Abstract

Data center (DC) development, driven by the AI boom and increased demand for cloud computing, has reached a critical juncture. While DCs are central to AI infrastructure, their environmental impacts are drawing increasing attention. Green initiatives for DC development have long focused on their prodigious power and cooling requirements – i.e. the sustainability focus was on technical issues associated with DC operation. General efficiency standards, such as LEED and ISO50001, have been applicable to sustainable DC development for some time. More recently however, sustainability issues surrounding DCs have broadened beyond their technical operation and, over $64Billion in US projects have been delayed or blocked between 2022 and 2025 due to community opposition or infrastructure limitations resulting from spillover effects. For example, the massive water requirements for cooling has been linked to water shortage for neighboring residents. The unprecedented power requirements of DCs has increased the potential for brownouts, utility bills for residents, and strain on aging power grids. Cooling systems and backup generators produce significant, persistent noise that degrades the quality of life for surrounding neighborhoods. The local economic benefits of DCs have been questioned because the number of new jobs they generate is usually quite limited. Taxpayers are increasingly questioning the generous tax abatements and sales tax exemptions given to tech giants that often yield minimal local employment, with some officials now moving to revoke or renegotiate these incentives. NIMBYism and Political Opposition to DC development is on the rise. Ashburn Virginia is considered the Data Center Capital of the world and its experience with DC development vividly captures both the benefits and evolving concerns (Carey 2025). The foregoing discussion shows how sustainable DC development, ipso facto, is a multifaceted and long-term decision-making problem. Multifaceted implies multiple interacting factors and objectives, some of which may conflict with each other. Long-term implies that nonlinearities in relationships among factors can manifest themselves, and there will be feedback effects among them. In other words, factors that could be considered exogenous for short to medium term decision making now have to be considered endogenous. In this talk, we will make the case that the system dynamics (SD) methodology (Forliano et al 2024) is particularly well suited for creating IS artifacts that provide decision support for sustainable DC development. Relevant characteristics of the methodology include the ability to capture nonlinear relationships (e.g. DC size and cooling needs) and feedback effects among interacting factors (DC tax revenue and local business growth), as well as delayed effects of selected factors (e.g. NIMBYism). Furthermore, SD models are systems of difference-equations which can be simulated, making them ideal to explore policy alternatives computationally in a decision support setting. This experimentation can help the multiple stake holders involved in DC development – i.e. technology companies, state and local government, local citizen representatives, local business groups etc., arrive at a common mental model of the tradeoffs involved and devise a sustainable solution that works for all in the long run.

Download

Share

COinS