UBS: Improving investor sentiment boosts valuation of Chinese internet industry, but stock prices remain relatively cheap.

UBS Investment Bank's Director of Research for the Chinese Internet industry, Fang Jinchong, said that this year, the overall favorable market sentiment in the Chinese Internet industry has driven valuation multiples higher. The emergence of DeepSeek has improved investor sentiment towards this sector. Chinese stocks started to catch up this year. From a valuation perspective, the valuation multiples of major Internet companies have expanded by about 58%, to an adjusted P/E ratio of about 17 times, corresponding to a compound annual growth rate of about 10% in adjusted earnings per share (EPS) forecast for 2024-2026. In comparison, the valuation of the "Big Tech Seven" in the US is about 31 times, corresponding to an 18% EPS compound growth rate. It is clear that Chinese Internet stocks are relatively inexpensive. Fang Jinchong pointed out two major trends in the Chinese Internet industry. The first trend is that Chinese Internet giants are actively increasing capital expenditures and increasing investment in artificial intelligence (AI). Geopolitical uncertainties have made investors concerned about whether chip supply will affect the development of AI in China, but capital expenditures are more demand-driven. Compared to overseas, Chinese Internet companies are more focused on graphic processing unit (GPU) efficiency and utilization, and therefore can adjust investment targets more quickly and flexibly based on changing demands. At the same time, domestic computing power continues to develop. At the AI chip level, although there is still a performance gap, performance is rapidly improving due to the continuous self-research investment by Chinese Internet companies and the development of domestic GPU manufacturers. At the system level, positive progress is also seen, especially through the adoption of "super node" technology. "Super node" technology expands the number of GPUs in a single server cabinet, to some extent making up for the gap in domestic single GPUs, and achieving better computing performance at the cabinet level. At the level of large model algorithms, domestic developers are optimizing algorithms for domestic GPUs. Chinese cloud companies continue to steadfastly invest in AI. The second-quarter results show that leading companies generally maintain their full-year capital expenditure guidance, focusing on improving chip utilization and deployment efficiency. Faced with supply chain uncertainties, they emphasize flexible chip selection, including domestic alternatives. He said that when considering AI-relevant aspects that can drive revenue, the main focus is on the following businesses. Revenue from cloud-related businesses is accelerating, and large language models (LLMs) are expected to become a sustainable drive for cloud market growth, bringing opportunities for cross-selling traditional businesses and building long-term ecological potential. Chinese cloud service providers (also major suppliers of large models) bundle large model APIs with traditional infrastructure (IaaS) and platform as a service (PaaS) for cross-selling, thereby driving revenue growth and profitability. Industry data shows that as of the first half of 2025, the demand for cloud services related to generative AI has grown nearly tenfold compared to the first half of 2023. However, it is important to note the short-term profit margin pressure brought about by AI investments. Integrating existing businesses deeply, such as including e-commerce, movie recommendations, games, and education. For e-commerce and movie suppliers, the opportunity lies in achieving precise push through user behavior prediction; in terms of games, AI is applied to art design, programming design, etc., to reduce costs, while AI can create smarter virtual partners and NPCs with dynamic interactive behavior, enhancing the gaming experience. In addition, independent applications are rapidly emerging, including in content generation, AI search, information processing, and conversational assistants. This aspect requires attention to repetitive model operations, large research and development investments, and intense price competition. The second major trend is the investment in instant retail. Increasing instant retail is a platform's hope to drive low-frequency e-commerce businesses through high-frequency takeaway transactions and increase user activity. Currently, there are signs of short-term competition stabilization, with industry transaction volume growth slowing down, primarily due to seasonal declines in beverage orders and platform optimization of promotional strategies following regulatory talks. In addition, by using the percentage of delivery app usage time as a proxy indicator for the platform's share of orders, market share tends to stabilize. In the medium to long term, the industry still faces some challenges, with structural changes not yet reflected in prices. Firstly, competition is intensifying; secondly, how to accelerate the online penetration of takeaways in terms of merchant supply and user mindset. From the current investment effects, the number of orders is increasing, and users are gradually starting to cultivate the mindset of instant retail, accelerating the process of offline to online penetration. However, the medium to long-term impact on merchant profits etc. still needs to be observed.