Key Takeaways:
Goldman Sachs expects private infrastructure and real estate capital to take a larger role financing the AI data center boom.
Goldman Sachs raised its combined capex forecast for the four largest hyperscalers to $5.3 trillion between fiscal 2025 and 2030, saying private infrastructure and real estate capital will play an expanding role in financing the AI-driven data center buildout.
"Private infrastructure and real estate will play an even larger role in the years ahead," Goldman Sachs said in a note Tuesday, pointing to the blurring boundaries between the two asset classes as data center projects span land, power, building and equipment.
The Wall Street brokerage previously forecast $4.5 trillion in combined capital spending for Meta Platforms Inc., Microsoft Corp., Amazon.com Inc. and Alphabet Inc. before first-quarter earnings. The private infrastructure market grew at an annualized rate of roughly 11.5% from 2021 to 2024, and Goldman expects that pace to accelerate toward the 16% to 17% rate that prevailed for much of 2012 to 2021 — pushing infrastructure assets under management comfortably above $3 trillion by 2030.
The shift comes as traditional bank lending reaches capacity for the scale of funding required. BlackRock Chief Executive Officer Larry Fink has estimated the nationwide buildout of data centers and energy infrastructure could total $10 trillion over the next decade, with much of that capital drawn from Americans' retirement and savings accounts through index funds and pension allocations.
Infrastructure AUM on Track to Triple
From 2021 to 2024, the private infrastructure market expanded at an 11.5% compound annual rate, Goldman said. Accelerating that pace to the 16% to 17% range — the historical norm from 2012 to 2021 — would push infrastructure AUM past $3 trillion by 2030, up from roughly $1.5 trillion at the end of 2024, according to the brokerage's estimates. The growth reflects structural tailwinds including the data center boom, energy transition spending and digitalization of industrial assets.
"Infrastructure sits at the epicenter of multiple structural tailwinds, which we expect will drive its growth and provide additional capacity for financing," Goldman added.
Blurring Lines Between Infrastructure and Real Estate
The traditional distinction between private infrastructure and real estate is eroding as data center projects require investment across multiple categories simultaneously. A single campus may involve land acquisition, power procurement, building construction and equipment installation — each falling into a different asset class under conventional classification systems.
Goldman said private infrastructure's structured income generation and inflation-protection characteristics will likely boost further growth. The asset class has attracted interest from pension funds, sovereign wealth funds and insurance companies seeking long-duration, inflation-linked returns.
The financing opportunity extends beyond the hyperscalers. Applied Digital Corp., a Dallas-based data center operator, signed a 15-year take-or-pay contract in April for 300 megawatts of IT load with a U.S. investment-grade hyperscaler, adding about $7.5 billion in base-term contracted revenue. The company secured a $300 million senior secured bridge loan sponsored by Goldman Sachs to finance construction of its third AI data center in North Dakota.
For asset managers, the data center financing wave represents one of the largest deployment opportunities in a generation. Firms that can structure deals across the infrastructure-real estate continuum — and raise capital from both institutional and retail channels — stand to capture the widest share of fee income. The risk, as Fink noted, is that the concentrated bet on AI success means retirement portfolios are increasingly tied to the technology's performance.
This article is for informational purposes only and does not constitute investment advice.
