Shenwan Hongyuan Group: How far has AI "bubble" gone?

Shenwan Hongyuan Group released a research report stating that since April, AI has been leading the way in the technology sector, with companies exceeding expectations and increasing capital expenditure. Compared to historical data, there may still be room for the expansion of AI capital. The group pointed out that while the Federal Reserve rate hike may hinder the process of AI "bubble," the probability of a rate hike within the year is still not high. Financing pressure is not yet a constraint, but power bottlenecks and public opinion may become obstacles. The performance of AI-related companies may be the focus in the short term. Shenwan Hongyuan Group's main points are as follows: Hot Topic Analysis: How far has the AI "bubble" gone? 1. How is the AI revolution progressing? AI applications, employment substitution, and productivity enhancement are still in the early stages. Since April, AI has been leading the way in the technology sector, with companies exceeding expectations and increasing capital expenditure. In the first quarter of 2026, "hard technology" companies such as chip designers and manufacturers saw a high growth in profits, corresponding to their stock performance since April. In the first quarter, M7 capital expenditure increased by 65% year-on-year, accounting for 33% of the S&P 500, and capital expenditure guidance continued to rise. AI not only drives stock market prosperity, but its importance to the U.S. economy and global trade is also increasing. In the first quarter of 2026, AI investments boosted the U.S. economic growth rate by 1.1 percentage points, surpassing consumer spending and traditional investments. Strong AI investments in the U.S. also led to a significant increase in import demand, reflected in the strengthening global manufacturing PMI and global trade resilience. Besides investment, AI applications, employment substitution, and productivity enhancement are still in the early stages. In May 2026, the AI application rate in U.S. companies was only at 19.8%, significantly lower than the theoretical level in various industries. AI's impact on U.S. productivity is still comparable to the early stages of the internet revolution, and its impact on U.S. employment is still limited, focusing on high-exposure industries, young demographics, and high-paying positions. 2. How far has the AI bubble gone? There is still room for increased investment intensity and external financing, with solid financial indicators. Compared to historical data, there may still be room for the expansion of AI capital. Since the fourth quarter of 2022, U.S. AI investments as a percentage of GDP have cumulatively increased by 1.0%, indicating room for growth compared to the internet revolution (1.4%). The capital expenditure of top tech companies is approaching their operating cash flow levels, but the Nasdaq 100 capital expenditure/operating cash flow ratio is still lower than it was in the early 2000s. Although top U.S. tech companies have started moving towards external financing, concerns about financing pressure are premature. Measured by net debt/EBITDA, the debt pressure on top U.S. tech companies is significantly low; in the first quarter of 2000, the total GDP share of U.S. corporate bonds, loans, and equity financing was as high as 5.4%, but by the end of 2025, this level had dropped to 1.6%, showing no signs of excessive financing. The financial indicators of top U.S. tech companies in the stock market are robust, and valuations still match performance. During the internet revolution period, the market value of top U.S. tech companies far exceeded profits, but in this round of tech company growth, market value gains still correlate with profits. The financial indicators of M7 companies such as cash/market cap, ROE, net profit margin, etc., are stronger than those of top companies in the dot-com bubble era, indicating that AI investments are built on a stable financial foundation. 3. How sustainable is AI investment? Financing pressure is not yet a constraint, focus on the realization of AI performance While the Federal Reserve rate hike may hinder the process of AI "bubble," the probability of a rate hike within the year is still not high. Under a "weak balance" in employment, the probability of a rate hike by the Federal Reserve within the year is low; if economic impact from oil prices emerges, expectations of rate cuts may rise; the ratio of Nasdaq 100 corporate bond issuance/ capital expenditure is still below 40%, significantly lower than the peak of the past 10 years, indicating that corporate reliance on debt financing is still low. When will the growth rate of U.S. AI investment peak? Financing pressure is not yet a constraint, but power bottlenecks and public opinion may become obstacles. The capital expenditure growth rate of M7 in 2026 may exceed 60%, and attention should be paid to whether this high growth can be maintained in 2027; obstacles that AI investments may face include power grid bottlenecks, equipment shortages, and opposition from residents to data center construction, which may lead to project delays or even cancellations. The performance of AI-related companies may be the focus in the short term. Since 2023, the degree to which major tech companies surpass profit expectations and safe margins has narrowed, possibly leading to market questioning the rationality of capital expenditure in the future; the core observation indicators of the demand side of AI investment should focus on the performance of top tech companies, especially companies like OpenAI and Anthropic. Risk Warning Oil price centralizes move higher than expected; hawkish policy stance by the Federal Reserve; U.S. economic slowdown exceeds expectations.