(San Francisco = Yonhap News) Correspondent Kwon Young-jeon = Analysis continues that other semiconductor companies in the artificial intelligence (AI) field, such as Samsung Electronics and AMD, will be beneficiaries of the crisis at Nvidia triggered by the resignation of former CEO Jensen Hwang.
Bloomberg News reported on the 12th (local time) that analysts predict that the diversification of the supply chain across the AI industry will bring about such changes.
“There is no doubt that the aggressiveness Nvidia has shown over the past two to three years in setting and exceeding goals will disappear,” said Mandeep Singh, head of technology research at Bloomberg Intelligence.
This is based on the prediction that not only interim CEO Colette Kress, who is in charge of finance, but also the next official head of Nvidia will take a fairly conservative approach to management at the beginning of their inauguration.
In this situation, analysts believe that if Nvidia reorganizes the supply chain of components such as memory semiconductors, it could be a strategic opportunity for Samsung Electronics.
Jeong-in Yoon, CEO of Fibonacci Asset Management, told Bloomberg, “After Jensen Huang steps down, the new management may reevaluate supplier relationships and place greater emphasis on supply chain diversification,” adding, “It will provide Samsung with an opportunity to deepen its partnership with NVIDIA.”
In the AI accelerator sector, AMD, which has the widest competitive front with NVIDIA, including general-purpose graphics processing units (GPUs) and AI server racks, and Broadcom, a leader in the custom chip (ASIC) and network sectors, were pointed out as likely to enjoy the greatest benefit from NVIDIA's leadership vacuum and confusion.
Dan Ives, a senior analyst at Wedbush Securities, said on CBS, "Jensen Huang was the person that competitors and China feared the most. If he is removed, the playing field between domestic and foreign semiconductor companies will be leveled."
Anna Woo, founder of Floodlines Research, an investment strategy analysis firm, said, "The scandal of former CEO Hwang has raised public suspicion about NVIDIA's trade secrets and product competitiveness." She added, "Direct competitors AMD, Alphabet (Google) in the tensor processing unit (TPU) field, and Broadcom will naturally benefit in the short term."
This is because NVIDIA chip customers are likely to respond by diversifying their supply chains rather than taking the risk of relying on a single company in a chaotic situation.
In fact, on a day when the Nasdaq index fell 0.4% and Nvidia's stock price fell 4.34%, AMD and Broadcom rose 5.54% and 5.86%, respectively.
In particular, AMD's market capitalization, led by CEO Lisa Su, former CEO Hwang's 5th cousin (maternal niece), was $736.5 billion as of the 6th, but surged to $960 billion (about 1280 trillion won) in just four business days. The market capitalization ranking also jumped three places from 19th to 16th.
Broadcom's market capitalization also increased from $1.4 trillion to $1.74 trillion during the same period, and its market capitalization ranking, which was 11th, rose to 8th.
Korean startups such as Furiosa AI, a domestic AI chip manufacturer, and Mango Boost, supported by Samsung Electronics, are also attracting attention in that they are targeting the market dominated by Nvidia through alternative technologies.
Cerebras, a rival to NVIDIA that possesses wafer-scale engine (WSE) technology that turns a huge wafer into a single chip, also saw its stock price surge by more than 17% on the day, citing earnings announcements that exceeded market expectations and a new contract with Microsoft (MS).
Pepperstone analyst Chris Weston said, “The market appears to be distinguishing between Nvidia’s problems and AI’s problems.”
However, many predict that the monopoly structure of the AI semiconductor market will not be shaken in the short term because the platform 'CUDA' ecosystem, which Nvidia has worked hard to build for 20 years, is so strong.
This is because millions of AI developers and companies around the world have been developing large language models (LLMs) and software based on CUDA, and switching to competitor hardware will require enormous costs, effort, and time.
