Core Viewpoint - Current AI development does not exhibit significant bubbles, with supply chain finance for major AI companies still in its early stages, utilizing equity investment rather than debt, which reduces risk compared to the internet bubble period [1][2] Group 1: AI Development and Market Dynamics - The AI to B business model is gradually forming a closed loop, benefiting cloud service providers (CSPs) from accelerated AI growth, with industrial leaders like TSMC, UPS, and Walmart beginning to use AI to enhance production efficiency [3][4] - The combined capital expenditure (CAPEX) of the four major North American CSPs (Google, Meta, Microsoft, Amazon) reached $97.3 billion in Q3 2025, a 65% year-on-year increase, with expectations for continued CAPEX growth in 2026 [1][2] - The revenue of the four CSPs in Q3 2025 was $396.6 billion, up 17% year-on-year, while their net profit totaled $86.6 billion, reflecting a 6% increase despite Meta's one-time non-cash loss of $15.9 billion due to tax policy changes [1][2] Group 2: Financial Metrics and Projections - The CAPEX/EBITDA ratio for the four CSPs reached 62% in Q3 2025, indicating room for further capital expenditure growth, with historical data showing Amazon's CAPEX/EBITDA exceeding 100% in multiple quarters [2] - OpenAI's revenue is projected to grow from $2.8 million in 2022 to $13 billion in 2023, with expectations of reaching $130 billion by 2025, making it one of the fastest companies to achieve over $100 billion in revenue post-establishment [4] - Anthropic is expected to achieve $70 billion in revenue by 2028, with positive cash flow anticipated, highlighting the strong growth potential in the AI sector [4] Group 3: Stock Performance and Valuation - Current stock price growth for major AI companies is primarily driven by earnings per share (EPS) growth, with NVIDIA's stock price and EPS showing similar increases since Q1 2023 [5] - The forward price-to-earnings (PE) ratios for leading AI stocks and the four major CSPs are currently lower than their peaks in 2024, suggesting potential for further growth [5]
国金证券:当前AI发展并无明显泡沫 继续看好AI核心硬件标的