Analysis of Sino-US AI Industry Chain Competition Pattern and Investment Opportunities — Based on Jensen Huang's AI 'Five-Layer Cake' Theory

Based on Jensen Huang’s ‘Five-Layer Cake’ theoretical framework for AI, I will deeply analyze the Sino-US competition situation and its impact on the valuation of investment targets from three key layers: energy, chips, and infrastructure.

• Market cap of 169.37 billion USD, P/E ratio of 25.72x [0]
• Cumulative increase of 13.56% in 2025, with steady performance [0]
• Leading clean energy enterprise, in line with the greening trend of AI data centers

• Market cap of 798.2 billion USD, P/E ratio of only 14.00x, with more advantageous valuation [0]
• China indeed has ‘twice the resources’ of the US in energy resources [1]
• Benefits from the growth of AI computing power demand during the transition from traditional energy to clean energy

China has obvious advantages in energy cost and resource endowment, which provides a low-cost foundation for AI computing power infrastructure construction. Energy enterprises like Shenhua have relatively low valuations and have room for revaluation amid the AI-driven growth in energy demand.

• Market cap of up to 4.33 trillion USD, P/E ratio of 43.58x [0]
• Cumulative increase of 28.49% in 2025, 3-year increase of 972.54% [0]
• Data center revenue accounts for 88.3%, US market revenue accounts for 46.9% [0]

• Market cap of 120.8 billion USD, P/E ratio of 57.45x, relatively high valuation [0]
• Undertakes the mission of补全算力生态短板 under the background of domestic substitution [1]

The Trump administration recently announced that NVIDIA can export H200 chips to China, but with a 25% share requirement, and the latest Blackwell chips are still under embargo [2][3]. This contradictory approach reflects the US’s balance between technological leadership and market share.

The chip layer is where the gap between China and the US is the largest. NVIDIA enjoys an ultra-high valuation premium due to technological monopoly, but Chinese chip enterprises have advantages in cost and local market. Although Chinese chip enterprises under the domestic substitution theme have relatively high valuations, they still have investment value driven by policy support and market space.

• Market cap of 18.84 billion USD, P/E ratio of 23.79x [0]
• Server and storage system revenue accounts for 97.0% [0]
• However, cumulative increase in 2025 is only 5.36%, with weak recent performance [0]

• China leads in the speed of AI infrastructure construction [1]
• Rich application scenarios promote the rapid deployment of infrastructure

The chart shows that since 2024, US-based NVIDIA has performed the most prominently in the chip field, with a cumulative return rate exceeding 100%; Chinese enterprises in the energy and chip fields have performed relatively steadily, but infrastructure construction speed is faster [0].

1. Energy Layer:
   Focus on the transformation of traditional energy enterprises like China Shenhua to AI computing power services; the relatively low P/E ratio (14x) provides a safety margin.

2. Chip Layer:
   NVIDIA remains the AI chip leader, but policy risks need to be watched; Chinese chip enterprises have thematic investment opportunities under the domestic substitution background.

3. Infrastructure Layer:
   Focus on server and data center-related enterprises that benefit from the growth of AI computing power demand in both China and the US.

• Energy cost advantage provides basic competitiveness for the entire AI industry chain
• Rich application scenarios and fast infrastructure deployment capabilities
• Open-source model development and AI patent count accounting for 70% of the global total [1]

• Leading edge in cutting-edge AI model technology
• Perfect chip design and manufacturing ecosystem
• Global financial market and capital advantages

1. Policy Risk:
   The Trump administration’s chip export policy may be adjusted at any time, affecting the performance of related enterprises [2][3]

2. Technology Competition:
   AI technology iterates rapidly, and today’s leading advantages may be overturned

3. Valuation Risk:
   Some AI concept stocks have reached historical high valuations, with potential callback risks

Based on Jensen Huang’s ‘Five-Layer Cake’ theory, China and the US show a complementary rather than completely替代 competitive pattern in various links of the AI industry chain. China has basic advantages in the energy field, the US leads in chip technology but faces cost challenges, and China has faster construction speed in the infrastructure layer. Investors should build a diversified investment portfolio based on the competitive characteristics and valuation levels of different links, focusing on both the growth of technology leaders and the revaluation opportunities of cost advantage enterprises.

[0] Jinling API Data - Financial data of companies like NVDA, TSM, NEE, SMCI, 601088.SS, 603019.SS
[1] OFweek Vekenet - “The Watched Moore Thread, the Ready-to-Launch Muxi, and Anxious Investors”
[2] Guancha.cn - “Trump: NVIDIA H200 can be sold to China, but with a condition”
[3] Phoenix New Media - “US Media: A Major Lobbying Victory for NVIDIA”