Besides the bubble, a big AI risk could be growing

A sharp sell-off in US technology stocks on Friday highlighted investor nervousness about the massive AI spending by Big Tech companies. The tech-heavy Nasdaq fell 2.86%, its largest single-day decline since June, following disappointing results from Intel and Amazon, which deepened concerns over whether quick investments can yield returns.

While the primary worry about AI remains whether the billions poured into data centers, chips, and infrastructure will generate returns fast enough, a potential secondary financial risk is emerging.

As AI infrastructure expands, a complex web of financing arrangements is connecting hyperscalers, chipmakers, data center operators, and AI startups. These arrangements could create financial exposures that are not immediately visible to investors. Although not an immediate threat, these connections could become a more serious vulnerability if the AI investment boom slows down.

The recent technology stock decline has raised questions among investors about whether the expected returns from Big Tech's AI spending can justify the extraordinary capital expenditure. Credit-rating agencies are also increasingly scrutinizing the risk. Mark Habib, S&P Global's head of digital infrastructure, noted that the biggest risk to hyperscalers' credit ratings is whether expected profits from AI investments fail to materialize.

S&P projects that six major US hyperscalers, including Amazon, Microsoft, Alphabet, Meta, Oracle, and SpaceX, will spend more than $7 trillion on data centers and AI-related infrastructure between 2025 and 2030, expecting these companies to experience negative free cash flow until at least the following year. However, Habib also pointed to the possibility that hyperscalers may have more AI-linked debt and financial liabilities than currently visible, as they often support large data center projects or smaller AI companies through interconnected financing structures.

These financing arrangements, such as debt, leases, joint ventures, special-purpose vehicles, and residual-value guarantees, are becoming more intricate than traditional corporate borrowing. For example, Google is both an investor in AI company Anthropic and a major provider of its cloud computing capacity. Another example involves Google and Anthropic, where Google purchases Google's custom AI chips from Broadcom and leases them to Anthropic.

These complex structures can make it harder to determine the ultimate financial exposure, particularly when companies are simultaneously investors, suppliers, customers, and financial backers within the same ecosystem.

While these interconnected financing arrangements are not inherently problematic, they can create questions about how much reported demand is backed by independent customers rather than financing relationships within the same ecosystem. This issue becomes even more complex when investments, commercial agreements, and financing arrangements reinforce one another. LSEG's analysis suggests that AI financing should be viewed as a web of interconnected relationships rather than individual transactions.