Cloud Computing and Federal Data Center Consolidation

Many factors drive federal investment in cloud services. Agencies may be seeking to modernize antiquated systems, deliver a service more efficiently, or leverage the updated security of cloud-based capabilities. Consolidating federal data centers has been another cloud market driver over the last few years, with agencies closing thousands of facilities since the Data Center and Cloud Optimization Initiative (DCCOI) began in August 2016. Many of those workloads were then migrated to commercial cloud service providers. The number of closing facilities began to dwindle in FY 2020, however, and in September 2022 DCCOI requirements came to an end.

The federal government nevertheless continues to own thousands of computing facilities that it costs billions of dollars to maintain and operate. Comments in agency budget requests and from federal officials also still appear regularly concerning the need to further reduce the number of data centers. Just in June 2023, for example, Air Force Deputy Chief Information Officer Winston Beauchamp told Washington Technology, “We still have a thousand data centers. There is a federal mandate to reduce the number of data centers and consolidate them, but we're trying to get as much of this moved to the cloud as we can and reduce the management costs associated with running those centers.”

Given that federal officials are still focused on reducing data centers, and that they are seeking to leverage cloud computing as the primary way of doing it, the question is how this will shape the federal cloud market in the years to come.

Remaining Data Centers

According to data reported to the General Service Administration’s IT Dashboard, agencies are planning to close only 26 data centers over the next two fiscal years (see chart below). This number is not going to cut it. If the Air Force alone owns 1,000 data centers it would take 100 years to close them all at a rate similar to what is being reported for the entire federal government.

A more realistic estimate of the remaining data centers to be closed can be found in the overall number of them the federal government still operates. This is shown in the chart below.

The number of remaining data centers is 3,374, with the vast majority of those owned by the Department of Defense. These numbers represent data centers that have not been designated as Key Mission Facilities or KMFs. Key Mission Facilities fulfull a purpose critical to the mission of a given agency, and the Office of Management and Budget allows them to be excluded from previously mandated closure requirements.

Data Center Closures and Cloud Investment

Although there is no way to directly translate the closure of data centers into cloud investment, we can take a look at the correlation between the two trends and come up with some interesting observations.

• First, civilian agencies closed 400 data centers from FY 2018 to 2022. During that same period, identifiable civilian sector cloud spending rose from $3.7B to $6.5B. In short, closing 400 data centers took place at the same time as civilian agencies spent $2.8B on cloud.
• Second, the DOD closed 311 data centers from FY 2018 to 2022 and it’s identifiable cloud spending rose from $1.6B to $5.4B. In other words, the DOD’s cloud spending rose $3.8B as it closed 311 data centers.

Correlation is not causation, but it looks like closing 300-400 data centers correlates with a rise of between $2.8B and $3.8B in the civilian and defense sectors of the cloud market. What then can we expect in the years ahead?

Let’s look at the DOD first. A back of the envelope calculation dividing the DOD’s $3.8B in cloud spending by 311 closed data centers results in $12.21M in cloud spending per data center closed. Applying this number to the remaining number of DOD data centers, with the understanding that they won’t close all of them, gives us the potential that the DOD will spend $35B on cloud services in place of its current data center services. The period of years over which this spending will take place cannot be calculated because the DOD does not appear to have or to publish a data center closure plan.

Now for the civilian sector. Doing the same calculation (i.e., $2.8B divided by 400 closed data centers) yields $7.0M in cloud spending per data center. If civilian agencies close all 467 of their remaining data centers at $7.0M in cloud spending per data center it totals $3.26B in potential cloud spending.

The conclusion to draw from this back of the envelope calculation is that potential ceiling for cloud spending is more than 10x higher at the DOD than it is across the civilian sector. This should answer any remaining questions there might be about why the fight over the Joint Enterprise Defense Infrastructure (JEDI) single award was so protracted and bitter.

This exercise is merely a thought experiment, but it appears to suggest that cloud spending should rise much more strongly at the Department of Defense than in the civilian sector in the years to come.