This audio was created using Microsoft Azure Speech Services
There’s no question when it comes to prefabricated modular data centers, it offers real benefits in terms of speed of deployment, greater reliability, greater flexibility to scale, and improved predictability. But when the topic of cost comes up, there is often a lot of confusion and disagreement as to whether it’ll save you money or cost you a premium over traditional, stick-built sites.
White paper 218, Quantitative Analysis of a Prefabricated vs. Traditional Data Center, is our attempt to set the record straight. In this paper, we analyze the capital costs of two data centers whose capacity, cooling architecture, power architecture, level of redundancy, density, and the number of racks are the same. By keeping everything the same, we were able to isolate the cost differences of the design approach (prefab vs. stick-built). We used one of our reference designs – a 440 kW tier 3 data center, as the basis for the design attributes. Romonet data center modeling software was used to complete this analysis, and Romonet experts reviewed and validated our analysis model itself.
When we looked at all capital costs for the physical infrastructure of this particular design (including material, design, site prep, installation, commissioning), the prefabricated approach was 2% less than the stick-built. Given the margin of error in this type of cost exercise, this led us to the conclusion that the two approaches cost roughly the same.
The materials for prefab come at a capex premium over traditional systems. I don’t think this comes as a surprise given that prefab modules are shipped with the physical infrastructure preassembled, so it includes the physical housing (the containers), as well as the cost of labor for the factory integration.
Field labor and space costs, however, favor prefab designs, which is why the two approaches end up equal in the end. The amount of onsite labor (both electrical and mechanical) is reduced since integration work done in the factory results in less field work. And the space cost (cost of the land, building, site prep work) represented the biggest savings opportunity in this analysis.
Building cost was a key driver to the results. We chose a “typical” building cost in our analysis, but actual costs vary significantly from country to country, and site to site. (Compare costs for a data center with different attributes using our TradeOff Tool). In general, prefabricated designs become more cost effective when compared to a traditional data center within a more expensive building or space. Likewise, the traditional approach becomes more economical if there is a low cost space available.
The other key driver was power density per rack. As density decreases, the number of racks and rack PDUs increases, but more importantly the size of the building and/or containers also increases. Having more or larger containers would represent additional overhead which diminishes the savings, and may even swing the results in favor of traditional as densities continue to drop.
The bottom line is that a direct comparison of capital cost alone generally doesn’t provide a strong case to shift to a prefabricated design approach. It’s important to understand other business drivers in making this important decision, such as the value to deploy faster, simplicity in repeating solutions in multiple locations, or the predictability in project cost and timelines. These are more difficult items to quantify and compare, but are an example of some common reasons why actual owners and operators have opted for a prefabricated modular approach. Prefabricated data centers offer a different way to deploy infrastructure which can offer great benefits depending on the application and specific site challenges. As manufacturers continue to optimize and standardize prefabricated module designs, it is likely that the enclosure and implementation costs will continue to decrease.