Artificial intelligence (AI) has been a significant driving force in recent years for infrastructure investors, who have shoveled money into massive data centers and paid premiums for the generation assets powering them.

But that’s not the only way AI is shaping infrastructure investment. AI use is becoming increasingly ubiquitous in managing and operating assets and is being used as a tool when considering investments.

Many kinds of AI

One tool is agentic AI, a form of generative AI that can autonomously plan and make decisions with minimal human oversight, according to a report published last week by EY-Parthenon. Agentic AI could help manage the complexity, fragmentation, and siloing in the sector, as specialist AI agents “reason over data from multiple sources, coordinate with one another, and propose or execute actions within human-defined guardrails.”

“Infrastructure is becoming harder to govern through human coordination alone,” the report continues. “Agentic AI makes a different future possible: not by removing the systems, functions or organizational boundaries on which infrastructure depends, but by creating an intelligence layer above them that can reconcile fragmented information and deliver decision-ready intelligence to the people responsible for judgment and decision.”

The report was spurred by the need to address a massive global funding gap in infrastructure, according to Steve Lewis, EY’s global infrastructure technology leader and one of the report’s authors.

“By our own published definitions at EY, we’re about USD 64trn short of the funding we need globally to deliver the infrastructure that we need as a global society by 2050,” Lewis told this publication. “There’s a need to adopt technology and try to find something to make us more productive and able to keep up with the infrastructure demands of society that is ever-changing.”

Portfolio management

Investors are already using AI to manage their portfolios. Will Schleier, a senior managing director and co-lead of the Portfolio Value Creation team at Stonepeak, said he has been working with portfolio companies on AI use cases over roughly the past 18 months.

“We get visibility into what is working across a range of businesses, and then we take those use cases and test them at our other businesses,” he said. “Some of the best ideas for our portfolio companies have come by applying it to other companies in our portfolio or their partner companies. We are focusing on a few functional areas that seem to be the best, most fruitful areas to look into for AI use.”

Schleier pointed to using AI agents to field customer inquiries, back-office functions such as analyzing legal documents, and IT and security applications as a few of the most fruitful uses of AI. By analyzing the application of AI across a portfolio of companies, he noted, it is possible to have a more informed view, making navigating the nascent technology more manageable and effective.

“There are a lot of companies out there that feature untested new technologies saying that they are the AI partner of choice, and that companies can use them to supercharge their operations. But it has been such a short timeframe for enterprises using AI that there’s not a long track record of these businesses demonstrating success,” Schleier added. “So, our portfolio companies find it very valuable that we are able to help them find vetted partners and solutions for them.”

Schleier points to one transportation logistics business in which using an AI agent to assist the fulfillment team has cut the cost per order dramatically and reduced error rate to near zero.

“When there’s high-volume, repetitive tasks, it takes up a lot of employee time,” he said. “We found that in those instances you can integrate an AI solution, and it gives the employee base a lot more efficiency, and allows them to process a lot more.”

In a 2026 operational briefing in February, Ben Way, head of Macquarie Asset Management, said the firm was using AI to help leverage the firm’s vast portfolio and investing background. He said Macquarie Asset Management is rolling out Chronograph, an AI tool that allows asset managers to see what’s happening in their portfolio companies.

“It allows them to compare across regions, across different funds, across different sectors,” Way said, according to a published transcript. “And then as you can imagine, as the technology becomes more proficient, we then overlay an AI solution to that, something like Claude, and that allows us then to upgrade further our investment decisions.”

Intelligent investments

Investors are also using AI in ways that shape their dealmaking. Many are using it to cut down on repetitive tasks, or what some would consider grunt work.

“We are using it as a research tool to enable better screening and desktop due diligence,” Robert Shaw, managing director of CBRE IM’s private infrastructure strategy, said. “We also use it a lot for the more mundane, administrative parts of investment management, which is really first cuts of decks, presentations and materials.”

EY’s Lewis said that Agentic AI would be a tool to make skilled employees more efficient. He pointed out that engineers, as one example, may spend 20% of their time doing engineering work and the remainder doing tasks that require less skill and expertise.

“If you think about a lot of that back-office, non-frontline engineering, surveying or architectural services, finance or environmental advisory, whatever it might be; there’s a lot of stuff that goes on in the background that is important,” Lewis said. “It’s important. But does it have to be done by the people that were trained to do something else? Probably not.”

Others are going further. Justin Johnson, Arevon’s CEO, said he has found AI useful for a range of purposes, and has made it a key performance indicator (KPI) for everyone in his organization to explore how AI can make their job easier and more productive.

“I think it is going to impact every job in a meaningful way,” Johnson told this publication.

Johnson has also created an agent for himself, feeding it information to reflect how he writes and thinks.

“I can ask AI strategic questions, or ask it to play devil’s advocate to something I am proposing,” Johnson explained. “It has meaningful, pretty fantastic insight, even for a C-suite level function.”

Johnson has also used AI to give suggestions about lobbying strategies, financial metrics, and investment ideas.

“It comes back with interesting analysis,” he said.

Shaw, meanwhile, said he’s heard of some investors using AI as an investment committee member.

Proceed with caution

There are risks to relying on AI, the investors said.

AI occasionally experiences “hallucinations,” confidently conveying false or inaccurate information. One investor said that he’s found AI to be around 90% useful and accurate, but noted that facts must be verified.

While AI is generally used to make employees more efficient, exploring different uses could turn into a poor use of time, Schleier said.

“AI is best used when it’s enabling them to do their day-to-day jobs better,” he noted. “Not when it’s distracting them from their day-to-day execution responsibilities.”

And even eliminating seemingly mundane and repetitive tasks could come with drawbacks, Schleier continued.

“I learned a lot over the first part of my career from doing the financial analysis, the modeling, presentation preparation, myself. Checking that everything’s right, making sure that I understand where all the information came from,” he added. “We are very careful about balancing the risk of potentially eliminating some of the training benefits that come from doing the work yourself.”

Not a question of ‘if’

EY’s Lewis agrees that questions of judgment should be left to humans.

“Certainly, we can take reference points, and they should be reference points, but we are responsible for the interpretation and the governance,” he said. “That’s critically important.”

But there are also risks to eschewing AI, Lewis notes. In the past, he noted, an asset’s faults would often be detected by failure.

“We’d notice the failure and we’d see that in a control room, or we’d see it physically as something that’s going wrong,” he said. “We can [now] much more quickly determine where something is going wrong.”

“We can run thousands of scenarios in minutes or hours, whereas it would take an army of humans a very long time to consider all of the different ramifications of a failure in a certain point,” he adds.

Much of EY’s report focuses on silos within the infrastructure sector. But rather than “breaking down” these silos, the report explains, AI can form an intelligence layer above them to make connections “more automatic, more scalable and less dependent on human-driven coordination.”

“Decision latency, coordination overhead and information trapped within silos continue to absorb significant professional effort and budget across global infrastructure delivery, at a scale likely measured in the hundreds of billions of dollars annually,” the report concludes. “The critical shift is from isolated automation to connected intelligence.”

AI-led change is inevitable, the report continues. Infrastructure leaders can either play a role in shaping it, or wait for others to do so.

“Yes, there’ll be speed bumps,” said Lewis. “There are going to be little things in the road that cause us some challenges. But aren’t there always?”