Seven key considerations for data center investors

In view of the many moving parts involved in data center development, would-be investors in the sector need to consider the following factors when assessing projects: 

1. Forecasting actual energy needs: AI could potentially require significantly less energy in the future due to a combination of hardware and software advances, as well as a shift towards more efficient practices. The launch of DeepSeek in January 2025, for example, demonstrated that powerful large language models (LLMs) could run on considerably less energy than previously thought. 
2. Predicting compute power needs: The AI sector is evolving so quickly that it is difficult to accurately predict future demand for compute power, and therefore to decide how much to invest in it. Consultancy McKinsey, for instance, envisages three different scenarios based on how quickly companies can create meaningful AI use cases, differing intensities of experimentation and training across the AI market, and varying degrees of efficiency improvement in AI technologies. Depending on the scenario, the investment needed could be as low USD3 trillion between 2025 and 2030, or as high as USD8 trillion (see Figure 2). 
3. Advances in cooling technology: cooling is crucial to data center operation, accounting for about 40% of energy consumption. Replacing air-based cooling with liquid cooling promises to reduce power usage by as much as 90%. As liquid cooling is deployed more widely globally, it could significantly alter the cost calculus for data center investors. 
4. Location limitations: the rise of “Edge AI” will result in a large share of AI tasks being processed on edge devices, such as smartphones, Internet of Things (IoT) devices, or embedded systems, rather than on centralized cloud servers. This will require the development of smaller data centers on sites near population centers, rather than mega facilities in remote areas. But urban sites with adequate power and room for expansion are in short supply. 
5. Reliability of revenue streams: One of the attractions of data centers for investors is their potential to provide steady, utility-like cash flows. Co-location arrangements, whereby data center capacity is leased to customers on a long-term basis, provide a degree of revenue security for projects. 
6. Supply chains: Lead times for procuring essential equipment such as backup generators have increased from months to years and there are mounting concerns about potential shortages in crucial components such as advanced AI chips. This makes supply chain management critical to project success. 
7. Security and regulatory considerations: Data centers need to be secure from the constant threat of fire and cyber-crime. Investors also need to keep abreast of evolving regulations on data sovereignty as governments may increasingly require governance and control of data generated or processed within their countries’ borders. Data center investment could also be hindered by bureaucracy, including licensing and permitting delays, onerous investment screening procedures and unpredictable foreign investment rules.