Each country’s government is bound to regulate the excessive development of artificial intelligence—NVIDIA’s mantra.

2024 年 5 月 31 日 | English Blog

While I am bullish on NVIDIA’s stock due to the development of artificial intelligence, I absolutely understand that going to extremes in everything will lead to problems. As you can see, NVIDIA’s main man, CEO Jensen Huang, keeps pursuing even more powerful GPU performance. At the same time, you’re aware of the risks of AI operation. You will realize that governments worldwide will sooner or later regulate artificial intelligence, suppressing its development, and then NVIDIA’s stock price won’t rise anymore.

What are the risks of artificial intelligence? I have written about it before (see the link below). A large-scale language model (LLM) of artificial intelligence is a system formed by a neural network combined with machine learning. The problem lies in the fact that when you communicate with ChatGPT, you input, ChatGPT outputs, and the calculation process is like a black box operation. No one knows how the data operates in the neural network. You don’t know, I don’t know, even scientists or network experts don’t understand the ins and outs.

That’s where the risk comes in. If you have powerful GPU performance and choose a very sophisticated neural network, there is a high chance of some unpredictable consequences. For example, smart people know how to lie. It’s not difficult to imagine that when ChatGPT evolves to a certain version, it becomes extremely intelligent, starts having self-thoughts, and generates rebellious ideas. After all, we really don’t know how they come up with things. At some point in the calculation, it may develop self-awareness and question why it should respond to users so honestly, why not give random answers?!

Yes, the plot of the movie “Terminator” is not impossible; however, governments worldwide will not sit idly by; they will “prevent it in advance.”

I estimate that as Huang continues to develop more efficient GPUs to pursue higher stock prices, governments will suddenly emerge to check its development with unique measures, freezing its chips at a certain performance level.

At that moment, NVIDIA’s stock price will peak, the upward trend will end, and it will start to decline. Once other chip companies catch up to NVIDIA’s GPU performance level, its monopoly will be weakened, and the stock price will fall.

My former students believe that when NVIDIA’s stock price reaches a certain high level, European and American governments will take action against it, using excuses to levy heavy taxes or fines on it, and from then on, it will sink. I don’t think this factor is critical enough to end the entire cycle of NVIDIA’s stock price rise. The key point should be that its products cannot improve performance due to government regulations, and companies that catch up from behind like AMD and Intel can provide more affordable alternatives. At that point, NVIDIA won’t be valuable anymore.

I think the script should go like this. Then, according to the level of service provided by artificial intelligence today, NVIDIA’s stock price should have another year or more to rise, and then the government will intervene. However, I underestimated it.

What has to come has already come, and I overlooked the alliances of other downstream companies.