1. The significance of guessing the peak of this bull market is minimal; the key lies in the direction of the real estate market. If housing prices surge alongside the bull market and capital flows, the bull market may not have a peak, reshaping asset confidence; conversely, caution is needed, as history may repeat itself.

2. The current U.S. strategy has led to a significant repatriation of capital from the EU, Japan, and South Korea, benefiting NASDAQ and AI infrastructure; asset analysis must closely monitor capital flows.

3. To counteract involution, it is necessary to combine demand-side policies. Success on the supply side cannot be separated from support on the demand side; for example, the beer industry has no involution but is dragged down by deflation.

4. Fertility subsidies may see "layered increases"; if policies shift from the supply side to the demand side, situations similar to technology subsidies may arise, requiring attention to subsequent developments.

5. The 14th Five-Year Plan determines the direction of capital; analysis of various assets must use this as an important basis.

6. GPT-5 being "below expectations" is a strategy for managing expectations by OpenAI. The new consensus in Silicon Valley shifts towards the practicality of models, with Wall Street evaluating AI based on the "economic Turing test," where success is defined by the ability to enhance productivity.

7. When user numbers reach a billion, practical value becomes evident; even slight efficiency improvements can significantly boost GDP. Recent increases in U.S. AI hardware reflect market recognition of this.

8. U.S. AI capital expenditures are expected to account for 25% of its GDP growth by 2025, continuing its infrastructure-building tendency, and we are unlikely to miss this opportunity.

9. The total monthly active users of mainstream AI applications in the U.S. is 1 billion, while similar applications in our country have less than one-tenth of that, indicating a clear gap in AI development.

10. To evaluate AI targets, one can look at "human and card" (algorithm and computing power); many A-share companies lack talent and computing power, making it difficult to achieve AI value.

11. GPT-5 adopts new paradigms such as synthetic data; data barriers are not insurmountable, and small companies find it challenging to build a protective moat based on data.