Nvidia’s position as a leader in artificial intelligence (AI) chip technology is being challenged by two new players: Broadcom and Marvell. The companies are developing custom-designed semiconductors, or Application-Specific Integrated Circuits (ASICs), that can handle specific tasks more efficiently than Nvidia’s graphics processors.
Hyperscalers, large cloud computing companies, are looking to reduce costs associated with building massive AI data centers. ASICs have been around since the 1960s and offer greater efficiency than general-purpose processors like GPUs. Broadcom and Marvell are well-positioned to capture a significant share of this growing market, which could reach $50 billion over the next five years.
Nvidia’s CUDA software layer allows developers to program its GPUs for specific needs, but Broadcom is developing custom-designed high-performance accelerator chips that can be used in massive clusters. The company projects a serviceable market size of $60 billion to $90 billion in 2027, with networking chips expected to grow to 15% to 20% of its total AI silicon content.
Marvell has also announced a five-year deal to provide Amazon.com Inc. with custom ASICs for its AI applications. The company believes that it can capture a significant share of this market due to its intellectual property, roadmap, and capabilities.
Investors have taken notice of the growing competition in the AI chip market, with Nvidia’s shares falling 11% over the past month while Broadcom has soared 45%. However, both companies have other business segments that are not growing as fast as their AI investments. The possibility of customers slowing down massive data center builds remains a concern, but for now, Broadcom and Marvell appear to be well-positioned to challenge Nvidia’s dominance in the AI chip market.
Source: https://www.marketwatch.com/story/broadcom-and-marvell-are-a-growing-threat-to-nvidia-as-ai-chip-needs-evolve-defd7ba1