The artificial intelligence boom is creating a growing financing challenge for the companies building its infrastructure, while power shortages, political opposition and rising hardware costs threaten to complicate further expansion.
The investment firm estimates that 80% of AI-related capital spending over the next four to five years will require external financing, as major cloud and technology companies spend beyond what their operating cash flows can cover, Carlyle Group (NASDAQ:CG) said in its Global Private Markets Quarterly report.
The funding needs are already reshaping credit markets. High-yield companies have issued $39 billion in bonds this year to finance expansion and capital expenditures — more than the total issued for those purposes over the previous 20 years. Spreads on investment-grade bonds issued by hyperscalers have also widened by roughly 40 basis points on average.
Power and Politics Threaten AI Expansion
Carlyle identified financing, political opposition and electricity supply as three major constraints on the AI infrastructure buildout.
New grid connections can take three to seven years to come online, while demand is outpacing supply. In Texas, the Electric Reliability Council of Texas, or ERCOT, faces more than 143 gigawatts of data center requests, compared with current peak electricity demand of roughly 86 gigawatts.
Political resistance is adding another obstacle. At least 25 states have restrictions or moratoriums when local pauses are included, Carlyle said, making data center development an increasingly prominent issue ahead of the midterm elections.
"While we see no sign of a slowdown in AI capex, we expect enterprise IT spending growth may continue to moderate as companies continue to adopt policies to optimize token consumption," the report said.
That could pressure AI model developers already facing hardware prices that have climbed 140% over the past 12 months, alongside competition from lower-cost open-weight models. Carlyle added that this is an area it will continue to watch closely.
Investors Shift Toward Physical Assets
The AI boom is also changing private equity dealmaking. Technology companies accounted for just 8% of leveraged buyout deal volume in the third quarter, down from 45% a year earlier. Hard-asset and manufacturing industries, meanwhile, represented 52% of deal volume, up from 20%.
Carlyle said the shift toward the "real economy" is partly tied to demand for data centers, power and other assets supporting AI infrastructure.
The report cited Brookfield Asset Management’s (NYSE:BAM) $7 billion acquisition of Aypa Power from Blackstone (NYSE:BK), including debt, as a major private-market transaction. It also highlighted MGX Fund I, a UAE-backed AI investment vehicle that closed at $49 billion in the third quarter, helping drive a sharp increase in growth-capital fundraising.
AI’s Returns Remain Uncertain
Despite the spending boom, Carlyle warned that investors may underestimate how exposed their portfolios are to a single theme.
"The importance of understanding business fundamentals is difficult to overstate," the firm said, cautioning that portfolios that appear diversified may still depend heavily on AI.
The report said the technology’s potential outcomes remain difficult to quantify, leaving investors to weigh the prospect of major productivity gains against uncertainty over "if, how, and when the promises of the AI revolution will be realized."
Photo by Shutterstock