The rapid expansion of artificial intelligence investment faces several potential fault lines, including rising debt-financed spending, circular financing arrangements and the emergence of cheaper open-weight models, according to 360 ONE Asset’s August 2026 Panorama report.
While the report has not declared that the AI boom is a bubble but has highlighted vulnerabilities on both the demand and supply sides of the value chain that could trigger an unwind.
Debt and circular financing
AI infrastructure spending by US hyperscalers and other AI companies continues to grow at a strong pace and is increasingly being financed through debt, the report said and that this could transmit vulnerabilities in the AI buildout into credit markets.
The risks are amplified by circular financing arrangements, which account for a sizeable share of AI labs’ spending and hyperscalers’ forward revenue.
One potential supply-side trigger is a breakdown in circular financing. If suppliers stop funding buyers, sales could fall, creating pressure across the AI investment chain.
The report also flags private-credit risks, warning that collateral could be worth substantially less than it was when loans were underwritten.
Cheaper AI models could pressure spending
Another potential source of disruption is the shift towards cheaper open-weight AI models.
The report said Chinese AI models are narrowing the performance gap with US models and have reached comparable capabilities with a lag of only about six months. Chinese models also lead in cost-effectiveness, operating at a fraction of the cost of leading Western frontier models.
The shift towards cheaper open-weight models is already contributing to lower AI costs. The blended price paid per million tokens has fallen by roughly 45% since May 2026, partly as usage has shifted from frontier models to cheaper open-weight models.
The report identifies a shift to open-weight models as one possible way the AI cycle could start to unwind, as cheaper models could “de-rate” US frontier labs.
What happens if AI spending disappoints?
On the demand side, AI adoption could slow if productivity gains fail to justify the scale of spending. A broader macroeconomic slowdown, including higher interest rates or job losses, could also weaken corporate spending.
On the supply side, infrastructure could become overbuilt if companies expand capacity faster than demand while competing for market share. Technological advances could also reduce compute requirements, leaving existing chip and infrastructure investments vulnerable.
Regulatory tightening following AI-related incidents could further slow adoption.
The report stresses that these risks are not isolated. “The two sides feed each other as each shock can amplify the other”, it mentioned.
Global spillovers, but India is better insulated
Drawing a comparison between the current sharp rally in semiconductor stocks and the late-cycle rally of the 1990s, when gains became concentrated in hardware even as software stocks failed to rally, the report mentioned that a reversal of the AI investment cycle could have broader financial-market consequences.
Taiwan and South Korea could be particularly exposed because their equity markets and economic growth have benefited from the AI-led investment cycle. A reversal of the AI trade could trigger financial contagion and weigh on both markets and broader economic growth.
India, however, is relatively less exposed because its equity market is more diversified. The top 10 Indian companies account for 19% of total market capitalisation, compared with 33% for the top 10 US companies, 39% for the top company in Taiwan and 47% for the top two companies in South Korea.
India is not completely insulated though as the country has seen a surge in data-centre investment proposals, but a pullback in global AI capital expenditure could slow the buildout and result in more muted growth than currently anticipated.
Leave a Reply