On the evening of January 31, NVIDIA CEO Jensen Huang appeared at a nostalgic restaurant in Taipei, Taiwan, to host a dinner for 35 executives from supply chain partners collaborating with NVIDIA. However, nearly 40 executives from Taiwanese companies attended, with only one being from a mainland enterprise.
According to Taiwan's Economic Daily, during the group photo session of the banquet, the first row was specially arranged with seats, and sitting in the first row with Jensen Huang were ASUS Chairman Jonney Shih, MediaTek Vice Chairman Cai Lixing, TSMC Chairman Wei Zhejia, Quanta Chairman Barry Lam and Vice Chairman Liang Cizhen, Wistron Chairman Lin Hsien-Ming, which the industry humorously referred to as giving priority seating to the 70-year-old or white-haired, respected supply chain giants.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/7dce7dbc-e6c2-47d3-8c4b-ee7f44d5304b.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/7dce7dbc-e6c2-47d3-8c4b-ee7f44d5304b.jpeg"/>
The second row included Inventec Chairman Ye Licheng, ASE Chairman Cai Qiwen, Foxconn Chairman Liu Yangwei, Pegatron Chairman Tong Zixian, Wistron General Manager Lin Jianxun, Gigabyte General Manager Lin Yingyu and Li Yitai, Wistron Chairman Hong Lining, Qihong Chairman Shen Qingxing, Quanta Senior Vice President Yang Qiling, MSI Chairman Xu Xiang, Compal Chairman Chen Ruicong, Inventec General Manager Cai Zhi'an, Foxconn Industrial Internet Chairman Zheng Menghong, Lite-On General Manager Qiu Senbin, Delta Electronics Chairman Zheng Ping, Pegatron Co-CEO Zheng Guangzhi and Deng Guoyan.
Jingyuan Electronics General Manager Zhang Gaoxun, ASRock General Manager Xu Longlun, Yongqing General Manager Sha Weixu, Acer Chairman Chen Junsheng, ASUS Co-CEO Xu Xianyue and Hu Shubin, Advantech General Manager Zhang Jiahao, Hongbai Technology Chairman Zhou Taiyu and General Manager Ding Zhaobang. Additionally, Chen Tao, Chairman of mainland PCB supply chain partner VGT, also attended, showing the importance placed on NVIDIA.
Advantech was represented by General Manager Zhang Jiahao this year, after Chairman Liu Kezhen attended last year. He stated that they have been collaborating with NVIDIA for over 20 years, from industrial computers to now Edge AI, and in the future, they will implement Physical AI across various industries, with considerable cooperation expected. For example, in robotics or medical fields, the third area is Edge Servers, as many industries hope to localize computing power, which will become a significant business opportunity in the future. Advantech will officially transform from an industrial computer leader to an Edge AI company, integrating software and hardware to perfectly implement NVIDIA's computing power into factory automation, smart medical, and robotics verticals.
The only two women present were Wistron Chairman Hong Lining and another being Liu Yingjun, CEO of the graphics card giant Tongde Co., Ltd., who was also invited last time.
Before the dinner commenced, Jensen Huang stood on a chair to give a speech, saying, “Welcome everyone to be here! This cannot be called an annual gathering, but we have worked so hard and so diligently together, so we should gather every six months for a drink.”
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/3bbd33a5-2ef4-4db0-8e2f-94abd6befc85.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/3bbd33a5-2ef4-4db0-8e2f-94abd6befc85.jpeg"/>
Jensen Huang pointed out, &#34;2025 will be a very challenging year as we begin production of Grace Blackwell. Looking back, compared to Grace Blackwell, Hopper was simply too simple. The semiconductor system of Hopper was the most advanced at the time, but Grace Blackwell pushes advanced technology to the limit, and the system is quite difficult. Over the past year, we have challenged the limits together, and everyone here has accomplished an incredible task.&#34;
Grace Blackwell faced many difficulties during mass production, leading to forced design modifications that caused chaos in the supply chain. Jensen Huang admitted that this year's operational model will be different from the past. He is very satisfied with the achievements of the past year, which have indeed been challenging, and he thanked everyone for their collaboration. He sincerely wanted to say thank you and also apologize. When he said sorry, the key supply chain representatives who accompanied NVIDIA in changing the design smiled knowingly.
