China Galaxy Securities: Inflation of domestic large model Token continues to focus on opportunities for low-level layout of Hong Kong Internet technology giants.

China Galaxy Securities released a research report stating that with the continuous improvement of domestic large model capabilities, the battle of the large models has entered the knockout stage, and may move towards token inflation. Due to the purity of its business, native large model companies have advantages compared to traditional internet giants and have been favored and sought after in the short term. However, the struggle for traffic entry by traditional internet giants in the AI era is a necessary option, as traffic remains one of the most important assets of the AI era. In the long term, internet companies are expected to regain their status in the AI era with their traffic advantages. Native large model factories may become an important part of their industrial chain. It is recommended to continue to focus on low-lying opportunities for Hong Kong internet technology giants. The main points of China Galaxy Securities are as follows: Domestic large models enter token inflation, the competition for AI traffic entry among internet technology giants intensifies In February, the Hang Seng Tech Index fell sharply by 10.15%, the largest monthly decline so far this year. As 7 of the top ten weighted stocks in the Hang Seng Tech Index are internet technology companies, traditional internet technology companies are currently facing a battle for AI traffic entry. The market is concerned about whether the short-term traffic narrative can bring about performance realization, leading to uncertainties in future profit expectations for internet technology companies. In addition, with the unexpectedly hawkish stance of the Federal Reserve, the market is worried about a turning point in global liquidity, which has resulted in a significant drop in Hong Kong internet companies, leading to the largest monthly decline in the Hang Seng Tech Index so far this year. However, looking at recent developments, native model companies listed in Hong Kong, KNOWLEDGE ATLAS (02513) and MINIMAX-WP (00100), have seen significant increases in share prices by 154.2% and 64.42% respectively. According to OpenRouter data, the volume of AI models called in China surged in February, surpassing the United States for the first time. The weekly rankings from February 16 to 22 show that four of the top five models by platform calls are from Chinese companies, namely MiniMax M2.5, Kimi K2.5, KNOWLEDGE ATLAS GLM-5, and DeepSeek V3.2, which together contribute to 85.7% of the total top 5 call volume. At the same time, KNOWLEDGE ATLAS announced an adjustment in the price of its GLM Coding Plan package, cancelling the acquisition discount and retaining the quarterly and annual subscription discount package prices for structural adjustments, with an overall price increase starting from 30% and no change for subscribed users. China Galaxy Securities believes that with the continuous improvement of domestic large model capabilities, the battle of the large models has entered the knockout stage and may move towards token inflation. Due to the purity of its business, native large model companies have an advantage over traditional internet giants. In the short term, they are favored and sought after for funding. However, the battle for traffic entry by traditional internet giants in the AI era is necessary. Traffic remains one of the most important assets of the AI era. In the long term, internet companies are expected to regain their position in the AI era with their traffic advantages. Native large model factories may become an important part of their industrial chain. It is recommended to continue focusing on the low-lying opportunities of Hong Kong internet technology giants. Agent intelligence capabilities leapfrog, Saas industry landscape may face a new paradigm AI startup Anthropic launched 11 new plugins in early February, including a legal plugin that can assist teams in reviewing contracts, identifying compliance risks, and tracking legal terms directly, equivalent to the capabilities of a junior lawyer. This has made investors begin to doubt whether the moat of traditional legal technology companies is still stable, as both domestic and foreign Saas companies have experienced varying degrees of decline. China Galaxy Securities believes that AI's biggest advantage currently lies in programming, which may impact traditional software companies in the development of functional interfaces. The traditional software sector may undergo a new reshuffling. It is recommended to focus on three types of companies: 1) companies with industry-specific expertise and proprietary data resources; 2) companies deeply embedded in physical systems, production processes, or high complexity scenarios and their corresponding scenes; 3) companies actively promoting software agentization. Risk warning Risk of technological iteration falling short of expectations; intensified competition among technology giants; legal regulatory risks; supply chain risks; risks of lower-than-expected downstream demand.