NVIDIA Expands Beyond GPUs: Custom Chip Division Launched
Currently, nearly all cloud service providers have procured NVIDIA’s A100 and H100 compute cards for artificial intelligence (AI) and high-performance computing (HPC) workloads. Concurrently, cloud service providers, including Amazon, Google, and Microsoft, are deploying their custom-designed chips to optimize performance and cost efficiency, highlighting the market’s demand for diversification.
Reuters indicates that NVIDIA is in the process of establishing a new business division dedicated to designing custom chips for a broad array of applications, including but not limited to AI chips, aiming to expand NVIDIA’s future business prospects. This new division, led by Vice President Dina McKinney, is designed to cater to the needs of various applications such as automotive, gaming consoles, data centers, telecommunications, and others that may benefit from custom chips.
The potential client list includes automobile manufacturers, major cloud service providers, and telecommunications companies. It is rumored that NVIDIA has initiated preliminary discussions with tech giants such as Amazon, Meta, Microsoft, Google, and OpenAI to explore opportunities for creating custom chips, indicating an interest that extends beyond traditional off-the-shelf data center products. Although products like the H100 boast impressive performance, customers are unable to add custom features according to their specific needs, leading to deficiencies in intellectual property protection. Additionally, these products may not fully adapt to their tasks and come with a high price tag.
NVIDIA hopes that this new division will better meet the chip usage needs of major clients and enhance competitiveness in relatively lagging market segments. In markets such as automotive, where NVIDIA’s solutions lag in sales compared to data centers, gaming, and professional visualization, some manufacturers prefer to develop highly customizable platforms themselves, considering cost, competitive dynamics, and control over intellectual property.