Currently, the GB300 cabinet has entered the early stage of mass production. Jensen Huang also stated that the GB200 is NVIDIA's second-generation product, and mass production is going very smoothly. As for Vera Rubin, the third-generation product, he hopes mass production will become very simple. The supply chain is running faster than ever, and he thanked the partnership, saying he believes tonight is also a night of great interest for all of Taiwan.
Regarding changes in the AI industry, Jensen Huang observed significant changes, stating that AI has become useful, and large language models have become very useful for the industry. Therefore, tokens can now be profitable. He explained that tokens generated in 2024 are not as smart; when AI is not smart enough, it cannot generate profits. Now that AI has become smarter, profit models can emerge.
According to the Commercial Times, Jensen Huang also responded to key external issues such as the shortage of storage chips, AI demand, participation in OpenAI's new round of financing, and challenges with AI ASICs during a media interview.
Jensen Huang stated that 2026 will be an &#34;extremely tight year&#34; for the AI industry, whether in high-performance computing or low-power applications. &#34;For AI to be intelligent, it must have storage.&#34; This year, the demand for high-bandwidth memory (HBM) and LPDDR will surge significantly, putting unprecedented pressure on the overall supply chain, but it will also usher in &#34;a very good year.&#34;
Jensen Huang pointed out that although global semiconductor supply has grown at about double the rate each year, AI demand is growing even faster, leading to extreme tension in every aspect from chip manufacturing, packaging to storage this year. He emphasized that NVIDIA has fully mass-produced the Grace Blackwell architecture and is simultaneously launching the next-generation Vera Rubin platform. &#34;Vera Rubin consists of six of the world's most advanced chips,&#34; with a high level of complexity in process and integration, posing significant challenges for wafer foundries and advanced packaging.
Jensen Huang also revealed that NVIDIA will participate in OpenAI's next round of financing, and the amount may be the largest strategic investment in NVIDIA's history. He described OpenAI as &#34;one of the most influential companies of this era,&#34; and NVIDIA will continue to increase its funding and computing power support
In response to concerns that AI-specific chips (ASICs) will replace GPUs, Jensen Huang firmly stated that it is &#34;unreasonable and unrealistic,&#34; because NVIDIA does not only produce a single chip but builds the entire AI infrastructure, covering CPUs, GPUs, network chips, switches, and data processing systems, and collaborates with all major AI players globally, including Google, OpenAI, Anthropic, and xAI. &#34;This scale and intensity of research and development cannot be matched by a single ASIC team.&#34;
When discussing the role of the Taiwan supply chain, Jensen Huang candidly remarked, &#34;Without Taiwan, NVIDIA would not exist.&#34; He praised the technical strength and engineering culture of Taiwanese companies, specifically highlighting TSMC's critical position in advanced processes. He anticipates that TSMC's capacity will &#34;grow exponentially&#34; over the next decade, not just to serve NVIDIA but as one of the largest technological infrastructure expansions in human history.
Jensen Huang also revealed that NVIDIA's annual R&amp;D budget has reached $20 billion, and it will continue to grow at about 50% per year, from Hopper to Blackwell, and then to Rubin. &#34;The technical difficulty has shifted from difficult to impossible,&#34; but because of this, NVIDIA must continue to invest at a high speed to maintain its leading position in the global AI arms race.
At the end of the dinner, Jensen Huang once again expressed gratitude to the supply chain attendees, stating that this year is going to be &#34;very big.&#34; In other words, the supply chain will work very hard, but the excellent conditions of the Taiwan supply chain are unique.
Source: Chip Intelligence
Risk Warning and Disclaimer
The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this is at their own risk.

002453

300476

NVIDIA CEO Jensen Huang held a banquet in Taipei, inviting nearly 40 executives from Taiwanese companies to showcase the importance of collaboration with the supply chain. Attendees included senior executives from companies such as ASUS, TSMC, and MediaTek, emphasizing the long-term partnership with NVIDIA. Advantech stated that it will transform into an edge AI company and plans to deepen cooperation with NVIDIA in areas such as industrial automation and smart healthcare

Jensen Huang's "Night Banquet" in Taipei: Gathering nearly 40 executives from Taiwanese enterprises, with the chairman of VGT in attendance