At the end of 2017, there were nearly 400 hyperscale data centers worldwide. By the end of 2020, that number had leapt up to nearly 600. The question isn’t if the number of hyperscale data centers is growing, but where. Current data shows that around 40% of hyperscale centers are in the United States—specifically, Northern Virginia, Northern California, Phoenix, Dallas, and Chicago—have all seen hyperscale development and leasing.
Right behind the U.S., Europe and Asia Pacific have seen growth as well. According to recent data, China currently hosts 10% of the world’s hyperscale sites, followed closely by Japan, Germany, the UK, Australia, Canada, India, and Singapore.
In our conversation with Tag Greason, the Chief Hyperscale Officer at QTS, he shared that the hyperscale market requires four things: scale, speed to market, price, and location. So when it comes to the last variable, how do hyperscale developers decide where to build?
The factors that go into hyperscale site selection
Cost of power will always be a driving decision when it comes to building data centers, and for hyperscale centers, it’s an even bigger factor. The average provider designs for 150-175 watts per square foot, which comes to approximately 7-8 kilowatt per rack. Hyperscale data centers, however, can require 240-300 watts per square foot (15 kilowatt per rack) or more. Depending on the provider, planning for capacity and understanding the costs that go with certain needs is paramount.
In addition to understanding the cost of power, hyperscale centers also pay special attention to the reliability of that power. For a higher degree of reliability (in some cases 99.999%), hyperscale centers often utilize multiple diverse paths with underground service. Many hyperscale centers require feed from two separate substations—preferably from different grids and different providers. And in some cases, the use of different power sources like nuclear and hydro can be used effectively.
Of course, raising the subject of power sources inevitably leads to the topic of sustainability. Hyperscale data centers obviously utilize massive amounts of energy and are therefore under intense scrutiny for doing so in an environmentally conscious manner.
Some companies have gone as far to develop their own wind farms and solar energy plants near their hyperscale operations. And when it comes to issues like water conservation, indirect and direct evaporative cooling techniques have become a common method. All that to say, hyperscale data center developers have to weigh a lot of variables when choosing a location. But one way they’ve gone about doing that work is by following in the footsteps of others.
Current hyperscale developments attract more growth
One of the trends seen with hyperscale data center development is clustering, which allows data center users to take advantage of the infrastructure and connectivity that’s already in place. For example, many believe the Phoenix data center market has similar growth opportunities as Northern Virginia, where hyperscale development is at its highest.
Microsoft and Google’s recent investments into the Phoenix area have made it more attractive to others who are looking to build off of the growth that’s already started there. While Phoenix has the advantage of low operating expenses, we’re also seeing it have the gravitational pull of hyperscale users investing in the market one after another.
So, when looking for locations where hyperscale data centers will be in the future, look at the ones that are already starting to develop. Growth will beget more growth, creating a compounding effect as hyperscale data centers look to cluster near each other and grow together.