Across a wide range of industries and applications, the world’s data needs continue to expand at an unprecedented rate. From finance to oil and gas to health care to agriculture, data has become a vital component of day-to-day operations across all industries; as data needs increase, so too does the pressure on mission critical systems to support data storage and processing. The physical embodiment of these growing needs are data centers, an often-sprawling one-story facility housing the infrastructure necessary for computing, storing, and processing massive amounts of data. Data centers now represent a vital component of global infrastructure. By 2025, these buildings are projected to draw 20% of the world’s power.
As market demand for data continues to increase, demand will need to be matched by a dramatic increase in construction of data center facilities. However, building a new data center involves a host of challenges, from location to communication and power to land availability. With each new data center, the physical locations suitable for construction become scarcer. To make better use of both financial and real estate assets, the mission critical industry must rethink the legacy one-story box and instead consider a more vertical solution.
Taking as an example “data center alley,” the global data center hub of Loudoun County, Virginia, representatives from M.C. Dean, Avison Young, and CallisonRTKL analyzed the mission critical, architectural, real estate planning, and development costs and challenges facing data center construction. The research modeled a variety of building heights and programs, examining the results with regards to footprint, cost, and megawatts per acre. In addition to creating more capacity with a smaller footprint, multi-story data centers offer better value in terms of cost per square foot and megawatts per acre.
Data center construction is not as simple as finding a large enough piece of land and starting to build. The physical and technical requirements for a data center include access to water, affordable power, communications connectivity, environmental resiliency, and several acres of land. And even when these needs are met, complicated zoning requirements, local laws, sound ordinances, proximity to residential and water management could still make data center construction difficult.
The elaborate process of designing and developing data centers can be defined by four major challenges, each of which includes a series of requirements and moving targets that further complicate the situation.
While land is considered an abundant resource, sites embodying all requirements for data centers are becoming a scarce and valuable commodity in major markets. For legacy, one-story data centers, land has traditionally been evaluated by the following metrics:
A minimum of 20 acres of land
Sufficient water resources or geographic location for free cooling
Proximity to communications providers or a major peering point
Robust, reliable, and affordable power
Resilience against natural disasters
Data center construction may also be complicated by zoning laws and entitlement processes. Any time lost to paperwork and bureaucracy prevents an industry’s data needs from being met; therefore, an accommodating municipality with straightforward procedures is a great value.
A single data cabinet can weigh as much as a mid-sized sedan. To accommodate the hundreds of cabinets housed in a data center, the building must be robust and well designed to provide the necessary structural integrity. Potential seismic activity should be factored into the design, as a data center must be able to withstand any potential natural disasters in order to provide continuous availability.
Because of the physical demands of data center construction, it is important to bring together an experienced team of architects, contractors, and engineers to ensure long-term success. Local knowledge also is key to understanding the environmental demands that could be placed on the structure, such as snow loading or heavy winds.
When choosing the optimal design for a data center, the initial financial outlay must be weighed against the expected life of the facility and cost of long-term operations. The archetypal single-story data center has lower upfront costs when compared with a multi-story building.
Additionally, real estate prices must be factored into the decision, as the number of suitable locations for data center development continues to plummet. Even for developers who may be able to justify a one-story data center in the short term, may find it makes more financial sense to plan for inevitable expansion needs by building a two- or three-story data center from the start, or at least master planning the option to include additional levels in the future.
Given the volume of equipment involved in building and furnishing a data center, there are significant logistics that must be taken into account during the design and construction processes. For example, IT equipment seldom exits a tailgate and is delivered directly to a server environment. Typically, the equipment is processed to a secure storage location, after which there is a transfer of possession, possibly a burn-in or test environment, then migration to production. Experienced architects and engineers can compensate for these processes with flexible design principles and an innovative mindset.
Another important consideration is schedule. Due to the rapid scaling of major tech companies, the primary challenge is finding new ways to accelerate product time to market by reducing construction schedules and overall project durations. In addition to meeting considerable structural demands, engineers and contractors must be able to work fast enough to deliver an immediate solution to a company’s data needs.
