KEY TAKEAWAYS
· The amount of data in the world is growing rapidly. In 2024 alone, the world is expected to generate 1.5 times the amount of digital data it did just two years ago.
· Two megatrends are driving the demand for data center capacity and will continue doing so in the years to come: organizations migrating their data from on-site servers into the cloud and the rise of artificial intelligence (AI).
· Hyperscale centers, which are large facilities, have dominated data center development to date. These centers tend to cluster in specific areas, which have limited supplies of power and space. However, as AI develops and other technologies rise to the fore, such as the internet of things, different types of data centers – and different types of investment opportunities – may arise.
· Not all data centers are created equal: Knowing how to properly value and future-proof investments, strong relationships with the largest tech companies, and downside protection are key to investment success, in our view.
Just as railways and pipelines supported the world’s industrialization in the 18th century, digital assets provide the critical support for contemporary society’s increased reliance on data. Networks of data centers are like a new utility, housing the entire value chain of the digital economy. They support the production of massive amounts of new data, similar to farming and mining in the physical economy; they refine and analyze data to produce something new, like manufacturers, and they help to distribute data across transmission networks, like logistics hubs and warehouses.
As a commodity, data is mushrooming in size. In 2024 alone, the world is expected to generate 1.5 times the amount of digital data it did just two years ago (Exhibit 1). Demand for data centers has grown exponentially, too, with vacancy declining all over the world and at a decade-low in North America.1 Today, most existing data center capacity is under lease, with projects and campuses sold forward further and further out. The ongoing shift to cloud computing and the emergence of artificial intelligence (AI) are megatrends that promise even faster growth in the years to come. AI encompasses a broad range of capabilities focused on creating intelligent systems, including machine learning, which involves developing algorithms that learn from data to make predictions or decisions, and generative AI, which leverages machine learning to generate original and realistic content.
EXHIBIT 1: Recent Growth in Data Generation
Investment into data centers and general hype about the sector has also reached a fever pitch. We think the excitement in this area is justified given the global trends behind the tailwinds, but as investors, we also acknowledge the challenge of building and operating data centers, staying abreast of the fast-changing technological landscape, and maintaining discipline when deploying capital in a heady environment. We believe that a careful, risk-based approach may lead to better long-term outcomes than acquiring data centers at any price just to obtain exposure.
Data Center Megatrends: The Cloud and AI
It is important to understand the two megatrends driving the explosion in data center demand and investment activity: cloud computing and AI.
Businesses, governments, and organizations have been moving much of their data to the cloud over the past 10 years. Essentially, enterprises are moving away from storing data and hosting applications on site and instead are choosing to use software-as-a-service (SaaS) applications available over the internet. This typically requires the support of cloud providers who work with organizations and SaaS providers to store, process, and compute data in racks of servers housed in data centers.
A significant portion of enterprise workloads have already moved to the cloud, and more are expected to do so in the future. According to Credit Suisse, the three largest cloud service providers saw compound annual revenue growth of some 28% a year between 2015 and 2022, when revenues reached $500 billion. Future spending on cloud initiatives, meanwhile, is projected to again quintuple cloud revenues from some $500 billion in 2022 to $2.5 trillion in 2032.2
Even while the cloud computing cycle is still at a relatively early stage, new themes are emerging that require increased data computing, processing, and storage capabilities. The most prominent example is AI, a technology with such an enormous appetite for data and processing power that experts suspect it will soon overtake cloud computing as a data center demand driver (Exhibit 2). The largest tech companies are engaged in an arms race to create new and better forms of AI, which will in turn require sophisticated, powerful chips and servers housed in the newest, most up-to-date data centers.
EXHIBIT 2: How Data Center Workloads Are Expected to Change
The Data Center Investment Opportunity
Supported by these trends, global data center demand is expected to grow at a compound annual rate of 12%-15% between now and 2030 (Exhibit 3), while construction spending is expected to sum up to more than $303 billion between now and 2030, according to Synergy Research Group. While there are several different types of data centers, hyperscale centers have been the largest growth area to date and are expected to dominate growth for the coming years. These are very large facilities that can achieve economies of scale and superior performance because of their size. Hyperscale providers have more than quadrupled capital expenditures between 2015 and 2022 to $122 billion3 and are increasingly turning to trusted third-party infrastructure providers to meet their growth needs.
EXHIBIT 3: Forecast Global Data Center Demand
Hyperscale providers tend to cluster in specific areas around the world, including Northern Virginia in the United States, London in Europe, and Singapore in Asia Pacific. There are a variety of reasons for this, ranging from the need to be physically close to important connections and to transmit at high speeds, cloud architectures that require multiple independent nodes in close proximity, and a concept called data gravity, in which applications, services, and data tend to gravitate toward areas that are already storing and processing large amounts of data. Data gravity is intensifying exponentially and increasing compute and storage demand in the incumbent hyperscale markets, which have finite space and power—two elements that data centers need in large and scalable amounts. This structural scarcity provides a potential opportunity to invest both in the data centers themselves and the services supporting their operations.
AI and other trends may continue to open up demand in new markets and for different types of facilities. For example, data centers used for different purposes may have different requirements:
·To train an AI, data centers need enormous amounts of computing power, but have a lesser need to be close to large population centers than a data center used in cloud computing.
·For AI inference models, which solve problems or create predictions, data centers need to be both very close to end users and to experience low latency, or delays in transferring information.
Therefore, data centers that focus on AI training may be in more remote locations, ideally close to cheap and plentiful power sources, while those that focus on inference may need to be located near application users and commercial centers.
Looking ahead, next-gen applications, including augmented and virtual reality, the internet of things, and of course, the evolution of AI, should further expand the need for data centers and the functions they serve. Edge data centers, which are located near the enterprises they serve, could become more prevalent. That said, we believe it remains unclear whether demand for edge centers will materialize at any significant scale beyond major metropolitan areas.
The Ingredients for Success in Data Centers
Data center investments have many attractive aspects, including incredibly strong demand growth and an upper limit on supply due to scarcity of space and power. However, not all data centers are created equal, and broad exposure does not guarantee investment success. We think three key principles can help secure a differentiated risk-return opportunity.
1. We cannot stress enough the value of technical expertise and experience. The hyperscale players that are the largest customers for data centers are also the world’s most advanced consumers of the product. These customers are extremely selective about the vendors they use and are also discovering that there may be cost benefits to consolidating the number of relationships they have over time. Strong relationships with these players, as well as a demonstrated track record of success adding value for customers by building and operating data centers, has been key in our experience to winning trust and repeat business.
2. Knowing how to value these investments is also critical. We have observed other investors acquiring legacy data centers that may no longer be suited for tomorrow’s computing needs or that may not have the power sources in place to support costumer growth plans. It is important for data center players to be able to call on experts in areas such as real estate, energy, technology, policy, and climate issues to help identify and appropriately underwrite digital infrastructure investments.
3. Finally, we value downside protection, even in a market that looks so attractive today. Our risk-based approach to infrastructure dictates that we look for exposure to assets with contracted revenue profiles, tangible value, or other protective features, as well as making sure that our long-term investments are aligned with customers and their roadmaps for growth.
1. Source: CBRE as of July 2023.
2. Source: IDC, Morgan Stanley Equity Research as of March 2023.
3. Credit Suisse Equity Research as of January 2023.
