Managing Sustainability Risk in Data Center Development

Technology firms are racing to acquire the data center capacity they need for new generative artificial intelligence (AI) models and related applications, opening opportunities for commercial real estate developers to participate in hyperscale data center development and increasing demand for colocation data centers. Large new data center campuses are going up across the U.S., and vacancy rates at existing data centers have fallen to new lows.

At the same time, the IT equipment (ITE) needed for AI training and inference, machine learning, large-scale data processing and image/video processing is more energy- and water-intensive than for traditional data center uses such as enterprise cloud computing. Most large technology firms have deferred their carbon reduction goals, but state and local authorities and utility companies have increased their scrutiny of the demands that new data center projects place on power and water resources. Local community concerns about undesirable effects from new data center construction remain a perennial challenge for developers. The NAIOP Research Foundation commissioned this report to examine best practices in sustainable data center development. The authors interviewed data center owners, operators, investors, developers and designers to evaluate how development teams are balancing three competing concerns: meeting market requirements, acquiring adequate power, and managing near-term and long-term sustainability risk—the liability that issues related to a project’s real or perceived sustainability will result in a financial loss. Key findings from this research include:

• Developers should have a plan in place to address stakeholder concerns about a proposed project’s use of power and water resources and the noise a completed data center will produce.
• Local market, regulatory and environmental conditions influence how development teams balance power and water efficiency.
• Some developers are turning to on-site generation to bypass delays from local utilities, but this strategy carries additional regulatory risks and is typically less cost-effective than sourcing power from the grid.
• New data centers are increasingly being built out with liquid-cooled ITE, which is more power efficient than air-cooled equipment and is often required to adequately cool state-of-the-art chipsets.
• Data centers face greater long-term financial risks associated with greenhouse gas emissions than other commercial properties due to their large power requirements. This risk can be mitigated by either sourcing low-carbon or carbon-free power or by developing a longer-term plan to reduce the property’s direct or indirect emissions should the market or regulatory environment require it.