Power Hungry Data Centers
How can we inform non-expert AI users about data centers and their growth (both physically and in required rate of energy consumption) since the AI boom?
Looking on a global scale, US has the highest number of data centers, well above 4k. Within the states, Virginia has the highest number of data centers, and within Virginia, Loudon county has the highest number and concentration of data centers - more than many countries. So for this investigation, my partner and I focused on Data Center Alley in Loudon county as a case study for what happens when many data centers are built in a small area. By examining the physical impacts these data centers are having here, we can anticipate their future effects on more parts of the world as they increase in size and in number.
We began by compiling data centers locations, and calculated their water and energy consumptions per year. From this data, we were able to pull out some pretty interesting comparisons, which show the exponential growth of the power draw of Loudoun's ever growing data centers.
​
To best represent this change over time, we chose to have three main factors forming the shape of each data center, as shown below. Energy consumption is represented as Megawatt per hour per day, Land coverage is measured in acres and water consumption is measured in millions of gallons.
From our visualisation we want the users to be able to understand the speed of Data center growth is relevant to the growth and demand of AI. To highlight the battle between human and data consumption and for users to realize that even though data centers are getting more efficient, it's not at the speed we need. ​
​
One of the main hurdles we encountered during the research process was a lack of available data. Most tech companies are very non-transparent about the size, consumption, and impact of their data centers, contributing to the lack of public awareness of data center impacts. This is why we had to rely on educated estimates and calculations for many of our data points.
​​​
We want to convey that the use of AI is not the click of a button, there is a large physical infrastructure behind it. While it is out of sight to the average user, it is rapidly shaping our physical world.