UBS: The de-leveraging process of AI is nearing the end, and Chinese tech stocks may welcome a new opportunity for reallocation.

UBS China Equity Strategy Research Head Wang Zhonghao pointed out in an article that certain AI technology hardware stocks tracked by UBS's Investment Research Department have seen their prices drop by 32% in July, with 36% of the tracked stocks experiencing declines of 40% or more during the month. Meanwhile, the financing balance of A-shares has retreated from a high of 3 trillion RMB to 2.6 trillion RMB, approaching levels seen before the recent increase in leverage in early April. Therefore, it appears that the most severe phase of technical sell-offs/deleveraging has passed. At the same time, driven by the strong performance of major cloud service providers in the U.S., the global narrative around AI seems to have turned positive. With valuations of Chinese AI technology hardware stocks significantly dropping to just above historical averages and earnings per share forecasts continuously being revised upward, the bank believes it is now time to re-enter the market. That said, the bank feels that, given the recent price volatility and ongoing market concerns regarding AI monetization capabilities, investors may not embrace AI technology hardware as aggressively as before. Therefore, the bank anticipates that the breadth of stock performance in the future may be more dispersed than it has been this year, and some funds may flow back into previously pressured sectors, such as Chinese internet, power equipment, and non-ferrous metals. As the financing balance has dropped back to April levels, and many tech stocks' prices have also retreated to nearly April levels, this indicates that the market may have passed the most severe phase of technical sell-offs. Overall, the collateral ratio for A-shares still appears healthy, at 280% for margin financing, suggesting that if another technical sell-off occurs, the market will still have some downside protection. The recent support for technology ETFs from the Hong Kong government also helps to stabilize confidence, and a stabilization in stock prices may attract some fundamental investors back to the market. Communication with investors by the bank indicates that, given the advancement of semiconductor localization, Chinese technology remains a focus for global investors. From a global perspective, leverage in tech stocks also seems to have declined, with Korea's margin loan balance currently close to early 2026 levels, and the assets under management of leveraged ETFs in Korea down about 50% and about 30% in the U.S. The bank stated that views on the global AI market have improved, primarily reflected in the following aspects: (1) Microsoft's and Amazon's performance shows that OpenAI has strong growth, and the monetization capability of AI investments has improved; (2) strong demand from AI-native companies; (3) increased order backlog from large cloud service providers, indicating an improved growth outlook for cloud service vendors; (4) an accelerated pace of AI adoption by enterprises, with average AI spending increasing by 25% quarter-on-quarter. For many Chinese AI stocks, their prices have been dragged down by the global sell-off. However, throughout this process, their narrative and fundamentals have not changed: the domestic technology supply chain in China is catching up, and with improvements in the supply of domestic GPUs, the bank expects AIDC (AI Data Center) construction to accelerate in the second half of this year. The stabilization of the global AI narrative helps investors refocus on the fundamentals of Chinese AI supply chain stocks. UBS believes that the breadth of the market in the second half of the year may no longer be as narrow. While the bank remains optimistic about the fundamentals of technology, recent stock price volatility and ongoing uncertainties in the market regarding AI monetization capabilities may limit the stock performance of relevant companies in the short term. Therefore, the bank believes that inter-industry performance in the second half of the year may not be as concentrated as in the first half. Within the AI technology hardware sector, the bank's preferences still focus on semiconductor equipment, network chips, and advanced packaging, which are the lower-cyclical parts of the AI theme. Outside of tech hardware, the bank's preferred sectors include: (1) internet, benefiting from improving profit trends, a shift in AI narratives related to large cloud service providers, and cheap valuations; (2) power equipment, due to cheap valuations and benefiting from AI data center construction and energy independence; (3) non-ferrous metals, due to strong profit trends; (4) "overseas" stocks, as the impact of currency appreciation begins to fade.