As environmental challenges become increasingly acute, there is growing recognition that rising electricity consumption leads to greater reliance on fossil fuels, which in turn drives up carbon emissions into the atmosphere—creating a domino effect that exacerbates environmental degradation. Modern data centers, from websites to online banking platforms, power virtually every information application, resulting in a substantial increase in their energy footprint, with power supply and cooling accounting for the largest share of energy use. According to statistics, server operations across the United States consume more than US$10 billion worth of electricity annually, with over one-third of that expenditure devoted to cooling. An efficient, energy-saving, green cloud data center is one that maximizes energy and space utilization while enabling enterprises and institutions to achieve a sustainable computing environment. High resource utilization, automation, low energy consumption, and automated management have thus become key priorities in the development of next-generation data centers in China.

01 Scientific Site Selection

Geographical Location: First and foremost, the geographical location of a data center is the most critical factor and should be thoroughly considered from the outset of the decision-making process. Key factors to analyze include: the probability and frequency of natural disasters at potential site locations (such as floods, hurricanes, and tornadoes); environmental risk factors (the extent to which the data center will impact its local environment); and climatic factors (whether the proposed site offers free outdoor air for cooling—a highly advantageous resource that can provide significant cost savings).

Power and Energy Supply: Electricity—or, more broadly, power—represents a critical factor, as it constitutes a major component of data center operating costs. Key considerations include: availability—while assessing the local power supply infrastructure, it is essential to evaluate whether alternative sites have access to multiple robust grid connections; cost—comparing various power-cost structures is also crucial; in other words, the cost per kilowatt-hour of power must be sufficiently low; and the availability of alternative energy sources—decision-makers should also determine whether potential sites can tap into renewable energy sources such as solar, wind, or geothermal power, which can help organizations cultivate a greener corporate image.

Communications Infrastructure: Communication connectivity is one of the most critical components of a data center. When selecting a site for a data center, it is essential to consider a wide range of factors from the perspective of communications infrastructure, including: the presence and proximity of fiber-optic backbone networks; this will help determine the precise capital expenditure required to connect the backbone network to the data center site; the type of fiber optic cable, which affects transmission speed; the type of local telecommunications service provider and the service models they support; and latency—both transmission latency and delivery latency are important considerations.

Tax Rates: Another important consideration is the tax rates imposed by the governments of the alternative locations. Factors to be considered include property taxes, corporate taxes, and sales taxes.

Construction Factors: When locating a data center at any specific site, construction costs play a critical role. Before making a data center site-selection decision, enterprises should consider the following factors: the maturity of the local construction industry, the availability of experience in building data centers, the adequacy of relevant construction technologies, the availability of local construction labor, and whether labor costs are within the enterprise’s affordability.

Transportation: The availability of various modes of transport is also a factor influencing data center site selection. Data center equipment must be delivered, and workers and related suppliers need convenient access to the data center location.

Cost of Living: The cost of living in the location of a data center should also be a key factor in enterprises’ data center site selection.

02 Flexible Scaling

Over recent years of exploration in data center construction, the industry has evolved from a purely building-based approach to adopting three primary deployment models: building-style, warehouse-style, and container-style. By comprehensively comparing the strengths and weaknesses of these three models and aligning them with specific business requirements, large-scale data centers at present increasingly emphasize a modular construction philosophy, which is manifested in the micro-modularization of server cabinets, the modularization of server rooms, and the modularization of entire data center buildings.

In cabinet construction, micro-modules are employed, characterized by modular integration, system simplification, flexible configuration, and rapid deployment. These micro-modules incorporate racks, integrated cabinets, rectifier cabinets, battery cabinets, AC/DC distribution cabinets, cabling, cable trays, anti-static flooring, cooling units, sealed hot/cold aisles, and an environmental monitoring system, thereby forming a relatively complete micro-module system that requires only external power supply and cold-aisle containment. Micro-module components can be prefabricated in the factory and assembled on-site, which simplifies the construction process, enables flexible configuration, and facilitates rapid deployment.

Based on a modular design philosophy, the data center is engineered to accommodate mixed-use scenarios meeting Tier 2, Tier 3, and Tier 4 standards, with construction requirements emphasizing low cost, high standards, and rapid deployment. The project is planned in one go and implemented in phases. For example, a micro-module can be defined as a single 10 kV power feed serving two floors totaling 4,000 m², with gas-based fire suppression zones delineated accordingly. Power and HVAC systems are modularly configured and integrated to meet the diverse requirements of different data center tiers. All utility pipelines are horizontally routed following the principle of shortest possible paths to minimize crossings, while power distribution and IT equipment are installed on the same floor and supplied from the same level. This approach reduces civil engineering costs, shortens deployment time, and delivers key benefits such as high versatility, shorter pipeline runs, phased scalability, and resource reuse.

