Author: Zhou Yuan / Wall Street News
NVIDIA also finds it hard to resist the charm of ASIC (Application-Specific Integrated Circuit).
Following NVIDIA CEO Jensen Huang's confirmation in June 2024 that Nvidia will begin ASIC chip design, recent industry news has revealed that NVIDIA has selected Taiwan as its ASIC design and R&amp;D center and has initiated a more aggressive talent acquisition effort.
This move has put companies like MediaTek and WorldChip-KY on high alert to prevent NVIDIA from poaching their talent.
WorldChip-KY, or WorldChip Electronics Co., Ltd., is a global leader in ASIC design without a foundry, established in 2003, and listed on the Taiwan Stock Exchange on October 28, 2014, with the stock code &#34;3661&#34;.
Currently, NVIDIA's GB200 mass production is facing difficulties. According to the established plan, the GB300 is set to launch in the third quarter of 2025, making the GB200 likely a transitional product, which raises market concerns about NVIDIA's performance; NVIDIA's competitors are also making efforts to &#34;de-NVIDIA&#34; by implementing AI training using ASICs.
For example, Apple announced the first preview of iPhone AI in July 2024, followed by a paper stating that the AI model was trained on Google's TPU (Tensor Processing Unit). Recently, Apple also announced at the Amazon AWS Reinvent conference that it will use Amazon's self-developed ASIC AI chips for model training.
TPU is also a category of ASIC.
Therefore, with its own general-purpose AI accelerator card GB200 facing production issues, and competitors like Broadcom, as well as Amazon and Google's self-developed ASICs &#34;poaching&#34; NVIDIA's market, Jensen Huang is increasingly restless.
Recent market news indicates that NVIDIA's ASIC layout is accelerating. In Taiwan, NVIDIA continues to recruit senior engineers in the ASIC field, including talents in front-end design verification, IP integration, and physical layer design, with engineers being poached one after another.
In fact, whether it is Amazon, Google, Meta, or even Microsoft's Maia or Cobalt projects, there are Taiwanese ASIC chip design engineers involved in guiding these efforts.
Thus, this group of individuals has become NVIDIA's target for poaching.
NVIDIA offers RSUs (Restricted Stock Units), and from 2023 to 2024, NVIDIA's stock has increased by 239%+ and 180%+, respectively. Engineers who are poached typically receive RSUs, and the stock appreciation profits far exceed their salaries/annual income.
With the GenAI (Generative AI) technology brought by ChatGPT, NVIDIA has transformed from a leader in gaming chips to the &#34;computing power master&#34; of GenAI.
This high level of market dominance has stirred a sense of &#34;caution in peace&#34; and &#34;unity against a common enemy&#34; among global tech giants, prompting these companies, including cloud service providers (CSPs), to seek sources of computing power for AI training beyond general-purpose AI accelerator cards.
Clearly, these companies are looking towards ASICs, which forms a solid foundation for Broadcom's impressive profit performance in ASIC chips announced on December 12, 2024
This market performance and collective behavior cannot be ignored by NVIDIA.
Jensen Huang previously revealed his personal thoughts on ASIC business: he believes that NVIDIA's push into ASIC business will further expand its customer base. Although CSPs are seeking to break free from the awkward situation of relying solely on NVIDIA, and are even engaging in ASIC design themselves, if NVIDIA participates in ASIC design, CSPs will still become NVIDIA customers through ASIC services.
Does this logic seem a bit convoluted? Jensen Huang is aware of this, so he provided three explanations.
Overall, Jensen Huang believes that whether it is general AI accelerator cards or ASICs, the systematic capabilities of NVIDIA bring low costs and a rich ecosystem, which is a clear advantage compared to CSPs directly entering ASIC design.
Due to these two advantages, Jensen Huang is very confident in NVIDIA's establishment of an ASIC department.
Of course, there are differing opinions in the industry; some believe that general AI accelerator cards and ASICs are not actually opposing or substitutive, but rather complementary forms that support each other.
General AI accelerator cards (GPUs) have numerous processing cores and excel at large-scale parallel computing, especially suitable for deep learning. Their production costs are relatively low, and development time is short, making them suitable for diverse market demands.
What about ASICs? These are custom-designed for specific tasks such as AI inference, requiring extensive preliminary research, design, validation, and production, thus having higher initial costs and longer development cycles.
Google and Amazon began ASIC development in 2013 and 2015, respectively, while Microsoft and Meta started relatively later, in 2019 and 2020.
In fact, before Broadcom released its stunning financial report for fiscal year 2024, the industry did not realize the computational support value of ASICs for GenAI technology, and the stock price increases of related stocks in the secondary market were essentially a realization of expectation differences: the greater the difference, the more astonishing the stock price increase.
As a result, Broadcom achieved an astonishing increase of 24.51% the day after releasing its financial report, propelling its market value to over $1 trillion.
Broadcom is confident in the shipment volume of custom AI accelerator cards (ASICs), expecting that the shipment volume in the first quarter of fiscal year 2025 will double compared to the previous quarter.
In the context of the A-share market, direct ASIC designers have little advantage, but in the field of active cables supported by short-distance data transmission in data centers (IDCs) composed of ASICs, this direct mapping relationship will remain effective now and in the future

NVDA

Jensen Huang also came to boost the copper cable connection business?

- Nvidia is accelerating its ASIC chip design efforts, establishing Taiwan as a development center and recruiting talent from competitors.  
- This move raises concerns among companies like MediaTek and SiFive, as Nvidia faces production challenges with its GB200 chip and competition from firms like Amazon and Google, which are also developing their own ASICs.  
- Nvidia's CEO believes that expanding into ASICs will enhance their customer base, despite differing opinions on the relationship between general AI accelerators and ASICs in the market.

Track Hyper | NVIDIA ASIC layout accelerates