NVIDIA has taken a major step in the evolving AI semiconductor market by creating a department focused on ASICs (Application-Specific Integrated Circuits). This move, reported on January 2 by Taiwan’s Commercial Times, shows NVIDIA’s strategy to strengthen its custom chip design capabilities. Already a leader in AI and GPU markets, NVIDIA aims to stay ahead in the competitive landscape.
In June 2023, NVIDIA’s CEO, Jensen Huang, announced plans to open a research and development (R&D) center in Taiwan within five years, hiring 1,000 engineers. The new ASIC department aligns with this plan, and NVIDIA is recruiting top talent from companies like Taiwan’s MediaTek to boost its expertise.
The global AI boom, driven by the rise of generative AI like OpenAI’s ChatGPT, has increased the demand for NVIDIA’s GPUs. While GPUs excel in AI training tasks, the industry is shifting towards inference AI chips, which apply trained AI models to real-world tasks more efficiently.
Market research firm Omdia predicts the inference AI chip market will grow from $6 billion in 2023 to $143 billion by 2030. This growth has pushed tech companies to invest in custom ASICs tailored to specific needs. For example, Google introduced its AI chip “Trillium” last year, designed for both training and inference tasks.
The memory market is also advancing, with Samsung and SK Hynix developing the 6th generation High-Bandwidth Memory (HBM4) for mass production by late 2025. These chips will handle the large data loads required for AI applications.
NVIDIA’s new ASIC department highlights its commitment to innovation. By focusing on custom chip design and leveraging local talent, the company aims to solidify its position as a leader in the AI semiconductor industry.