The primary characteristics of modular data center construction are “modular architecture, compact layout, and efficient organization.” In terms of spatial configuration, the data center building and its supporting power center are treated as basic modules, enabling a flexible, phased spatial layout that balances short-term and long-term needs. The initial phase involves constructing a limited number of data center buildings and power-center components to enable rapid commissioning and operation, with the cluster size of data centers and power centers gradually expanded in line with business requirements. This approach helps mitigate investment pressure and delivers favorable economic performance.

03 Green and Energy-efficient

First, select a rational geographic location and make full use of natural resources.

As is well known, a favorable geographic location translates into a more optimized infrastructure and application environment. Factors such as the probability and frequency of natural disasters at alternative sites, the presence of environmental hazards, and local climate conditions all directly determine whether a data center enjoys a conducive external environment—for example, access to outdoor air for cooling.

For example, the three major telecom operators that have pooled their resources in Guizhou are leveraging the region’s mild winter and cool summer climate to achieve significant energy savings. Meanwhile, Alibaba’s Qiandao Lake Data Center utilizes deep-layer lake water from Qiandao Lake for cooling and, through custom-designed hardware, has achieved an annual average PUE of below 1.3—dropping as low as 1.17 at times. This translates into annual electricity savings of tens of millions of kilowatt-hours compared with conventional data centers, while reducing carbon emissions by more than 10,000 metric tons of coal equivalent. As such, it stands as one of the most energy-efficient subtropical data centers in China today.

Secondly, energy conservation and emission reduction should be prioritized by focusing on lowering the energy consumption of data centers.

The infrastructure of a typical data center is often referred to as the “air, fire, water, and electricity” of the server room—namely, air-conditioning and cooling, fire protection, humidity control and moisture prevention, and power supply. Accordingly, the primary energy-consuming equipment in a data center can be categorized into four main groups: air-conditioning systems, IT equipment, cooling equipment, and power supply, distribution, and lighting systems, along with other electricity-intensive components such as fire-protection systems.

Data indicate that cooling equipment accounts for the largest share of data center energy consumption, at 45%, followed by IT equipment, which accounts for 30%. While the energy consumption of IT equipment cannot be reduced directly or effectively, conventional air-cooled HVAC systems consume substantial amounts of electricity and contribute significantly to overall power usage. By adopting sound planning and design, along with innovative green and energy-efficient technologies, it is possible to substantially reduce the energy consumption of cooling systems, thereby achieving energy conservation and environmental protection more directly and effectively.

04 Economically Reliable

Economic efficiency and reliability require a comprehensive assessment of equipment redundancy, utilization rate, safety, maintainability, and compliance with relevant standards to strike an optimal balance between investment and benefits. The main approaches include:

(1) Optimization of the backup power supply system. By deploying high-voltage diesel generator sets, large-capacity power can be transmitted over long distances, reducing transmission losses compared with traditional low-voltage generator sets and making this solution particularly suitable for centralized power supply in large data centers. This approach also enables a better match between power demand and data center utilization, thereby improving overall facility utilization. A centralized power center building is established within the campus to house generator sets in a shared configuration, creating the conditions for their common use and resource pooling. Reusing generator sets reduces the total number of units required across the campus, lowering both capital and operational costs for data center construction and operation. With a centralized power center in place, different levels of power supply reliability can be tailored to meet varying user needs, enabling tiered power delivery. Furthermore, based on business classification, the optimal combination of generators and UPS systems can be configured to minimize UPS usage.

(2) Intelligent control. The data center environment leverages centralized monitoring, unified operations and maintenance, and integrated management capabilities to achieve high efficiency, enhanced security, green energy savings, and intelligent operations in data center maintenance and management, thereby improving reliability and reducing the intensity of O&M activities.

05 Intelligent and Controllable

The intelligent control system for data center server rooms comprises subsystems such as the operations and maintenance management platform, integrated power and environmental monitoring, specialized monitoring platform, and data center resource management.

The operations and maintenance management platform provides data center O&M personnel with functions such as daily operational maintenance, incident and work-order management, and O&M analytics, while gradually enhancing the O&M knowledge base. The integrated power and environmental monitoring system conducts real-time, dynamic monitoring of the data center’s power and environmental conditions, promptly identifies issues, and leverages long-term data statistical analysis to gain timely insights into trend developments and proactively mitigate potential problems. In addition, when issues do arise, the system enables precise fault localization and rapid resolution. The specialized monitoring platforms cover building equipment monitoring, intelligent lighting, power distribution and transformation monitoring, intelligent network monitoring, security surveillance, and fire safety monitoring. Data center resource management focuses on managing resources directly related to business operations—such as space and power—and thereby supports business operations.