Data center alley
Northern Virginia — Loudoun County in particular — is known throughout the world as “Data Center Alley” due to its tremendous density of data centers. As a result of low power costs, robust fiber connectivity, and an accommodating entitlement process, more than 650 facilities related to data storage and processing have been established in Northern Virginia. These facilities support and connect thousands of companies and billions of endusers. These businesses are rewarded handsomely by the local government, as Loudoun County offers significant tax incentives to data centers that hire more than 50 employees and compensate them at a rate 1.5 times higher than the local average. Even more incredible than the number of businesses is the extent of their reach — 70% of the world’s internet traffic passes through the sprawling network of data centers in Loudoun County. The county itself leans into its status as the global data center hub. Their fast-track commercial incentive program pairs dedicated employees with new data centers to help them move quickly through the entitlement process.
But as land has been developed to meet the rapidly growing need for data center construction, the cost of land has skyrocketed. In just two and a half years since May 2016, the average cost per acre in Northern Virginia has increased by roughly $1.2 million, from just over $400,000 to more than $1.6 million.
Because of its reputation in mission critical industries as a global data center hub, and because of the area’s significant sample size with regards to existing data centers, Loudoun County provided a useful reference point when modeling and analyzing the potential costs of multi-story data center construction.
Fixed price cost analysis
To better understand the cost and performance implications of constructing multi-story data centers, a consortium from M.C. Dean, CallisonRTKL, and Avison Young evaluated a series of design options for data centers in Loudoun County, Virginia. Each design was measured with a fixed price cost analysis, with the relevant metrics being total cost, cost per square foot, cost per megawatt, and megawatt per acre. The fixed cost of land for the exercise was set at $1.5 million.
The baseline example of a legacy one-story data center has a surface area of just over 20 acres and has an initial cost of $306 million. The one-story building delivers 2.39 megawatts per acre, with each megawatt costing $7.65 million. The cost per square foot of a one-story data center is significantly larger than in multi-story buildings at $920.
A two-story data center, built on approximately 14.5 acres, offers immediate benefits with regard to overall cost and cost per square foot. The total cost of the two-story data center is $302 million, while the cost per square foot is approximately $877. A two-story data center also offers a significant improvement in cost per megawatt: just under $7.55 million.
Offering the best balance between cost, area, and future flexibility, a three-story data center requires just 10.5 acres of land. While costing slightly more than a two-story data center at $303 million, the three-story data center offers the best result in terms of cost per square foot ($875). A three-story building offers 4.58 megawatts per acre (compared to 3.27 in a two-story building), while cost per megawatt is slightly higher at just under $7.6 million. When considering rising real estate costs in major technology centers like Northern Virginia and Silicon Valley, the financial benefit of a three-story data center is likely to increase as developable parcels become scarcer.
With each additional story added to a new data center project, the building’s megawatts per acre increases accordingly. However, a four-story data center is significantly more expensive to build than a three-story structure. Carrying a total cost of more than $314 million, a four-story data center could be considered prohibitively expensive when considering the similar performance delivered by a two- or three-story structure. The issue of cost is compounded when adding even more floors, and many municipalities require a special exception to build a structure of four or more stories.
While this analysis was conducted in the context of Northern Virginia, it is likely that the same principles and results would apply in any region offering the necessary attributes for data center construction. Because of the insatiable need for data in mission critical industries, it is unlikely that the market for suitable real estate will slow down in the near future.
The sweet spot: three stories
Mission critical industries must consider the context of dwindling land resources and rising real estate costs when planning data center construction. Bearing in mind the long-term need for increased data resources, one-story data centers should be eschewed in favor of multi-story structures that can deliver more performance in less space. According to the recent analysis, the sweet spot for data center construction is three stories. Comparatively low construction costs and an unquestionable advantage in long-term performance ensures that going vertical is the right choice for new data center builds.
Regardless of the final design, the complicated details of data center construction require careful planning and an intelligent long-term strategy. Mission critical industries must engage with experienced engineering, architecture, and systems integration firms to ensure that their data needs are met efficiently and with a durable solution.