facilities target a PUE of 1.1–1.2, and direct liquid cooling (DLC) and immersion cooling are spreading to accommodate high-density AI racks.
Network and Latency
Large-scale traffic is handled through spine-leaf architecture, 400G/800G Ethernet, and optical interconnection. Physical distance from users directly affects latency.
Economic and Industrial Significance
A data center is a complex industry intertwining real estate, power, semiconductors, and cloud. While it creates large-scale investment and employment, it also brings issues such as burden on the power grid, water usage, and land and tax issues. Competition among countries to attract them is fierce, and the ability to secure power has emerged as a key variable for location.
Latest Trends
In 2024–2025, the data center industry is being led by AI. As demand for training and inference of generative AI and large language models (LLMs) surges, high-density racks centered on GPU clusters (over 100kW per rack) have emerged, and in response, the adoption of liquid cooling is rapidly increasing. With the surge in power demand, interconnection queues have become a bottleneck in the United States, Europe, and Asia, and power purchase agreements (PPAs) for nuclear and renewable energy and direct power generation investments are expanding.
Hyperscale operators are announcing new campuses worth tens of trillions of won, and discussions on linkage with small modular reactors (SMRs) are also active. At the same time, regulations and disclosure requirements regarding power efficiency, water usage, and carbon emissions are being strengthened, and the EU is expanding energy efficiency guidelines and reporting obligations. In Korea as well, policy discussions are underway to ease concentration in the capital region, disperse to local areas, and link with renewable energy. Meanwhile, edge computing and the decentralization of AI inference, along with next-generation technologies such as optical interconnect and CXL, are expected to reshape the next-generation data center architecture.
Related Topics
- [[Cloud Computing]]
- [[Artificial Intelligence]]
- [[Server]]
- [[Semiconductor]]
- [[Cooling Technology]]
- [[Power Grid]]