Several points worth paying attention to in Nvidia's third quarter financial report were announced in the early morning, with Nvidia's financial report exceeding expectations and future guidance also exceeding expectations. On Monday, I talked about my personal preference for Nvidia's strong financial guidance this time, and it doesn't seem like I've been slapped. Let's talk about the points worth paying attention to in this financial report: 1. Reversing the slowing trend, the growth rate rises and accelerates again In the second quarter, Nvidia's revenue increased by 56% year-on-year and earnings per share increased by 59% year-on-year. The third quarter revenue was 62% year-on-year, and the earnings per share were 60% year-on-year. The official forecast for fourth quarter revenue and median earnings per share are both 65% year-on-year. This means that downstream demand has not slowed down, but rather begun to accelerate. And the guidance for the fourth quarter is based on the assumption that there is no data center computing revenue from China. 2. The speed of valuation digestion is very fast I don't know how many people will pay attention, but Nvidia's static P/E ratio for the past 24 years is 63, and its rolling P/E ratio for the past four quarters (Q3 Q2 of 24) is 53, while its expected dynamic P/E ratio for the past 25 years has dropped to 30. If the latest third quarter financial report and fourth quarter financial report forecast are added, the 25 year dynamic P/E ratio corresponding to the current stock price is less than 30, only around 28. What is this concept? By the fourth quarter, Nvidia's expected revenue and median earnings per share are expected to increase by 65% year-on-year, but the current market's dynamic PE for the whole year is less than 30 times. From this perspective, it can be said that the valuation is underestimated, as it was digested too quickly. 3. The three main drivers of growth During the conference call, Lao Huang believed that Nvidia's growth came from: The transition from CPU to GPU accelerated computing in the post Moore's Law era The transformation of existing applications by generative AI And the new revolution brought by agentic AI. The CFO revealed that the cumulative revenue visibility of their next-generation chip platforms Blackwell and Rubin has reached $500 billion, and demand continues to exceed expectations. 4. Bottleneck physical limitations An analyst asked what is the biggest growth bottleneck? Lao Huang did not avoid it, stating that "electricity, heat dissipation, and liquid cooling" are all huge challenges. He admitted that building a gigawatt level data center not only requires chips, but also complex energy infrastructure coordination. It's necessary to take a look at Lao Huang's words. There will be three large-scale platform transformations in the future: 1) The transition to accelerated computing is fundamental and necessary, and is crucial in the post Moore's Law era; 2) The transformation towards generative AI is transformative and necessary, empowering existing applications and business models; 2) The transformation towards intelligent agents and physical AI will be revolutionary, giving rise to new applications, companies, products, and services. When considering infrastructure investment, please take into account these three basic dynamics. Each one will drive infrastructure growth in the coming years. Nvidia was chosen because our single architecture can achieve all three transformations and is applicable to any form and modality of AI, spanning all industries, every stage of AI, and all diverse computing needs in the cloud, from cloud to enterprise to machine.