Utilizing Power Modeling Solutions for Data Center Planning & Siting

The data center market is expanding at an unprecedented rate. The U.S. colocation market alone has doubled in size over the last four years. Colocation, or “colo,” facilities are third-party data centers that house the servers and computing hardware of multiple businesses. As of mid-year 2024, U.S. colocation capacity, measured in the power consumption of the maximum number of servers a data center can house, stands at 12 GW with an additional 5.3 GW under construction and a further 22 GW in the development pipeline.

The global data center market is seeing similar growth, with nearly 20% year-over-year (YOY) growth in Europe and 15% YOY growth in Latin America. The data center market in Singapore is so tight that existing data centers have just a 1% vacancy rate.  

This extraordinary market expansion shows no sign of slowing down. By 2030, the global data center market is expected to be worth more than $437 billion, growing at a 10.9% compound annual growth rate between 2023 and 2030.  

However, one significant challenge tempering the data center industry’s rapid expansion is the lack of reliable electricity to power these critical facilities, which makes site selection for new developments incredibly complex.  

In this article, we’ll look at what’s driving data center growth and explore how  modeling solutions can help data centers optimize their siting and forecasting efforts.

What's driving data center growth? 

The amount of data processed by data centers has been on a meteoric rise for nearly a decade, thanks to the growing number of connected smart devices, remote work, high-speed streaming and increased access to the internet. Between 2015 and 2019, data center workloads tripled, and today these critical facilities are an essential part of our modern economy. 

 Interestingly, the rapid workload expansion witnessed between 2015 and 2019 did not correspond to a significant uptick in energy consumption, with consumption remaining at roughly 200 TWh per year during that time frame. This can largely be attributed to improvements the industry made around energy efficiency, which offset any increases in demand. Today, however, the landscape is entirely different. The explosion of power-hungry artificial intelligence (AI) applications is causing data center electricity loads to skyrocket well beyond the mitigation capabilities of energy efficiency programs.  

AI models, especially those involved in generative tasks such as those performed by large language models (LLMs), require considerably more electricity than typical internet browsing. With the increased power demand of AI, global data center power consumption is expected to increase 160% by the end of the decade. Today, data centers account for 1% to 2% of total global power consumption; by 2030, it’s expected to be 3% to 4%.  

Key considerations for data center siting

As AI adoption accelerates across every sector of the economy, data centers are rushing to keep pace with the growing demand for facilities to house the technology’s supporting hardware and infrastructure –– and that means building more data centers. However, challenges around power and water availability, weather, and even local regulations and policies make data center site selection in the AI era increasingly difficult.  

Fortunately, robust analyses can help optimize data center site selection. When considering site locations for these power intensive data centers, it’s important to take the following into account: 

• Electricity demand forecasting - This analysis forecasts the amount of electricity the new facility will require at any given time. This data point is key to understanding whether a site will be suitable for a data center.  
• Generation capacity - It’s also important to understand whether the region has enough available power to reliably meet the new facility's electricity demand.  
• Transmission congestion - Electricity delivery is another key component of any siting analysis. In an electricity market, a power plant may be able to produce electricity at a competitive price, but it’s critical to understand whether that power can be transmitted to the data center.  
• Proximity to renewable energy - Data centers are under increasing pressure to decarbonize their operations. As a result, access to renewable energy, such as wind and solar, can be a crucial data point in site selection. Equally important is access to energy storage, which mitigates the intermittent nature of wind and solar.    
• Solar and wind Forecasts - Solar and wind power generation capabilities vary by location, so having accurate data for each potential site is critical. These forecasts help operators better understand the risks of a given location and better predict the impact of any planned on-site renewable energy assets.
• Production cost forecasting - Accurate energy price forecasts across any time interval allow data center operators to understand the actual cost of operating a data center at a given site.  
• Water access - The high-performance computers and graphic processing units (GPUs) used in AI applications generate tremendous amounts of heat. Data centers use large amounts of water to cool these systems, so the ideal site will have ample access to water.  
• Policies - State and local governments looking to attract high-tech industries often offer incentives that make investing in a particular region financially attractive. Incorporating these incentives in the site selection process ensures that decision-makers have a complete and accurate understanding of the benefits of selecting a given site.  

Energy Modeling with PLEXOS® for data center site selection

Gathering, managing and leveraging all these critical data points has made modern data center site selection increasingly complex. Fortunately, energy modeling with powerful solutions like Energy Exemplar’s PLEXOS® Platform, can simplify the process of analyzing and visualizing data needed for data center siting decisions.  

PLEXOS® features state-of-the-art data center siting capabilities that allow for modeling of traditional and renewable power accessibility, transmission congestion analysis and the development of highly accurate demand forecasts for each new data center. The platform also allows data center owners to determine where data center assets should be built for optimal outcomes and with the lowest risk.  

Highly visual, the platform is designed to enable faster, better-informed decisions around the value of potential data center sites. PLEXOS® also excels at nodal analysis and Monte Carlo stochastic analysis, which means the platform can handle a wide variety of hypothetical future states, empowering decision-makers to identify which have the best benefit-to-cost ratio.