Jen-Hsun Huang will once again take the stage at the GTC summit. Can he save NVIDIA's stock price?

On March 17th, U.S. time, Jen-Hsun Huang, the founder and CEO of NVIDIA, will face an important opportunity to revitalize the company's stock price. On that day, NVIDIA's annual technology summit, GTC, will be held in San Jose, California. During the keynote speech, it is expected that Jen-Hsun Huang will elaborate on how he leads NVIDIA to explore the next frontier of AI. Over the six-day conference, 900 business partners and startups will also introduce how they use NVIDIA's technology and showcase related products.

Suddenly encountering a "black swan", the stock price has not yet recovered.

The genuine and confident Jen-Hsun Huang also experiences anxiety—he needs to lead NVIDIA to create an endless demand for AI chips and steadily maintain a dominant position. He has witnessed the dramatic changes in the fortunes of companies within the semiconductor industry and is well aware of the ups and downs in the history of technology infrastructure companies, with Intel being the latest example.

At the CES conference held at the beginning of this year, Jen-Hsun Huang made a high-profile appearance. Dressed in a shiny crocodile-patterned leather jacket, he talked extensively about physical artificial intelligence on stage and claimed that "the ChatGPT moment of general robots is about to arrive."

In the following week, Jen-Hsun Huang visited China and reaffirmed NVIDIA's commitment to the Chinese market to developers and partners in China. He also stated that NVIDIA has the lowest employee turnover rate among global tech companies.

Until Jen-Hsun Huang left China, Nvidia's stock performance was sufficient for his personal net worth to easily exceed 100 billion USD. However, at the end of January, Nvidia's stock trend reversed—Chinese AI startup DeepSeek became a "black swan" that attacked Nvidia's stock price. The release of a low-cost AI model by this company caused Nvidia's market value to evaporate by nearly 600 billion USD in one day, setting a record for the largest single-day drop in history for any company.

Despite Nvidia's quarterly revenue and profit exceeding expectations, and the company making an optimistic forecast for sales in the new quarter, the stock price still fell. The market has begun to question the demand for Nvidia chips for the first time. Amid these doubts, Nvidia's stock price has not fully recovered. As of pre-market on March 17, Nvidia's market value was less than $3 trillion.

On March 17, when Jen-Hsun Huang stands on the GTC stage again, he is expected to announce NVIDIA's next-generation AI chip platform, which will be named Rubin, after the American female astronomer Vera Rubin, who focused on the study of "dark matter" and is known for observing the rotation speeds of galaxies.

On Tuesday, Jen-Hsun Huang will introduce the speed of the Rubin chip, the configurations to be adopted, and when it will start shipping. At last year's GTC conference, Jen-Hsun Huang announced the launch of the Blackwell chip, which is named after the first African American scientist elected to the National Academy of Sciences in the United States. As of the fourth quarter of fiscal year 2025, the chip has contributed $11 billion in revenue to NVIDIA.

Customers demand more autonomy in their choices.

As the capabilities of NVIDIA's AI chips surpass one generation after another, their prices are also becoming increasingly expensive. This not only deters some small and medium-sized enterprise users, but even large cloud service providers are trying to break free from their dependence on NVIDIA chips. Against this backdrop, NVIDIA is seeking new sales strategies, and related technological products are expected to be announced at this year's GTC summit.

One connection technology known as Socket has attracted attention in the industry. With this technology, it will be possible in the future to directly assemble and disassemble independent GPU chips, whereas current GPUs must first be packaged into a board, which is then connected to the motherboard, and the GPU chips on the board are non-removable.

Research institution Gartner analyst Sheng Linghai told First Financial reporters: "If this technology is really launched on GTC, it may open up a new market, and GPU boards can become products with disassembly freedom."

