On June 2nd, Kong Rong, Deputy General Manager of the Research Institute and Chief Overseas Analyst at Guolian Minsheng Securities, provided an in-depth analysis during an online broadcast, addressing market sentiment of 'fear amidst rising prices', the diverging landscape of global tech stocks, and key investment themes for the second half of the year. She pointed out that AI commercialization has exhibited non-linear revenue growth, and that the tech sector still presents allocation opportunities following adjustments. The epic surge in the South Korean stock market is driven by a memory super-cycle, while the underperformance of the Hang Seng Tech Index relative to global peers is primarily due to the 'AI narrative' for heavyweight stocks like Tencent and Alibaba failing to impress the market. However, she noted that the overall valuation of Hong Kong stocks is 'extremely cheap', suggesting considerable room for subsequent recovery.

Non-Linear AI Commercialization Emerges, Post-Adjustment Opportunities Remain

Faced with market anxiety over 'fear amidst rising prices', Kong Rong offered a clear assessment. She stated that while the technology sector will experience a degree of adjustment, she remains optimistic about the subsequent market trend. The fundamental reason lies in the substantive progress of AI commercialization. Citing Anthropic and OpenAI as examples, she noted that these companies' annual recurring revenue is growing rapidly and continuously, demonstrating that AI has already shown strong commercial viability, which will continue to support the tech rally ahead.

Kong Rong believes that over the past three years, the market was largely in a phase of 'anticipation', whereas this year investors are finally witnessing AI transition from an investment phase to an output phase. "The biggest difference is moving from the three-year anticipation period to seeing AI commercialization demonstrate strong capabilities this year." She defines the current stage as the 'golden time for global technology', emphasizing that opportunities for allocation persist even after adjustments.

Regarding commercial revenue, Kong Rong presented striking data: taking Anthropic as an example, its annualized revenue jumped from around $9 billion at the end of last year to over $10 billion in January, and to nearly $50 billion by April, with monthly leaps of tens of billions of dollars—a phenomenon unseen in recent years. Similarly, following the launch of OpenAI's Codex product, the growth rates of API calls and user revenue have been astonishing.

Trillion-Dollar IPOs Reshape Tech Stock Valuation Logic, AI Investment Shifts from Hardware to Software

The successive IPO plans of three trillion-dollar tech giants—SpaceX, OpenAI, and Anthropic—have sparked widespread discussion about capital siphoning and valuation restructuring. Kong Rong pointed out that the market's enthusiasm for these giants essentially reflects a search for certainty amidst uncertainty, and more capital will flow towards the most certain directions and targets in the future.

She also emphasized that the IPO logic for SpaceX and AI model companies is not entirely identical. SpaceX represents a 'new narrative', and the market will price it based on its ultimate growth potential. In contrast, the listings of OpenAI and Anthropic will directly impact the current valuation anchor for the AI industry.

Kong Rong believes the deeper impact of large model company listings on tech stock pricing lies in breaking the past three-year trading theme of 'hardware dominance'. The mainstream narrative over the past three years was that large models would consume all applications, leading to significant drops for software companies and gains for hardware companies. However, with model companies going public, capital will be reallocated—partly flowing to the model companies themselves, and partly seeking out software companies that are not consumed by AI but instead see increased demand due to AI. She specifically mentioned that the recent earnings reports of data infrastructure software companies like Snowflake and MongoDB, which significantly exceeded expectations, leading to notable stock price rebounds, validate this logic.

Against the backdrop of continuously improving Agent capabilities, the business demand for companies in data warehousing, data security, and identity authentication is not being consumed but is actually increasing. These companies possess data moats and barriers in vertical scenarios, which is key to becoming 'AI winners'.

Chinese Large Model Companies Hold Revaluation Potential via 'Token Globalization', Memory Enters Super-Cycle

Regarding Chinese domestic tech assets, Kong Rong provided a systematic valuation analysis. She noted that after their listings, Zhipu and Minimax have shown strong stock performance, while some overseas large model companies have faced valuation pressure during the same period. This divergence stems from both trading factors and deeper differences in valuation logic.

On the trading front, the relatively small free float of Zhipu and Minimax means that less capital can drive their stock performance. However, the more critical factor is the core logic—overseas large model company valuations primarily focus on commercial Annual Recurring Revenue (ARR), future growth rates, and are assigned certain Price-to-Sales (PS) multiples. For Chinese large model companies, there is an additional, very important logic: these companies are essentially global from birth, having already 'gone global'.

