Story of the month
DATA CENTERS: ARTIFICIAL INTELLIGENCE IMPLICATIONS
The global AI data center (DC) market could reach 50-60 GW within 5 years, and traditional DCs could experience incremental growth of 15-20 GW over the same period. At an all-in cost of $50mn/MW (including land, construction, power supplies, hardware, and connectivity), the total market opportunity for AI DCs could reach $2.5-3.0 trillion. However, McKinsey forecasts that US data centers demand alon will reach 80 GW by 2030 and 35GW in Europe.
A DEEP DIVE INTO THE WORLD OF DATA CENTERS
Data centers represent the very essence of global connectivity and the cornerstone of the technological revolution. Data centers are physical locations for the computing infrastructure that IT systems require, such as servers, data storage drives, and network equipment. It is the physical facility that stores a company’s digital infrastructure and makes it accessible from anywhere.
We can also distinguish between AI DCs and traditional DCs. AI DCs, as the name suggests, are suited for AI-related activities such as AI training and AI inference. Since training is somewhat latency agnostic, AI-training DCs can be located in more rural, less developed areas that offer land and power at a larger scale. Compared to traditional DCs that require 5-10 KW per rack, AI DCs require 60 KW or more per rack.
DCs are also divided into 3 categories. Hyperscale DCs are large facilities used for off-premises cloud applications such as Microsoft Azure and AWS. The second type are colocation/multi-tenant DCs, where a DC owner sells space, power, and cooling to multiple enterprise and hyperscale customers in a given location. Finally, enterprise DCs are smaller facilities that are owned and operated by the company they support.
WITH AI DRIVING NEAR-TERM DEMAND FOR DATA CENTERS, SUSTAINABILITY TARGETS MAY TAKE A BACK SEAT
The surging adoption of digitalization and AI technologies has amplified demand for DCs across the globe. Sustainability is becoming less of a priority for hyperscalers as speed to market has become the top concern. In previous years, sustainability was a top-
five consideration when building DCs but has become a bottom-five factor considered by companies as the need to get DCs online and in a short timeframe has outweighed all other factors. However, the DC industry needs to decarbonize and achieve net zero by 2030–2040. This creates an opportunity for countries with abundant carbon-free energy and lower temperatures that could be targeted as they represent a strategic location and reduce the need for cooling- related power consumption.
DATA CENTER GROWTH DRIVES GROWTH AND OPPORTUNITIES IN RELATED MARKETS AND INDUSTRIES
The boom in DCs is driving growth in several other industries. Firstly, liquid cooling is gaining popularity over air cooling, growing at c.46% compared to c.13% for air cooling. Liquid cooling has been shown to be more energy efficient than air cooling and could result in c.25-30% efficiency savings. Investment in cooling is expected to grow at a CAGR of c.18% to 2028. Secondly, energy consumption of DCs has been growing at c.17% per year and, according to McKinsey, is expected to continue to grow at c.13% per year in Europe. Finally, DCs are also driving infrastructure growth as security requirements for DC buildings are high.
As a European small/mid cap player, we aim to benefit from the growth momentum in DCs and have positioned ourselves at several levels of the DC value chain. Through our investment in Munters, a leader in climate control solutions, we have gained exposure to both air and liquid cooling for DCs. Finally, our recent investment in Hill& Smith, a UK infrastructure leader with operations in the UK and US is well positioned to benefit from DCs growth by providing infrastructure security solutions such as fences, gates, sentinel road blockers, lighting, galvanizing, …
