如何在不新建数据中心的情况下提升现有设施密度
Increase Data Center Density Without New Construction
面对选址困难、审批周期拉长和电网受限,提升现有数据中心密度往往比新建更快更省钱。密度常用 kW/sf 和 kW per rack 衡量,例如 10 MW IT 负载分布在 10 万平方英尺白空间即为 0.1 kW/sf,而 AI/HPC pod 在 5000 平方英尺提供 5 MW 相当于 1 kW/sf。可通过替换高功耗 GPU 服务器、优化机柜、重排机柜行和回收闲置空间扩容,但需应对供电、散热、维护通道及监管审批等挑战。
正文
当前显示原文,中文全文译文尚未就绪。
Densifying existing data centers is often faster and cheaper than building new ones, but it requires careful management of power, cooling, access, and regulatory constraints.
September 14, 2026
Building new data centers is becoming increasingly difficult due to siting challenges, longer approval timelines, and grid constraints. For many operators, the faster, more economical path to capacity is to increase the density of existing facilities.
This article defines data center density, explains why densification is worth pursuing, details how to execute it safely, and highlights the operational, mechanical, and regulatory considerations you must manage along the way.
What Is Data Center Density?
Data center density describes how much compute workload a facility supports within a given amount of space. It is commonly expressed in two complementary ways:
kW per square foot (kW/sf), calculated as IT load divided by net white space. For example, 10 MW of IT load across 100,000 square feet of white space equals 0.1 kW/sf. By contrast, an AI/HPC pod delivering 5 MW in 5,000 square feet equates to 1 kW/sf; kW per rack, which indicates how much power and cooling one cabinet can reliably support. This measure is especially useful for planning distribution, cooling methods, and aisle containment at the row level
The Case for Densification in the AI Era
Densification can be faster and less capital-intensive than new construction because it uses existing real estate, interconnection, and support infrastructure. It reduces exposure to site selection and permitting risks, as well as community opposition.
Rising compute intensity further strengthens the case. Modern systems, especially GPU-accelerated systems that power AI and simulation workloads, often draw substantially more power per rack than traditional CPU-based servers. Instead of adding more low-power racks, operators can upgrade racks and power and cooling systems to concentrate workloads where space, power, and operations can support them.
Operators can tap multiple tactics to increase data center capacity:
Replace or augment lower-power servers with higher-power, higher-utilization systems. Consolidating GPU-equipped or otherwise high-throughput platforms increases IT load and performance without expanding the physical footprint, but it requires corresponding changes in rack-level power delivery and cooling; Optimize cabinets. Use taller or deeper racks to increase per-cabinet capacity, and standardize row geometry and depth to boost packing density while preserving service clearances; Reconfigure the floor. Adjust row layout and aisle spacing where safe and compliant, and align power and cooling distribution to the plan to increase racks per square foot; Expand space utilization. Many facilities underutilize available white space. Consolidating stranded capacity, reclaiming partially used areas, or converting adjacent rooms can raise effective utilization and density
On the flip side, these density expansion strategies present challenges.
Power supply. Higher density requires delivering more power to IT equipment. In regions where the grid is constrained, operators may add on-site generators or energy storage (subject to permitting and emissions limits). These solutions are feasible but capital-intensive and may lengthen project timelines; Cooling capacity. Increased IT load translates into greater heat output. If density is raised by moving racks closer together, airflow can become less efficient, intensifying cooling challenges. Solutions, ranging from enhanced air cooling and containment to liquid cooling, are available but require careful design and significant investment; Access and maneuverability. As floors become more crowded, serviceability can suffer. Insufficient clearances impede rack moves and maintenance. Facilities can mitigate this by adopting overhead handling systems, such as ceiling hoists, to move equipment safely and ensure alignment with ASHRAE TC 9.9 thermal guidance, NEC/NFPA electrical codes, and applicable local permitting requirements; Regulatory constraints. Even if a data center was approved at a certain capacity, local authorities may question plans to increase density, especially where power or water resources are limited. Early engagement with regulators and utilities, backed by clear impact assessments, helps reduce approval risk
Increasing data center density is a strategic, often more economical way to meet surging compute demand, especially for AI, without breaking ground on new facilities. While it introduces real challenges in power, cooling, serviceability, and compliance, thoughtful planning and targeted investments can overcome them. In many cases, densification offers a faster, less costly route to capacity than building anew, with the added benefit of maximizing the value of existing assets.
Christopher Tozzi is a technology analyst with subject matter expertise in cloud computing, application development, open source software, virtualization, containers and more. He also lectures at a major university in the Albany, New York, area. His book, “For Fun and Profit: A History of the Free and Open Source Software Revolution,” was published by MIT Press.
Want more Data Center Knowledge stories in your Google search results?
Editor's Choice
Aug 21, 2026
Want more Data Center Knowledge stories in your Google search results?
Join the entire data center ecosystem September 21 – 23, 2026, at the Gaylord Texan Resort & Convention Center in Dallas, TX.