Data Center-Related Investments Available Across Most Asset Classes

It’s no secret that the investment potential of data centers is huge. According to information from investment firm Colliers, $580 billion was invested in data centers in 2025. That number will be eclipsed by the reported $750 billion hyperscalers including Amazon, Microsoft and Google plan to spend this year.

For investors, there are a variety of ways to participate in the data center buildout. Bessemer Venture Partners calls this the data center “stack.” Beyond the data center itself, investors can gain exposure to the compute boom by investing in the land, power generation, power transmission, construction, maintenance and cooling costs. Investors can also participate in the decommissioning of older data centers as hyperscalers refresh their computing power with newer, higher-performing equipment.

Bessemer noted that all of these components make up the ecosystem necessary to power artificial intelligence and they each have their own unique challenges. For investors, there is a diversity of opportunity and many of the themes will develop over the next several years.

“We are seeing a variety of strategies emerge,” explains Alan Synnott, global head of real assets at Mercer. “There are large-, mid-, and small-cap opportunities as well as financing [and] private placements. Where investors ultimately choose to invest is going to depend on whether they see it as a sector exposure or if they want to go deeper and look at asset-level exposure. We advocate doing a total portfolio analysis so that investors can understand what they might already own and how best to add on. This can help mitigate concentration risk.”

Building Digital Infrastructure

Data centers can be tricky assets to place into portfolios. Depending on what piece of a data center an investor owns, that exposure can be part of the equities portfolio, credit, real estate, or infrastructure. As investors diversify their exposure to the opportunity set it is likely that all asset classes will have some link to digital infrastructure.

Mercer’s Synnott sees some similarities with how the growth of private financing in the energy sector has evolved over recent years. “We are starting to see a bit of differentiation in terms of the development of both generalist and specialist funds in this space,” he says. 

Adam Patti, CEO of VistaShares agrees. VistaShares manages a listed AI Infrastructure strategy and is also moving into private markets. “What we’re seeing right now is really a race among sovereigns to develop AI technology and build up their available compute. As AI models become more powerful and specialized, they are going to need more resources,” he explains. “From our perspective if you want to make money in this as an investor, you have to look at the supply chain and understand where the money is being deployed and critically who is making money right now. All data centers need the same things to get up and running so you have to see where the money is being spent and where the bottlenecks are because that can impact profit.”

Patti adds that the bottlenecks are not necessarily indicative of long-term problems and most will likely evolve over time. “If you think back to when we were very early in the AI cycle, it was memory. They couldn’t get enough memory to support running the models. So providers had to respond and that is still ongoing. Now we’re seeing a bottleneck emerge around electricity because of demand on the grid from the new data centers that are fully up and running. So there are going to be opportunities there as utility providers respond.”

Both Synnott and Patti say bottlenecks are a natural part of building out any type of significant infrastructure. There are also points where investors might see pockets of upside as demand increases for particular things. Another recent example is copper. Copper prices started rising in late August and are likely to remain elevated as a result of demand from data centers. The potential for higher tariffs on copper from Canada, as a result of the latest breakdown in negotiations between the Trump administration and Canadian authorities, could put further upward pressure on prices and provide adjacent opportunities for investors if they see a point of entry via the metals market.

Cross-Aasset and Stakeholder Disruption

 The size and scope of the digital infrastructure needed for AI is not the only unprecedented aspect of this technology. AI’s potential for cross-asset disruption throughout the total portfolio is high. Many software investors are already seeing immediate and profound impacts to their portfolios as a result of the rise of generative and agentic AI. Given this, institutional investors are also signaling that they are keenly focused on risk management as part of their investment strategy.

Speaking during a July board meeting, Eric Lang, senior managing director for private markets at the Teacher Retirement System of Texas said, “We are thinking about every investment we make and asking is this going to be interrupted by AI?”

Texas Teachers has created an investment task force around AI and identified five key themes. The task force sees AI as an industrial buildout via data centers and power generation; a business-model disruptor; a labor and workflow reallocator as a result of potential efficiencies derived from AI; a security and governance question, and finally, as an uneven force across the global economy. “The U.S. is a leader in AI at the moment,” Lang said. “But we understand that could change.”

The interlocking nature of those five investment themes could also lead to unexpected concentration risk if not actively monitored, Lang added. “We understand that this is a transformational investment opportunity. But at the same time, it is complex.”

That complexity is also front of mind at the California Public Employees’ Retirement System, which has been a significant investor in data center buildout in California and nationally. “CalPERs wants to play a role in ensuring that data center buildout is done responsibly, to get the benefit of increased compute because the potential for positive change coming from AI is significant,” Peter Cashion, managing investment director of sustainable investments at CalPERS says. “The trick is to tap that without causing damage in these earlier stages of development.”

Cashion says that risk management is key to the buildout of the new digital infrastructure. There are a lot of opportunities but given how early in the development cycle it is,  it’s important not to lose sight of fundamental investment principles, he notes. Projects still need to have clear-cut plans and pathways to profitability. He says investors should maintain their long-termist perspectives and not get caught up in a fear of missing out. The opportunity set for AI is vast and will evolve over the medium and long-term.

“We are very attuned to responsible data center buildout in terms of maximizing water efficiency, energy efficiency, and where you put your power demand so that it doesn’t disrupt prices for consumers,” Cashion says. “I think it’s in everyone’s interest to manage the buildout in a way that takes into account all stakeholders. Otherwise you risk a legislative reaction, particularly at the state level, which we have already seen in some states including New York.”