What are the key takeaways from NVIDIA's GTC keynote?

Today, let's talk about NVIDIA's GTC keynote speech.
Why should we pay attention to NVIDIA? Because NVIDIA is the top trendsetter in this wave of AI.
This AI wave can be said to be a once-in-20-years top narrative; previously, it was the internet in 2000.
We absolutely cannot miss this opportunity again.

Why do I say that? First, let's look at a device NVIDIA announced in March 2025 — the SpectrumX photonic packaging network switch, and the concept of AI factories.
It mentions optical communication.
I believe that those who carefully read this article and deeply research the underlying industry are already "winning big."
Whether it's the fiber optic company Corning in the US stock market, or optical chip companies like Lumentum and Marvell Technology,
the latter being Zhongji Xuchuang, Xin Yisheng, and SourceJie Technology, Shijia Photonics in China;
and CPO modules from Broadcom, Bocon Technology, etc., which have already surged tenfold or more.

So, subsequent NVIDIA releases and speeches are definitely worth watching.
Follow NVIDIA's line of thought to explore if there are any opportunities.

Let's first review the key points announced at the conference:

First, they mentioned starting to develop CPUs.
NVIDIA is also entering the CPU market, collaborating with Microsoft on a new PC AI.
In the future, this will extend to robots, satellites, and cloud base stations.
I also believe that the future will be dominated by robots.
So, which robot companies are there now? Tesla? Yushutec?
For satellite launches, there’s SpaceX, and who else?
As for cloud base stations, which companies are working on this?

Then, they talked about developing robot medical services and robot taxi services.
Which companies and core components are involved in this area?

I did a preliminary search — here are some companies (do your own AI research; this is just my initial AI-based search, as covering all aspects requires huge time, effort, and data organization):

What about companies and components related to robot cars?
Car-related components are numerous.
I roughly identified: chips — NVIDIA, Qualcomm, Broadcom;
LiDAR — LAZR, OUST;
Camera sensors — Sony, ON (Siemens);
Millimeter-wave radar — NXP;
In-vehicle networks and optical communication — Broadcom, Marvell, GLW;
Motors and power semiconductors, etc.

Huang Renxun then announced the development of an AI PC with Microsoft, developed over three years, to be announced tomorrow night.
My personal feeling is that it might be similar to the current Xiaolongxia (small lobster).
Will Microsoft take this opportunity to see a small rise in the stock tonight?

NVIDIA also launched its own AI models and AI CPU — Vera Rubin, and the super factory DSX.
I won't go into too much detail about these products.

I think this is very important — the future operation mode of AI computers.
We only need to describe our needs and intentions, and AI will automatically generate results.
For example, if I want to write a research report on NVIDIA, I just tell AI, and it will open the browser to search or launch relevant software.
It's similar to ByteDance's AI phone before — it can automatically open all applications, maybe even appearing on our PC.

Based on this thinking, what other mainframes do computers need?
Just go fully cloud-based.
This should be the main trend because, in the past, each person's PC had its own local computing power, which was not centralized.
But in the cloud, computing power becomes collective, increasing resource utilization.

Tokens are assets that can generate profits!