NVIDIA's latest earnings report analysis

Key points to watch in NVIDIA's Q3 earnings report: NVIDIA's earnings released early this morning comprehensively exceeded expectations, with future guidance also surpassing expectations. On Monday, I mentioned my personal inclination that NVIDIA's earnings guidance this time would be very strong, and it seems I wasn't wrong.

Let's talk about the key points to watch in this earnings report:

1. Reversing the slowdown trend, growth is picking up again. In Q2, NVIDIA's revenue grew 56% YoY, with EPS up 59% YoY. In Q3, revenue grew 62% YoY, and EPS grew 60% YoY. The official Q4 forecast shows both revenue and EPS median growth at 65% YoY. This means downstream demand is not only not slowing down but actually accelerating. Moreover, the Q4 guidance was made under the assumption of "no data center computing revenue from China."

2. The speed of valuation digestion is very fast. I wonder how many people noticed: NVIDIA's 2024 static P/E is 63, the trailing P/E for the past four quarters (Q3 2024 - Q2 2025) is 53, and the expected dynamic P/E for 2025 has already dropped to 30. If we include the latest Q3 earnings and Q4 forecast, the current stock price implies a 2025 dynamic P/E of less than 30, around 28. What does this mean? By Q4, NVIDIA's expected revenue and EPS median growth is 65% YoY, but the market is currently pricing it at less than 30x forward P/E. From this perspective, it's fair to say it's undervalued—valuation digestion is happening too fast.

3. Three major growth drivers. On the earnings call, Jensen Huang attributed NVIDIA's growth to: the shift from CPUs 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 (Agentic AI). The CFO revealed that the revenue visibility for their next-gen chip platforms, Blackwell and Rubin, has already reached $500 billion, with demand continuing to exceed expectations.

4. Physical bottlenecks. When asked about the biggest growth bottleneck, Jensen didn't dodge the question, bluntly stating that "power, cooling, and liquid cooling" are all huge challenges. He admitted that building gigawatt-scale data centers requires not just chips but also complex energy infrastructure. It's worth listening to Jensen's comments on this.

There are three major platform shifts ahead:

1) The shift to accelerated computing is foundational and necessary, critical in the post-Moore's Law era;

2) The shift to generative AI is transformative and necessary, empowering existing applications and business models;

3) The shift to agentic and physical AI will be revolutionary, spawning new applications, companies, products, and services. When considering infrastructure investments, keep these three fundamental dynamics in mind. Each will drive infrastructure growth in the coming years. NVIDIA was chosen because our unified architecture can enable all three shifts, applicable to any form and modality of AI, across all industries, every stage of AI, and all diverse computing needs in the cloud—from cloud to enterprise to robotics.