In the view of Shenglinghai, detachable boards will give customers more choices without the need to purchase complete server machines. "From DGX to NVL72 clusters, Nvidia has shifted from selling boards to selling servers, and product prices are getting higher and higher. This sales model can significantly increase revenue," he explained. "However, many customers now believe that its products are too expensive, and the willingness to buy has decreased. In this situation, Nvidia may need to consider reintroducing a model of selling detachable boards individually to diversify its product strategy."

Due to the fact that large cloud service providers may only want a piece of hardware at the bottom of the stack in Nvidia servers, they are not satisfied with Nvidia's sales model of bundling GPUs with some other products, and instead require more autonomy in their choices. "If Nvidia's products lack flexibility, then deliveries will also be delayed due to the delays from system integrators," said Sheng Linghai.

Although the four tech giants, including Meta, Microsoft, Amazon, and Google, are expected to invest hundreds of billions of dollars in AI infrastructure by 2025, these companies account for about half of NVIDIA's data center business revenue. However, they are now also independently developing AI chips to reduce their reliance on NVIDIA's expensive chips, which to some extent puts NVIDIA in competition with its largest customers.

In addition to the need for further diversification of its product sales model, NVIDIA also faces a challenge regarding the significant hurdles in the practical application of AI in real-world scenarios. Although it is widely believed that artificial intelligence has disrupted everything, most of the economic impact of generative artificial intelligence so far has come from infrastructure development, and truly transformative applications have largely yet to be realized.

Not content to be the "shovel seller".

Currently, the applications for NVIDIA's industrial metaverse platform Omniverse and the physical AI large model Cosmos are still in the early stages. Jen-Hsun Huang's prediction about the imminent arrival of self-driving cars ten years ago has also not come true.

The economist from the Massachusetts Institute of Technology, 2024 Nobel Prize winner Daron Acemoglu, estimated in a paper published last year that artificial intelligence will only increase total factor productivity (a key measure of economic efficiency) by about 0.5% over the next ten years. "I have no doubt about the capabilities of generative artificial intelligence, but how to apply it in the business sector remains an open question," he said. "Currently, companies are under pressure to use artificial intelligence because there has been too much hype."

Venture capital firm New Enterprise Associates has co-invested in several artificial intelligence startups with Nvidia. The firm's partner Aaron Jacobson stated, "We need to provide real applications for artificial intelligence. Nvidia is not content with just being the shovel seller; it needs to take miners to the mine and point out to people from the first scoop of dirt, 'Look! There's gold over there!'"

In this context, Jen-Hsun Huang and executives from Nvidia are trying to find the next AI application industry with disruptive potential, eagerly hoping to discover where the gold lies. Fields such as healthcare and robotics are areas that Nvidia is betting on.

In January this year, Jen-Hsun Huang hosted a party with a group of executives from the healthcare technology industry at the J.P. Morgan Health Conference in San Francisco. According to attendees at the event, about 400 guests filled the luxurious Golden Ballroom of the Fairmont Hotel in San Francisco. As night fell, Jen-Hsun Huang, slightly tipsy after having at least two glasses of red wine, kept joking, "If a robot is better than you at finding drug molecules, what would you name it?" He also told Iqvia CEO Ali Buzbee that the name of his company looked like "you fell asleep in front of the keyboard and accidentally sent out those letters you bumped into."

Kimberly Powell, Vice President of NVIDIA Healthcare, emphasized at the J.P. Morgan Healthcare Conference that artificial intelligence models will accelerate the development of new drugs. She described a scenario of a virtual hospital to the attendees, where robots will "watch" surgeries and annotate surgical videos.

"The healthcare system will face a series of challenges such as population aging, rising supply costs, and labor shortages, and Nvidia can directly help hospitals operate more effectively using artificial intelligence." She pointed out, "Nvidia has already collaborated with hospitals to develop an AI agent that can answer patients' pre-operative questions, and the chatbot can ensure that all important questions are answered before surgery, thereby reducing costs for healthcare providers. This is a whole new way of thinking."

(Source: Yicai Global)

Source: East Money

Author: Yicai