How AI is Fueling the Rapid Growth of Data Centers

February 5, 2025 • Ella Krygiel, BOMA International 

As the demand for data processing and storage continues to surge, data centers are experiencing unprecedented growth. McKinsey’s analysis “suggests that demand for AI-ready data center capacity will rise at an average rate of 33 percent a year between 2023 and 2030 (reflecting a trend that is already underway.)This expansion is transforming data center design, resulting in increased power consumption and new cooling strategies. In fact, Axios reports that data center jobs have doubled in Washington state since 2018. These changes are not only critical for the technology sector but also have broader implications for global power grids and sustainability efforts. 

To provide perspective on these pressing topics, we turn to Kristen Vosmaer, Managing Director at JLL Data Center Work Dynamics. With extensive expertise in the data center industry, Vosmaer addresses three key questions that illuminate the future of data centers and their essential role in the changing technological environment. 

  1. How are the latest advancements in GPU technology changing the way data centers are designed and powered, and what does this mean for the future of AI? 

As JLL’s 2025 Global Data Center Outlook report determined, graphics processing units (GPUs) are becoming substantially more powerful, which is enabling faster training of large language models. According to the Nvidia GPU-Ready Data Center report, they summarize the difference that scaling with GPUs (Graphics Processing Units) can make for today’s data centers. In their description of GPUs, “GPUS drive AI with massively parallel compute and optimized high-memory bandwidth, enhanced for AI-class matrix multiplication and convolution.” Simply put, GPUs in the data center sector are what helps speed up complex calculations and process large amounts of data quickly, making advanced applications in AI, machine learning and data analytics possible. Vosmaer’s insights align with Nvidia’s findings. “The transition to higher GPU (Graphics Processing Unit) workloads results in data centers having to operate at a higher density per rack,” explains Vosmaer. “The total power utilization is the same, however, that power services fewer racks that operate at a much higher rate and generate higher temperatures.” This is an important consideration as we continue to understand the evolving demands placed on data center infrastructure. 

In addition to the advancements in GPUs over the past two years, liquid cooling technology, once viewed with skepticism, is now considered a “necessity for those responsible for deploying the next generation of high-density compute,” according to Semiconductor Engineering. Their research explains that liquid cooling technologies, such as the Direct-to-Chip Liquid Cooling method and the Liquid Immersion Cooling method, offer significant benefits for managing higher power densities. This is because they “minimize the reliance on air as an intermediary,” which is essential for repurposing waste heat and utilizing significantly less water. Vosmaer agrees with this evolution, stating, “Often the traditional air-cooled facilities will not be able to handle the extra heat being produced.” In addition to the need for advanced cooling solutions, Vosmaer highlights broader industry trends: “As an industry, the transformation is well underway for the large greenfield data centers the hyperscale companies have planned for AI chips. Concurrently, adaptive reuse of existing data centers is also being looked at to meet future inferencing demand.” This shows how data centers are evolving to keep up with the growing needs of AI and machine learning. 

  1. How are data centers impacting global power grids, and what steps are being taken to ensure they can meet the growing electricity demands? 

According to GlobeSt., spending on data centers could help grow the US economy by 10 to 20 basis points by the end of 2026, J.P. Morgan estimates. As part of this anticipated growth, their research found that due to the increase in data center spending of $20 billion, “power consumption is expected to reach record highs over the next two years.” This projected increase in power consumption highlights the critical need for collaboration. Vosmaer notes, “As the industry continues to grow, utilities and data center developers have been working actively together to provide solutions that enhance the capacity and resiliency of the grid.” Developers are proactively addressing challenges by building new substations, investing in transmission lines and increasingly incorporating onsite power sources to supplement the grid.  

While each region faces unique obstacles, Vosmaer emphasizes that “data centers continue to be an economic booster for the region, so staying ahead of the grid challenges serves all stakeholders.” This positive outlook is tempered by a recent Bloom Energy report, which highlights that “grid interconnection takes longer, there is more congestion on the network and capacity is increasingly expensive.” If the U.S. continues to build high-voltage transmission infrastructure at its current rate, it will take decades to meet the growing demand. 

  1. How is the shift to liquid cooling changing the design and operation of data centers, and what are the main challenges and benefits of this technology? 

Given the earlier discussion on cooling solutions, liquid cooling is becoming essential for high-density racks. According to the JLL 2025 Global Data Center Outlook report, a hybrid approach is typically employed today, with 70% liquid cooling and 30% air cooling. This shift is crucial as immersion cooling will likely become a common thermal management strategy as GPUs push above 150 kW per rack. However, while these technologies promise many benefits, they also face challenges related to liquid quality, reliability and maintenance. 

Vosmaer emphasizes the importance of transitioning to liquid-cooled designs, noting that “this offers several benefits, including reduction in equipment needed, thus less consumables and embedded carbon. Closed loop systems allow for zero water systems, and that ultimately reduces the average footprint by accommodating high powered servers.” He also points out that “there will be a learning curve as the industry advances the supply chain and operational aspects to this technology,” which may lead to challenges around equipment. 

Overall, data centers are rapidly evolving to meet the growing demand for AI, leading to increased power consumption and the adoption of advanced cooling technologies. These changes are crucial for sustainability and require collaboration between developers and utilities to enhance grid capacity. 

Interested in more content like this? Read our recent articles, Data-Driven Approaches to Warehouse Inventory Tracking or 2025 Trends: Technology and Talent in Industrial Real Estate. You can view all this content and more when you click here to sign up for our Industrial newsletter!