The surge in data centers, driven by AI and cloud computing growth, has altered the landscape of commercial real estate (CRE) site selection. While factors like transport links and workforce availability were once key, the need for steady and plentiful power now takes center stage. As data centers require non-stop utility access to handle constant data processing, power availability has become the main factor in choosing ideal locations.
Three SIOR experts with deep data center industry knowledge recently shared their views on this changing scene with the SIOR Pulse blog. They portray data centers as a complex yet profitable asset class, one that needs careful balancing of many interlinked factors to work well. From power supply and infrastructure strength to rule compliance and sustainability, each element plays a vital role in these facilities' success or failure.
They also stress that insatiable demand for power and non-stop data processing abilities renders old site selection methods obsolete, calling for new ideas to meet rising energy needs. And as data centers grow more crucial for AI, cloud computing, and digital economies, grasping and navigating these complexities is key for industry players aiming to tap into this booming market.
"For data centers, you need constant power, enough land, and just as important, the infrastructure to support it all," says Ryan Moen, SIOR, principal at Versa Real Estate Services in Chicago.
"The power factor is the deal-maker or breaker for data centers, which really highlights how important utility infrastructure is for site selection." ~ Jim Kerrigan, SIOR
Power infrastructure and its patchwork availability on a regional basis determine more than just where data centers can be built. As Jim Kerrigan, SIOR, of North American Data Centers in Chicago, points out: "The power factor is the deal-maker or breaker for data centers, which really highlights how important utility infrastructure is for site selection. There's no point in even suggesting a site to a data center developer unless these criteria are already met, or the area's power infrastructure is being actively upgraded."
Three main factors shape data center site choices: power availability, cooling options, and closeness to end users. Reliable and affordable power is crucial given how much energy data centers use. Kerrigan explains that the industry has shifted "from buildings to land" because the focus now is on which land plots can handle the necessary power.
With the vast expanse of land in the continental U.S., most areas aren't suitable for future data centers. Kim Ford, SIOR, CEO of Rise Pittsburgh, points out: "Most power grids across the country can't support new data centers. They already face major latency issues." Latency is the delay between a change in power demand or cooling and the system's response. Ford emphasizes that cutting latency is key to maintaining optimal power efficiency, cooling, and resource allocation, which helps prevent overheating, cut energy waste, and boost the data center's overall performance and reliability.
She says that finding a reliable energy source has become so important that tech companies are exploring nuclear power. In Pennsylvania, home to the second-highest number of nuclear plants in the U.S., giants like Amazon and Microsoft are even considering reviving Three Mile Island, a site known for a historic nuclear accident.
Nuclear power, being zero-carbon, aligns with their ambitious carbon-neutrality goals. It also offers a stable, long-term energy source that doesn't rely on weather or fossil fuel supply chains.
AI is driving up demand for data centers as its uses grow, requiring more robust and extensive infrastructure. At the same time, AI is changing how these facilities run, making them smarter and more energy-efficient. This shows how AI not only increases the need for more data centers but also helps them run more sustainably and efficiently.
Ford mentioned AI-driven systems like Powerwall, which uses AI to predict and store energy for peak load times, ensuring smooth operation even when the power grid is stressed. "You have to make these intelligent, like a Powerwall," she said, highlighting AI's future role in energy efficiency.
However, finding effective and sustainable solutions remains challenging. Moen notes: "Water-based cooling works great, but some towns have rejected data center plans due to the water needed for cooling." This has led to innovations like air-based systems, though they're still new and not widely used.
Ongoing supply chain issues also pose big challenges for data center developers. Kerrigan highlights long wait times for key components, like circuit breakers and transformers, now delaying projects by years. "Three years ago, the wait time for these items was three years, and it's still three to five years today," he notes, adding that utilities can't order these parts until they get customer requests, which makes delays even worse.
Beyond technical issues, data center development also faces legal and money hurdles. Local rules can speed up or slow down projects. Moen says that some states are attracting new data centers with perks and by teaming up with power companies to ensure energy supply. "It's a team effort," he says, "with officials and even investors working together to make these big projects happen."
While data centers have a bright future, it's getting more complex. Builders must tackle power needs, green goals, and rule challenges. As AI drives up demand, the industry needs to quickly upgrade its systems and build stronger ties with local governments and communities to keep up.
ARTICLE CONTRIBUTORS
Kim Ford, SIOR, is CEO of Rise PIttsburgh an Exclusive Tenant Consulting Firm based in Pittsburgh, PA.
