AI infrastructure investment sustainability faces scrutiny over debt dependency

The sustainability of the artificial intelligence (AI) investment supercycle is increasingly being scrutinized through the lens of debt dependency and cash flow. While major hyperscalers like Meta, Alphabet, Amazon, Microsoft, and Oracle continue to drive massive capital expenditures, concerns are rising regarding whether their operating cash flows can keep pace with the required investments in data centers and GPUs.

Meta Platforms, for instance, reported a significant gap between its operating cash flow and capital expenditures in the second quarter of 2026, with an annual CAPEX outlook reaching up to $145 billion. Gavin Baker, CIO of Atreides Management, noted that if AI infrastructure construction relies heavily on credit markets rather than internal cash flow, it could pose significant risks. The ability of cloud providers to reprice long-term GPU contracts will also be a critical factor in maintaining profitability.

Furthermore, the Bank of Korea has highlighted risks associated with the use of Special Purpose Vehicles (SPVs) and private credit funds to finance data centers, as these arrangements may not appear on corporate balance sheets, obscuring true financial risk. While global AI investment is projected to remain high—with Goldman Sachs estimating $5.3 trillion in cumulative investment from four major US tech firms between 2025 and 2030—analysts warn that a slowdown could occur if financial conditions tighten or if actual revenue generation fails to meet expectations.