The AI build-out, including data centres and supporting infrastructure, is creating one of the largest financing opportunities in modern markets. Yet investors must navigate growing competition, supply constraints and concerns over stranded assets as private credit establishes its role in the field and takes advantage of a $700bn opportunity. Aysha Gilmore reports…

The momentum behind artificial intelligence (AI) has accelerated rapidly and has led to significant concerns for software companies – a mainstay in private credit portfolios. And while much of the financing for AI-related businesses has been supplied by the public markets, private credit managers are keen to get in on the action.

But to do so, investors need to navigate growing competition, supply constraints and issues around stranded assets.

The sheer scale of capital required to support the build-out behind AI means there is a big enough pie for everyone to grab a slice.

The development has been a central topic in credit markets over the last six months, with the “AI capex cycle now in full swing,” according to Mahmoud El-Shaer, fixed income portfolio manager at Wellington Management.

No longer are tech giants such as Google and Meta funding AI investment solely from their own profits. Increasingly, they are turning to external financing, with significant debt issuance and even equity issuance from the hyperscalers.

Meanwhile, many private markets managers are building out capabilities to invest through their infrastructure arms. Last month, private capital heavyweights Apollo and Blackstone partnered with Broadcom to launch an AI infrastructure platform backed by $35bn (£26.2bn). Overall, the platform is expected to provide 20 gigawatts of compute capacity for AI labs through to 2028, the firms said.

KKR has launched Helix Digital Infrastructure, a new company established in June to finance AI infrastructure and other alternative asset managers, including Ares Management, have set multi-billion-dollar fundraising targets for data centre investments.

However, this financing is still only scratching the surface. According to Morgan Stanley research published in May, global data centre capex (excluding power) is expected to exceed $3.2tn through to 2028. Of this, approximately $1.75tn is expected to be financed through credit products, with private credit accounting for around $700bn.

Nascent field

AI infrastructure investment remains a relatively “nascent field” for private credit, explains Anant Kumar, global investment strategist at Benefit Street Partners.

Currently, the majority of funding is coming from public markets, including high-yield and investment-grade bonds, with private markets still too small to accommodate some of the largest deals. However, capital is beginning to flow through the asset-based finance (ABF) market, particularly where borrowers seek flexibility and speed, areas that play to private credit’s strengths.

Kumar tells Alternative Credit Investor that Benefit Street Partners, Franklin Templeton’s private credit arm with $93bn in assets under management (AUM), currently lends to data centres within AI infrastructure as part of its ABF allocations, and is looking at chip and turbine deals.

He explains that over the last six to nine months, the business case for AI has become “a lot clearer”, with value now being observed in workplaces and returns on investment becoming increasingly visible.

“I think some of the fear is that this is all a bubble, like the telecom bubble, that we are building all of this capacity right before the bubble pops. I don’t think this is the case,” Kumar says. “Right now, the AI build-out that is happening is being fully utilised – users are maxing out on tokens and data centres are running at full capacity. Unlike the telecom bubble, it is not as if you are building a bunch of speculative capacity. For this reason, more and more companies will raise money from all available sources – investment grade, high yield, direct lending, and ABF.”

Increasing competition

However, despite strong demand and a compelling investment case for AI infrastructure, the maturing data centre market and the declining risk profile raises questions over whether private credit can compete with banks in financing the sector. Alongside this, whether the improving fundamentals could ultimately result in overcrowding.

Joel Holsinger, co-head of alternative credit at Ares Management, recently told ACI that within the firm’s ABF portfolio, despite these areas still being attractive, “digital infrastructure and data centres are nearing the end of relative value for us”.

This is largely due to the significant volume of capital already flowing into the data centre space, particularly on the lending side, with banks becoming increasingly active, especially in the US, he says.

“Banks are building capacity in the data centre space, so there is a lot of concentrated risk,” Holsinger adds.

AI is increasingly dominating venture capital as well, accounting for around a third of the funding in the asset class in the UK and close to two-thirds in the US, with a handful of mega-rounds driving a growing share of overall investment.

Read more: Private credit could get ‘boost’ from higher rates despite software concerns 

SRT deals

However, Holsinger argues that while bank involvement is making digital infrastructure less attractive within ABF allocations, it is opening up deals within the significant risk transfer (SRT) market. As banks seek to expand AI infrastructure lending, they are also looking to shed exposure to data centres.

