NVIDIA reportedly plans to launch a custom chipmaking unit.

Nvidia is establishing a new business unit to design custom processors for a diverse range of applications, including artificial intelligence, according to Reuters reporting citing nine sources. The unit, led by Vice President Dina McKinney—who previously oversaw AMD's Cat-series CPU microarchitectures, some of Qualcomm's Adreno GPU design, and Marvell's infrastructure processors—is designed to address the needs of automotive, consoles, datacenters, telecom, and other applications that require custom silicon. While Nvidia has not officially confirmed the new unit's existence, McKinney's LinkedIn profile indicates she serves as VP of Silicon Engineering overseeing silicon aimed at "cloud, 5G, gaming, and automotive," reflecting the breadth of her responsibilities.

The move represents Nvidia's strategic response to a rapidly growing market trend. Leading cloud service providers including Amazon Web Services, Google, and Microsoft are already deploying their own custom processors for AI and general-purpose computing workloads. This approach allows them to optimize costs by eliminating premium payments to Nvidia, tailor their datacenters' capabilities, and optimize performance and power consumption, saving substantial sums of money. By controlling silicon design, these companies can add custom capabilities such as new data formats to their chips quickly and protect their intellectual property. While Nvidia's AI and HPC GPUs remain irreplaceable for certain workloads, many deployments now run on custom silicon, a trend that Reuters notes is extensive and growing rapidly—essentially, as the article states, allowing cloud service providers to "eat Nvidia's lunch."

Nvidia has reportedly engaged in preliminary talks with tech giants including Amazon, Meta, Microsoft, Google, and OpenAI to explore opportunities for creating custom chips, signaling an expanded focus beyond traditional off-the-shelf datacenter offerings. The automotive sector represents a particularly significant opportunity. While Nvidia's datacenter, gaming, and professional visualization solutions remain highly profitable, sales of its automotive offerings have lagged. Many automakers are developing custom silicon for software-defined vehicles, preferring their own highly-customized platforms over Nvidia's Drive platform due to cost, competitive, and intellectual property control considerations.

This business unit strategy places Nvidia in direct competition with established custom chip designers including AMD, Alchip, Broadcom, Marvell Technology, and Sondrel. However, Nvidia possesses highly competitive intellectual property across CPU, GPU, AI, HPC, networking, and sensor processing technologies already deployed in its off-the-shelf products—including A100 and H100 processors, RTX-series graphics processors, and Mellanox connectivity and networking products. Packaging these technologies into custom configurations for specialized applications could significantly expand Nvidia's total addressable market and increase its earnings.