Last night, the U.S. stock market dipped slightly after non-farm payrolls exceeded expectations. Why did the semiconductor sector strengthen? Why is it Astra driving this instead of the fab5.1 driver released a day earlier? Let’s dive deeper into this based on last night’s analysis. From a personal perspective, valuation is one factor, but the core reason is the breakthrough in model capabilities brought by Astra (the leap in model capabilities driving higher expectations). A few key points: 1. Officially disclosed: GPT-6 Astra is OpenAI’s first model pre-trained on over 100,000 GPUs (Texas Stargate base). Model training still heavily relies on computational resources. 2. Astra focuses on long-term agents that can truly operate computers, browsers, and professional software—models capable of interacting with real GUIs. It emphasizes creative production, 3D/game-related capabilities, scientific research, etc. The core is the breakthrough in capabilities, not just performance improvements. This confirms to the market that cutting-edge model development is still delivering valuable advancements in capabilities. This also explains why fab5.1, released on September 1st, didn’t boost the AI semiconductor sector: fab5.1 is still focused on long-range encoding and knowledge work, essentially making models smarter. Astra, on the other hand, demonstrates that models can now handle tasks they previously couldn’t do well—or couldn’t do at all. 3. The stronger the model’s capabilities, the more new applications can be developed, and the more likely the AI spending cycle will continue. If new models expand the range of tasks that can be commercialized, then the servers already purchased and the data centers already built will have stronger future utilization rates and revenue conversion. This means infrastructure has a better chance of being filled with real workloads. This is the core of AI: the competition between models is shifting from price and efficiency battles to direct competition in intelligent capabilities. For the market, if leading advantages can be reflected in the intelligence of the models themselves, the sustainability of the AI capital expenditure cycle will gain new support. This likely explains why the semiconductor sector strengthened yesterday despite the broader market weakness. It’s the fundamental logic driven by industry dynamics. From a personal perspective: In the short term, this momentum should continue until next week’s CPI and PPI data. If inflation doesn’t show a significant unexpected rise, there’s still room for further growth—Astra is the new driving force. Of course, if inflation resurges or remains stubbornly high, the market may still face macroeconomic pressures. However, once these macro pressures ease in the mid-to-long term, Astra—or future models with even greater breakthroughs in capabilities—will continue to drive new market trends.