Source: Zhitong Finance.

JPMorgan released a research report on March 9th stating that generative AI in China has entered an application explosion phase. In this research report, JPMorgan established a four-phase framework for the development of generative AI in China and provided investment strategies for each phase.

The bank believes that the development of generative AI in China will impact the fundamentals and stock prices of the companies it covers for many years; from a 12-month investment perspective, it is expected that IaaS (Infrastructure as a Service) value chain will continue to outperform the Large Cap, as the development of generative AI in China has just entered the second phase of its framework, which will be accompanied by rapid growth in computing power. Moreover, the bank believes that high-potential beneficiaries of generative AI applications may also outperform the Large Cap, although they come with higher uncertainty and risk; it recommends investing in the companies it covers. $Alibaba (BABA.US)$ / $BABA-W (09988.HK)$ and $KUAISHOU-W (01024.HK)$ 。

1. The four stages of Gen AI development in China.

JPMorgan expects that the development of Gen AI in China will go through four distinct stages, each accompanied by significant changes in market dynamics. The launch of DeepSeek's latest V3 and R1 models has pushed the industry into the second stage, and the bank believes that 2025 will be a critical turning point for the adoption of Gen AI.

JPMorgan pointed out that the four stages are:

Stage One: Development of Large Language Models (LLM). The focus of this stage is to create and refine LLMs to lay the technological foundation for subsequent AI applications.

Stage Two: Adoption of Gen AI in existing applications and services. By 2025, the development of Gen AI in China has entered the second stage, characterized by the rapid integration of Gen AI features and functions into existing Internet applications and operations. This stage involves a trial-and-error process, as businesses are still unclear on how Gen AI will create value for their operations. Entering the second stage requires a combination of commercially viable model quality and cost-effectiveness. In the bank's view, DeepSeek's V3 and R1 models help to advance this stage.

Stage Three: Growth in Consumer Spending on Internet Services. In this stage, it is expected that consumer and business spending on Internet services will see significant growth. With the popularity of Gen AI, Internet operators may achieve considerable economic benefits, driving further growth in the Industry.

Stage Four: Emergence of Local Gen AI Killer Applications. In this stage, the rise of local Gen AI killer applications will reshape the competitive landscape of key market segments. These applications may drive new business models and revenue streams, creating opportunities for market leaders and challengers.

JPMorgan added that the corresponding investment strategies for each stage are:

Stage One: Leading LLM Developers.

Stage Two: IaaS Value Chain. The firm believes that the development of Gen AI in China has just entered the second stage, characterized by explosive growth in Gen AI consumption. Although the financial returns from Gen AI applications have not yet emerged, it is expected that Stocks in the IaaS value chain will outperform the market, benefiting from upward adjustments in revenue expectations.

Stage Three: Applications and service providers that gain practical economic benefits from the popularity of Gen AI. The firm stated that when Stage Three begins, investors are advised to shift funds from the IaaS value chain to applications and service providers capable of achieving practical economic benefits through the adoption of Gen AI, with positive surprises in revenue or profit expectations acting as catalysts for this rotation.

Stage Four: Local Gen AI Killer Application Developers.

In 2025, it will be a turning point for the adoption of Gen AI in China.

JPMorgan stated that the development of Gen AI in China in 2025 can be compared to the growth of the mobile Internet in 2010, as there are several key similarities between the two. The firm believes that by 2025, all Chinese Internet companies will recognize the importance of integrating Gen AI into existing business operations and will invest substantial resources to embrace Gen AI capabilities. This is similar to the strategy of the mobile Internet transformation in 2010. However, by 2025, the actual benefits of Gen AI to existing businesses have not yet been verified, and there have yet to be any killer applications for Gen AI in the local market, similar to the situation in 2010.

JPMorgan believes that 2025 will be a turning point for the adoption of Gen AI in China, triggered by the V3 and R1 models from DeepSeek. The firm indicated that these high-performance, cost-effective models will accelerate the integration of Gen AI with existing Internet applications and services in the coming quarters, significantly increasing the consumption of computing power. The firm asserts that the facts in the following table demonstrate that Gen AI in China has entered a phase of rapid adoption.

What will the Gen AI era look like?

JPMorgan expects the following trends to emerge in the coming years:

Improvement in financial performance: Gen AI will significantly enhance existing Internet services through new revenue streams and increased cost efficiencies, thereby improving the overall financial performance of Internet operators.

Change in competitive landscape: As Gen AI applications disrupt traditional business models, the market competitive landscape will undergo significant changes. Companies that adapt to and leverage Gen AI will gain a competitive advantage.

The emergence of killer applications for Gen AI: While predicting the exact timing and nature of killer applications for Gen AI is challenging, their emergence is expected to redefine market dynamics and create growth opportunities.

Increasing demand for computing power: To accommodate the growing demand for Gen AI computing power, large IaaS cloud operators will maintain high levels of capital expenditure in the coming years, benefiting downstream stakeholders such as datacenter and server providers.

4. Positioning of several large-cap Internet stocks in AI trade.

$Alibaba (BABA.US)$ / $BABA-W (09988.HK)$ JPMorgan believes that Alibaba is a Stock in the IaaS value chain, which will perform better than its Internet peers in the second phase, and may also become a beneficiary of Gen AI applications in the third phase.

$TENCENT (00700.HK)$ Although TENCENT has significant exposure in the IaaS market in China, JPMorgan views the stock more as a beneficiary of application in AI trading rather than as an IaaS provider. The firm expects TENCENT to actively integrate Gen AI into its existing product portfolio to enhance Consumer engagement in terms of traffic, time spent, and product dependency, which may convert into incremental advertising inventory and revenue in 2026 and beyond.

$KUAISHOU-W (01024.HK)$ JPMorgan stated that among the Chinese Internet companies it covers, KUAISHOU is a less conspicuous beneficiary of Gen AI development. The firm believes that Gen AI will significantly drive user engagement and monetization of KUAISHOU's main application; KUAISHOU's leading AI video generation large model, KLING, has a deep synergy with the KUAISHOU application in video generation, offering rich monetization potential across various business verticals (E-Commerce, professional content creation - such as gaming/movies).

$Baidu (BIDU.US)$ / $BIDU-SW (09888.HK)$ JPMorgan believes that Baidu is both an IaaS value chain stock and a potential beneficiary of Gen AI applications, with its stock price prospects depending on changes in the narrative of Baidu's core advertising business and cloud business. Among these factors, changes in the narrative of the core advertising business are more important, as this business generates the majority of the company's profits.

Editor/jayden