On November 10, Shenwan Hongyuan Group Co.,Ltd. (000166) and SWHY (06806) hosted its 2027 Capital Markets Investment Annual Meeting in Shanghai. Chen Dafei, Chief Macro Analyst at Shenwan Hongyuan Research, delivered a presentation titled "Bubble Revaluation" outlining the company's 2027 overseas macroeconomic outlook.
He stated that the growth rate of US data center investment may peak around the fourth quarter of 2026, slightly lagging behind market expectations. Based on the data center construction cycle and micro-level project data, a real-time forecasting framework for US AI capital expenditure has been constructed.
As of the first half of 2026, real-time forecasts looking forward one year indicate that US data center investment growth may peak in the fourth quarter of 2026, lagging market expectations by one quarter. Monetary tightening is a necessary but not sufficient condition for the revaluation of financial bubbles; the bubble itself is the sufficient and necessary condition for its revaluation.
Financial instability arises from excessive leverage in the real economy sector. From 2021 to the present, the intensity of US non-residential investment has increased by only about 1.5 percentage points, while corporate leverage and debt service ratios remain at historically low levels. AI capital expenditure and corporate leverage behavior may not yet have entered a frenzied stage.
The debt servicing capacity of AI companies remains robust and is still improving at the margin, providing a higher margin of safety compared to the dot-com bubble period of 2000. Among the full sample of enterprises, hedge financing, speculative financing, and Ponzi financing account for 72.2%, 12.4%, and 15.3% respectively, with hedge financing still dominating.
In recent years, the proportion of Ponzi financing enterprises has declined, significantly lower than during the 2000 dot-com bubble period. From the peak of asset prices to the peak of capital expenditure growth, this marks the transition of the economic and credit cycle from positive feedback to negative feedback.
The AI "bubble" has not yet evolved to an uncontrollable stage, and financial risks may still remain non-systemic. However, experience shows that general-purpose technology revolutions and financial bubbles are a twin phenomenon, and once a bubble develops to a certain extent, it will ultimately move toward revaluation.