Last week, the Nasdaq index fell 4.5%, semiconductor ETFs dropped over 10%, and the CNN Fear & Greed Index slid to 18, entering the "extreme fear" zone. AI stocks, which were previously chased by the market, are now being sold off en masse. This coincides with earnings season. On Tuesday, Netflix, SpaceX, General Motors, and GE Aerospace reported results; on Wednesday, Meta, Tesla, IBM, and Lam Research followed. Of these, I’m most focused on Meta, as its earnings report may determine how the market views the entire AI infrastructure investment cycle going forward. For the past two years, Wall Street largely operated under one assumption: that tech giants would continue ramping up capital expenditures, guaranteeing sustained growth for NVIDIA, memory, data centers, and energy. But now, investors are asking: when will these investments actually pay off? Market expectations for Meta’s Q3 revenue exceed $63 billion, yet its AI spending is squeezing profit margins. Meta will now need to prove not only that AI has improved ad monetization efficiency, but also clarify how much it plans to invest next year. Compounding the pressure, the U.S. 10-year Treasury yield has risen to around 5%. Higher financing costs are making the entire AI supply chain—built on massive capital spending—increasingly strained. What I find most telling this earnings week is how the market reacts to good news. If major companies report revenue beats but their stock prices still fall due to higher capital spending, it signals that Wall Street is re-pricing AI growth. Two years ago, announcing an additional $10 billion in GPU purchases was enough to excite investors. Now, everyone is finally asking: how much profit will these GPUs actually generate?