Huang Renxun: The market has completely misunderstood DeepSeek.

The emergence of DeepSeek has raised doubts about the logic of large-scale AI computing power, leading to a sharp drop in the stock price of NVIDIA Corporation (NVDA.US) at one point. However, Huang Renxun stated in a recent interview that the market's understanding of DeepSeek is "completely backward". He believes that the release of DeepSeek R1 will not only not end the demand for AI computing, but will instead inspire the pursuit of more efficient AI models in the market, thereby driving the development of the entire industry. "It makes everyone realize that there is a chance to make models more efficient than we imagined. Therefore, it is expanding and accelerating the adoption of AI." NVIDIA Corporation will release its fourth quarter financial report on February 26th, and is expected to provide a more detailed response to the market reaction. Huang Renxun: Technological advances drive the popularization of AI, while market reactions vary greatly The open source release of the DeepSeek R1 model caused a sensation. After the announcement, the stock price of NVIDIA Corporation plummeted, closing at $142.62 on January 24th and dropping to $118.52 on January 27th, evaporating $60 billion in market value in just three days. Currently, NVIDIA Corporation's stock price has largely recovered to pre-fall levels, trading at $135 in after-hours trading. Market confidence in NVIDIA Corporation seems to be recovering. The market seems to believe that the technological progress of DeepSeek means the end of AI computing demand, posing a direct threat to chip giant NVIDIA Corporation. However, in an interview recently released by Alex Bouzari, CEO of DataDirect Networks, Huang Renxun refuted the market's interpretation. Huang Renxun pointed out directly that there is a fundamental "mental model" misconception about AI development in the market: "Investors think the world is only about pre-training, then inference. And inference is you ask AI a question, and it immediately gives you an answer, a one-time answer. I don't know whose fault it is, but it's clear that this model is wrong." He stated that the release of R1 is "incredibly exciting" and emphasized that the market's reaction is "completely the opposite". "It makes everyone realize that there is a chance to make models more efficient than we imagined. Therefore, it is expanding and accelerating the adoption of AI." In other words, the emergence of R1 does not mean that the market no longer needs computing resources, but rather it has sparked a pursuit of more efficient AI models in the market, driving the development of the entire industry. Huang Renxun further pointed out that while DeepSeek has made progress in the pre-training of AI models, later training is still crucial and requires a large amount of resources. He emphasized: "Inference is a part that consumes a considerable amount of computing power." This means that even with improved pre-training efficiency, the inference process of AI models still requires strong computational support, and NVIDIA Corporation's chip products will continue to play a key role in the development of AI. Meanwhile, on February 21st, DeepSeek announced that starting next week, they will open-source five code repositories to transparently share their research progress with the global developer community, unlocking new content daily and defining this plan as "Open Source Week". DeepSeek stated that the building blocks in their online service have been documented, deployed, and tested in practice, and they hope that every line of code shared will become a strong potential energy to accelerate industry development. The team also stated that there is no ivory tower in this field, only the pure entrepreneurial spirit of the garage and the innovative force built by the community. This article is reprinted from "Wall Street View", author: Zhang Yaqi; GMTEight Editor: Jiang Yuanhua.