"AI Makes Major Contributions! Global Tech Giants Report Better-than-Expected Earnings with Significant Increase in Capital Expenditures"

On July 31, the world's leading computer software provider announced its financial results for Q4 of fiscal year 2025, reporting revenue of $76.44 billion (compared to $64.7 billion in the same period last year) and net profit of $27.233 billion (up from $22.036 billion YoY). The company stated that its cloud business growth exceeded expectations and highlighted record-high spending on AI infrastructure. Its closely watched cloud computing division saw a 39% increase in sales during Q4, with 16% attributed to AI growth. In terms of capital expenditures, the company expects to spend over $30 billion in Q1 of fiscal year 2026.

On the same day, the world's leading social media platform released its Q2 financial report, showing revenue of $47.52 billion (a 22% YoY increase) and net profit of $18.34 billion (up 36% YoY). The company projected its full-year capital expenditures to be between $66 billion and $72 billion.

Driven by AI, computing power demand is surging. Whether from large models, consumer-facing AI applications, or enterprise AI applications, the potential for computing power expansion is vast! AI demand is expected to boost the semiconductor equipment supply chain, with advanced manufacturing processes continuing to expand, driving growth in the semiconductor sector.

"AI Computing Power Demand Soars, Stimulated by Large Models and Application Scenarios"

Dongwu Securities stated that it is optimistic about the development of AI computing power demand, noting that large model inference costs have significantly declined. With low-cost, high-performance, and open-source models like DeepSeek, advancements in inference and training chips are accelerating, promoting AI applications and hardware development. (Source: Dongwu Securities report, July 21, 2025, "Focus on Rapid Growth in AI Computing Demand, Domestic Equipment Suppliers to Benefit Significantly")

According to comprehensive forecasts from CCID Consulting, IDC, Inspur Information, and CITIC Securities, the exponential demand for AI chip computing power from large models will continue to rise in the coming years. By 2026, CCID Consulting predicts China's intelligent computing power scale will reach 1,271.4 EFLOPS, while a joint report from IDC and Inspur Information estimates it at 1,460.3 EFLOPS—both indicating rapid growth. In terms of AI chip market size, CCID Consulting projects China’s market to reach ¥120.6 billion, while CITIC Securities is more optimistic, expecting domestic scale to exceed ¥300 billion.

Beyond large models, application-side demand for AI computing power is also surging. Everbright Securities analyzed the core drivers of AI computing power growth from multiple dimensions:

1. Consumer AI Applications: Demand is shifting toward high-token-consumption areas, supporting AI computing power growth. Generative AI applications show steady traffic growth, with AI programming and AI music/video generation experiencing rapid adoption, while writing, customer service, and image-generation tools see declining usage.

   • AI Programming: A key area for general large model performance iteration with vast market potential.

   • AI Video/Audio: Continuous model upgrades create commercialization potential from scratch.

2. Enterprise AI Applications: Customized AI Agents are the primary implementation scenario, with long-term commercial value. AI Agents are the ideal form for enterprise AI adoption overseas, emphasizing strong personalization and customization to meet complex industry needs. Standardized protocols like MCP and Agent2Agent are accelerating Agent development, with cloud providers and SaaS companies prioritizing Agent strategies, leading to rapid growth in Agent development platform usage. (Source: Everbright Securities report, July 29, 2025, "Hot Demand for AI Computing in U.S. Stocks: How to View Downstream Application Trends?")