Large Language Models Breakthrough Triggers Computing Power "Acceleration Race" - Caitong Securities Recommends Focus on LENOVO GROUP (00992), KINGDEE INT'L (00268) and Two Other Stocks

Stock News
Sep 18

According to recent reports, the research team led by Hao Yanhui at Caitong Securities released a research report stating that based on current AI deployment comparisons between Chinese and US internet companies, the gap in models and capital expenditure is narrowing. Global large language models have initially formed a competitive landscape dominated by OpenAI, Anthropic, Google, and Tesla, with DeepSeek and Alibaba's Tongyi Qianwen from Chinese enterprises entering the first tier.

According to GPQA test results, the top 25 positions are primarily occupied by large models from OpenAI, Anthropic, Microsoft, Google, and Meta, with relatively low representation from Chinese companies. Among them, DeepSeek-V3/R1, released in December 2024/January 2025, disrupted the global AI landscape and officially represents Chinese/open-source large models achieving state-of-the-art alignment; Alibaba's Qwen-3 follows closely behind, with performance entering the first tier.

In the near term, leading tech companies are heavily investing in large model training and adapting models to their business needs through inference; training and inference together boost internal computing power demand. On the external supply side, cloud providers are also offering large model APIs on their proprietary cloud computing platforms, with the MaaS business model driving external computing power expansion.

With strong demand from both internal and external sources, companies are accelerating computing resource investments, driving high CapEx growth. The capital expenditure intensity (capital expenditure as a percentage of revenue) for Microsoft/Google/Meta/Amazon reached 34.8%/23.3%/35.8%/18.7% in Q2 2025.

Regarding domestic internet giants, in Q2 2025, Baidu/Alibaba/Tencent's capital expenditures increased by +10.2%/+162.7%/+319.1% year-over-year to 23/318/366 billion yuan respectively. Domestic giants began accelerating AI investments approximately one year later than overseas counterparts, and currently their CapEx-to-revenue ratio still has a considerable gap compared to overseas giants.

Investment recommendations include:

MEITUAN-W (03690): Possesses recovery potential after valuation adjustment

KINGDEE INT'L (00268): Cloud business maintains high growth, subscription transformation brings sustainable ARR growth, AI+ERP expected to open new growth curve

LENOVO GROUP (00992): Expected to emerge from slow bull trend, AI PC product cycle begins, data center business benefits from increased AI computing power demand

Tencent Holdings (00700): Long-term preferred choice

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10