The AI chip market has shifted towards custom ASIC chips over the past six months, with potential growth rates believed to outpace commercial GPUs significantly. This trend has affected the stock performance of Nvidia (NVDA, Financial) and AMD (AMD), as both have faced challenges in the market. However, Morgan Stanley analysts, led by Joseph Moore, suggest that ASICs may not threaten the long-term dominance of commercial GPUs.
Morgan Stanley's report highlights that ASICs and commercial GPUs serve similar purposes but achieve them differently. While ASICs excel in specific applications, they rely heavily on custom client needs. Despite lower development costs, ASICs may incur higher system and software deployment expenses than scalable GPUs. The mature CUDA ecosystem of Nvidia allows for easy workload deployment, minimizing total ownership costs.
According to Morgan Stanley, Nvidia's leadership position remains robust, barring unexpected changes. While cheaper processors may initially attract clients, they often revert to Nvidia due to its ecosystem maturity and support. The stronghold of Nvidia in the AI chip market comes from its technological prowess and robust ecosystem, reinforced by substantial R&D investments.
This year, Nvidia plans to invest around $16 billion in R&D, compared to the typically under $1 billion spent on ASIC development. These investments enable Nvidia to maintain a 4-5 year development cycle, with continuous high-performance chip releases. Nvidia's global presence across cloud platforms further cements its market dominance, ensuring global support for investments in its ecosystem.