This logic is underpinned by the concept of 'Token Globalization': Chinese AI companies generating revenue in markets like the US, Southeast Asia, and the Middle East. Their potential market space extends far beyond the domestic market. Compared to their overseas peers, their revenue and market capitalization might be only one-tenth or even one-twentieth, which itself represents a huge growth gap.

Kong Rong believes that as more large model companies like Kimi and DeepSeek enter the capital markets, 'Token Globalization' will become one of the core narratives for China's AI sector.

In the memory sector, Kong Rong clearly stated that driven by AI, the memory industry has entered a 'super-cycle'.

Kong Rong believes that since the surge in AI demand in the second half of last year, memory company stock prices, both overseas and domestically, have performed very strongly this year. The fundamental reason is that the iterative improvement of AI model capabilities will not stop, and data storage demand will only grow larger. This is not a demand logic that can be broken by a single round of short-term capacity expansion.

She specifically mentioned that global semiconductor capacity is ultimately constrained by HBM (High Bandwidth Memory). Chinese memory companies—ChangXin Memory Technologies (CXMT) and Yangtze Memory Technologies Co. (YMTC)—have demonstrated very strong competitive advantages in global competition. Their listings are significant landmark events for the A-share market this year.

South Korean Stocks Driven by Memory, Hang Seng Tech Divergence Due to Weak 'AI Narrative' of Heavyweights

Against the backdrop of diverging global tech stock trends, Kong Rong provided a comparative interpretation of the performance of the South Korean and Hong Kong stock markets. Regarding the epic surge in South Korean stocks, Kong Rong believes its core driver is highly consistent with the global tech theme: the resonance of AI and hardware.

Although the new South Korean government has provided policy support for the capital market, the more crucial driving force comes from the memory industry. Samsung and SK Hynix have driven the performance of the entire South Korean market. Whether the opportunity in the memory sector can continue will directly determine the subsequent trend of South Korean stocks.

However, in stark contrast to the fervor in South Korean and US tech stocks, the Hang Seng Tech Index faced pressure in May. Kong Rong pointed out that the key reason is the temporary failure of the AI narrative for heavyweight stocks like Tencent and Alibaba to impress the market. Current market concerns focus on their resource allocation not being sufficiently concentrated on AI, and their model capabilities not yet dazzling the market. Meanwhile, the strong post-listing performance of AI-native new companies like Zhipu and Minimax has created a 'challenger' narrative against traditional giants.

But she drew an analogy with Google's experience. Before Google launched Gemini 2.5, the market once believed that despite sitting on a technological treasure trove, Google could not produce a leading model, leading to prolonged stock price weakness. It was only with the strong debut of Gemini 2.5 that the valuation logic reversed in an instant. She believes that if Tencent, Alibaba, and other Hang Seng Tech heavyweights can deliver large model products that convince the market, a similar 'Google moment' could occur. It is worth noting that the weakness of the Hang Seng Tech Index does not obscure the structural bright spots within the Hong Kong market. Hardware stocks like Lenovo have performed strongly due to the AI PC concept, and newly listed Hong Kong chip companies and model companies have also shown impressive stock performance this year.

Hong Kong Stocks Overall 'Extremely Cheap', Two New Tech Directions to Watch in H2

Regarding cross-market allocation, Kong Rong emphasized that these Hong Kong-listed companies are 'extremely cheap'. Although some new tech companies in Hong Kong have risen significantly this year, many large platform companies have underperformed, and their valuations remain reasonable. Following the stock price and market performance of overseas counterparts, Hong Kong stocks have greater potential for subsequent movement.

For seizing global allocation opportunities, she suggested adhering to one principle: invest in the best companies in the field. While overseas companies are indeed leading, Chinese companies also have their own areas of expertise and advantages.

Beyond the AI theme, Kong Rong highlighted two entirely new technological directions. The first is the space economy. SpaceX presents a new opportunity. While many might think space sounds like science fiction or even 'bubble creation', careful research reveals it has important commercial logic and is achievable in the future. According to SpaceX's prospectus, this is a multi-trillion-dollar sector. The second is Physical AI/World Models, which digitize the real world for simulation and testing, applied in scenarios like autonomous driving and robotics. This is a very important direction following large language models.

Addressing the widespread 'fear of missing out' anxiety among investors, Kong Rong advised adjusting mindset and controlling risk. She emphasized that ultimately, profitability depends on risk control capability—those who last the longest are the final winners. Investors should establish their own judgment framework based on the direction of global industry evolution, rather than being swayed by short-term emotions.