“We are seeing second derivative interest such as capital relief on that data centre piece, but the first derivative is less interesting to us,” Holsinger says.

Roopa Murthy, head of EMEA infrastructure debt at Ares, adds that this opportunity is becoming more prominent in Europe, where regulatory pressures are intensifying.

“We have seen banks face concentration issues in this space, and they are looking at innovative ways to reduce exposure,”  she says. “This includes working with private credit lenders through SRT-style structures.”

Although the US market might be becoming more saturated, European data centres and related AI infrastructure still need “huge” amounts of private credit investment, argues Anne-Laurence Roucher, group head of private markets at Edmond de Rothschild.

“Half of the world’s data centres are located in the US, whereas the US accounts for only around four per cent of the global population,” she says. “Europe is therefore likely to continue investing heavily in data centres, alongside power generation, given the significant energy requirements.”

Bubbles and constraints

Indeed, supporting infrastructure, particularly power and networking, is central to the investment case for data centres, and without it, the investment case weakens significantly.

Kumar notes that concerns about an AI bubble are less pressing than the structural constraints facing the sector.

“I’m not worried about the fundamental prospects; I am worried about about the ability of the frontier labs to actually get access to the compute they need,” he says.

Beyond these constraints, there is also the risk of technological obsolescence, says Roucher, with Edmond de Rothschild allocating to digital infrastructure through its infrastructure debt portfolio.

Data centres require upfront investment, yet the rapid pace of innovation in GPUs and computing architecture raises the risk of stranded assets for investors, she says.

“There has been more technology risk with infrastructure in the past decade as AI is very connected to data centres,” explains Roucher. “You need a lot of data centre capacity for AI, and you need a lot of energy for AI.

“So, the debt that is provided can be short-term to make sure that we cover the obsolescence risk as you are reimbursed quicker.”

Another fundamental bottleneck lies in hardware supply, with the semiconductor supply chain remaining highly concentrated, with companies such as TSMC producing a significant share of advanced chips. A key risk here is that any geopolitical disruption could severely impact AI and data centre development.

In addition, as data centres rely heavily on water for cooling, scarcity of the resource is an additional constraint on expansion and is a big risk for investors.

Climate risks

At the same time, electricity and water consumption, alongside construction activity, place significant strain on the environment. And despite the climate agenda facing headwinds in the US and elsewhere, participation in the AI infrastructure build-out can pose challenges for climate-focused investors such as European asset manager Ambienta.

“We appreciate it is a growing field, but we need to link that growth to environmental sustainability,” says Ran Landmann, co-head and chief investment officer of Ambienta’s credit strategy. “In and of itself, the growth of AI and data centres doesn’t address that.”

Nevertheless, Landmann tells ACI that Ambienta is exploring ways to invest in the AI build-out. While they do not currently invest directly in data centres, they are examining opportunities across the broader ecosystem, he says.

“It is a massively growing industry, and we need to find a crossroad as to where our companies play a meaningful role in this massive growth, while contributing to the environmental solutions that make it sustainable,” Landmann adds.

“We do touch on that arena, as you build out more data centres, there are technologies that support this build out to be more sustainable. Examples being companies that provide solutions to cooling technology to data centres, an area identified as a growth area for us.”

Read more: Managers hunt for software winners and losers amid AI panic

Doubts and market sell-offs

Despite the strong momentum, questions remain over whether expectations for AI have become overstretched. In early June, markets experienced a sell-off triggered by weaker-than-expected results from semiconductor company Broadcom, which failed to meet expectations for AI chip demand.

This triggered a broader tech rout, compounded by macroeconomic factors including stronger-than-expected US employment data. Although markets have since partially recovered, the episode highlighted underlying concerns about the sustainability of AI-driven growth.

Kumar, however, dismisses the volatility as temporary, suggesting that there “will always be pull backs as nothing never goes up monotonically”.

“If I just took one part of the semi’s space, which is memory, memory demand is backed up for a multi-year period and I don’t think a one-day sell off affects that thesis,” he says. “So, I would argue this daily volatility is mostly noise.”

For now, the consensus appears to be that an AI-driven downturn is unlikely in the near term. And with the predictions estimating that private credit needs to provide $700bn for the build-out it is inevitable that the industry will further invest in AI infrastructure and its supporting backbone.

This article originally appeared in the July edition of the Alternative Credit Investor magazine, to view the issue click here.