Huachuang Securities looks forward to the 25-year opportunity of the Internet: optimistic about the value reappraisal of Alibaba Cloud (09988) and the application ecosystem reshaping of Tencent (00700).

Huachuang Securities released a research report stating that if the industrial reshaping brought by open source large models is similar to the United States in 2023, the first enlightenment it may bring is that cloud revenue, subscription revenue, and advertising revenue are the first areas closest to implementation in the initial stage. Understanding the business model of M7 can help identify the most worthwhile investment direction in this round. The second enlightenment may be that the investment value of configuration-type weight stocks this year will be strengthened, different from the completely dominant logic of the previous two years; internet giants Alibaba (09988) and Tencent (00700) may show similar configuration value. Key points of Huachuang Securities: Review: Since 2023, generative AI has ignited a new round of bull market in US tech stocks, with cumulative gains of 24% and 43% for the S&P 500 and Nasdaq indices in 23 years, respectively. In this round of tech bull market, seven leading tech companies play a crucial role. Their cumulative gains in the past 23 years are as follows: NVIDIA/Meta/Tesla/Amazon/Microsoft/Google/Apple with gains of 239%/194%/102%/81%/58%/58%/49%, far exceeding the index gains. In terms of market share, based on the closing prices of the past 23 years, their market value sums accounted for ~55% of the Nasdaq 100 index weight, and also accounted for about 27% of the S&P 500 index market value, significantly affecting the overall trend of the industry index. What sparked the M7 bull market? Leading model capabilities Increased computing power demand Explosive downstream applications and AI demand Giant companies - Rapid surge in computing power demand: Overseas AI companies led by OpenAI in 2023 maintain a significant advantage in the large model field. On one hand, the advantages of large models themselves have led to a rapid increase in call frequency and inference computing power demand; on the other hand, large models are constantly iterating and upgrading, directly driving the demand for GPU servers, high-speed networks, and data centers. The huge demand for AI training and inference computing power has laid the foundation for the leading chip manufacturers (NVIDIA) and cloud providers (Amazon, Microsoft, Google, etc.) to become "power sellers", further externalizing the capital expenditure and revenue growth on the computing power side, ushering in a new era of AI infrastructure. Downstream - Prosperity of diverse AI applications: AI+ advertising marketing (Applovin), AI+ e-commerce (Shopify), AI+ education (DUOLINGO), AI SaaS (Salesforce), etc. are leading the way in realizing performance. Giants - Hardware + software super entry point: 1) Leading internet companies represented by Meta (social) and Amazon (e-commerce) have stable and rich application ecosystem and massive user data, as well as funding + technological advantages, positioning them well and with high scarcity, they are expected to develop super AI applications; 2) Leading hardware companies represented by Apple (mobile) and Tesla (autonomous driving) have scarce user value, and they can become new traffic entry points in the AI era, and have the value of reshaping application traffic distribution. US M7-AI layout summary: All essentially have AI full-stack capabilities and actively make extension investments to bet on startups, but there are slight differences in the direction of investment focus based on their own endowments and ecological scenarios. NVIDIA: GPU; Amazon: Cloud (AWS) + AI applications (e-commerce, logistics, smart home, etc.); Microsoft: Cloud (Azure) + large models (extension investments in Open AI, Inflection AI + self-developed multimodal and small models) + enterprise AI applications; Google: Computing power (self-developed TPU chips + cloud services) + large models (Gemini, etc., focusing on multimodal) + AI applications (search engines, etc.). Meta: open source large models (LLama) + high-value social application ecosystem (AI-enabled advertising/social/video, etc.) + hardware (AI glasses, Siasun Robot & Automation, etc.), in addition, Microsoft and Google also have application ecosystem advantages. Apple/Tesla: technology companies with stable hardware user base, focusing mainly on existing hardware ports. Outlook: How to view the trend of China's AI industry? From mapping to implementation. DeepSeek brings about demand and industrial chain reshaping. With outstanding model capabilities, DeepSeek drives APP products quickly to market, 1) The surge in user call demand has pushed up the demand for inference computing resources, thereby driving the explosion of public cloud demand; 2) The rapid embrace and deployment of AI applications on both the B-end and C-end. It has gradually begun to realize the increase in cloud computing power demand transform B-end efficiency software + remake internet age APPs innovate the trend of complex C-end AI experiences, echoing the strong industrial trends in the US stock market in 2023. Enlightenment: If the industrial reshaping brought by open source large models is similar to the United States in 2023, the first enlightenment it may bring is that cloud revenue, subscription revenue, and advertising revenue are the first areas closest to implementation in the initial stage. Understanding the business model of M7 can help identify the most worthwhile investment direction in this round. The second enlightenment may be that the investment value of configuration-type weight stocks this year will be strengthened, different from the completely dominant logic of the previous two years; internet giants Alibaba and Tencent may show similar configuration value. Alibaba: Benchmarking Microsoft/Amazon, AI full industrial chain layout, leading public cloud (leading in IaaS and PaaS market share in China) + leading open source large models (the Qwen2.5-Max model capacity of Tongyi Qianwen is leading in China and seventh globally, the champion of non-inferential Chinese large models) + e-commerce scenarios (rich application ecosystem covering e-commerce retail, office, local life, payment, etc., all expected to integrate AI transformation). Tencent: Benchmarking Meta, strongest application scenarios in the internet industry. Tencent has one of the strongest internet scenarios in the internet era - instant messaging, optimistic that with the improvement of base model capabilities, AI transformation of existing APPs will bring new growth points. Currently accelerating the integration of AI with applications, several products have been integrated into Deep Seek, including Yuanbao, Peace Elite, QQ Music, Tencent Finance, WeChat search, etc., and expects new growth.AI empowers its application under a state-card positioning. In addition, the company also values model research and development & technology reserve, self-developed hybrid large models, which are still being iteratively upgraded.Investment advice: optimistic about the reshaping of the AI ecosystem with large open-source models + policy cycle resonance, and focus on recommending TENCENT (expected improvement in profit capability of Tencent Cloud + leader in the trend of APP AI transformation), Alibaba (trend of cloud revaluation is confirmed); in addition, it is recommended to pay attention to internet platforms + games with user data and platform scenarios, including Kingnet Network, Giant Network Group, Zhejiang Century Huatong Group, Beijing Ultrapower Software, Perfect World, Kuaishou, ALI PICTURES, CHINA LIT, Tencent Music, Meitu, etc. In addition, in the upstream, it is recommended to pay attention to Tencent Cloud and Aliyun Chain, GPU, and in the downstream, it is recommended to pay attention to opportunities in interactive hardware.