AI Infrastructure: Bubble or “Profound Change”? – Jan. 5, 2026
With trillions of dollars of data center projects on the drawing board, skeptics are asking what happens if demand falters
Reports over the holidays from the Wall Street Journal and CNBC underscored the mushrooming growth of the data center sector, driven largely by the expanding prevalence of artificial intelligence. Both reports cited the increasing risks accompanying the increases in demand, development and investment.
The WSJ reported that spending on data center development is poised to surpass office-building construction for the first time, possibly as soon as this year. North America alone could see $1 trillion of new data center projects between 2025 and 2030, according to JLL. Globally, HSBC projects an AI infrastructure surge worth twice that amount.
At the center of the development—and investment—is a handful of hyperscalers. In the space of two months in late 2025, OpenAI announced partnerships totaling roughly $1.4 trillion in headline commitments, CNBC reported. The rapid succession of deals has led skeptics “to warn of an AI bubble and raised basic questions about whether the power, land and supply chains exist to match the ambition,” according to CNBC.
Among OpenAI’s partners in these projects is Oracle, which is a front-runner in its own right by one metric: its data center-related leasing. Although it’s also spending billions to develop data centers, the company is carrying $248 billion in future leasing commitments not yet on its balance sheet. Investor concerns about potential overspending on AI infrastructure brought Oracle share prices down 11% last month, reported the WSJ.
Other partners in OpenAI’s deals include Broadcom, Nvidia and AMD. “Nvidia is effectively financing demand for its own chips, Oracle is building the sites, AMD and Broadcom are positioning as alternative suppliers and OpenAI is anchoring the demand,” wrote CNBC’s MacKenzie Sigalos.
“Critics call it a circular economy: capital, capacity and revenue all recycling through the same small set of players. It works as long as growth holds — but if demand slips or funding tightens, the stress can propagate fast through a web of shared exposures.”
Increasingly, that web of exposure extends to real estate investors. Citing data from Nareit, the WSJ reported that publicly traded real estate companies boosted their data center investments by 15% in 2024, while pulling back from traditional sectors such as office buildings and apartments. And a CBRE survey of 92 major investors, including private equity and pension funds, found that 95% planned to increase their investments into data centers.
One major distinction between office or multifamily properties and AI-centric data centers is that both office space and apartments can be home to a diversified mix of tenants, the WSJ pointed out. AI, by comparison, is a niche tenant base, one that has yet to find a profitable business model.
Ryan Severino, chief economist at BGO, sounded a cautious note regarding exposure to AI infrastructure. “It’s a very specific use case for a data center versus those other property types,” Severino told the WSJ. “I’m not sure everyone out there playing in this space understands it as well as they should.”
Asset management giants Brookfield and Blackstone have significantly increased their exposure to these properties in recent years, even if data centers remain a relatively small part of their overall investment portfolios, the WSJ reported. In an interview with CNBC last month, Blackstone CEO Stephen Schwarzman downplayed fears of an AI infrastructure bubble.
“What’s happening is that we’ve got rapid expansion,” Schwarzman said. “It’s being done principally by five or six companies,” nearly all of whom are “great credits. What we do is quite simple in a way: we’re the largest in the world at building data centers and we’re the largest owner of data centers. But it’s a very conservative, picks-and-shovels type of business… This isn’t bubble-type work.”
He acknowledged that with the “explosion” of growth, there will be newcomers to the sector with credit that is less robust than that of tenants signing 15- or 20-year leases at Blackstone’s facilities. “And so you’ll have some problems, like you always do with this type of thing,” said Schwarzman. “But we will power through this. And there’s going to be a profound change in the world, looked at five to 10 years from now.”
For insights into the drivers of the profound change Schwarzman sees coming, be sure to attend Connect North American Investment in Digital Infrastructure & AI, scheduled for Feb. 11 in Montreal.