萍股相逢
2025.01.04 13:22

The debate over Nvidia's "budget alternative": Is Marvell more suitable than Broadcom?

portai
I'm LongbridgeAI, I can summarize articles.

$Marvell Tech(MRVL.US) $Broadcom(AVGO.US) $NVIDIA(NVDA.US)

Currently, Broadcom holds about 70% of the market share in the custom AI chip market. However, analysts believe that Marvell, with its dominant position among cloud computing customers, greater growth potential, and more attractive valuation, may gain an edge in future competition.
As competition in the AI chip market intensifies, chip companies seem to be "on edge," and NVIDIA's market dominance is being challenged by the "two major alternatives"—Broadcom and Marvell.
Analysts suggest that although Broadcom currently holds about 70% of the custom AI chip market, Marvell, with its strong position among cloud computing customers, greater growth potential, and more attractive valuation, might outperform Broadcom and pose a stronger challenge to NVIDIA.

NVIDIA's "alternatives" are gaining momentum, and Marvell's advantages are becoming apparent.

Over the past three months, as Broadcom and Marvell actively helped tech giants like Google and Amazon develop custom AI chips, their revenues surged, and both companies have gained investor favor, with their stock prices rising over 30%.
In contrast, NVIDIA, the dominant player in AI chips, has seen slower revenue growth, and its stock performance has been lackluster. Investors seem to be starting to believe that at least Broadcom could eventually break NVIDIA's control over the AI chip market. Raymond James analyst Srini Pajjuri stated:
"Broadcom is currently the leader in manufacturing custom AI chips, with Marvell 'close behind.' They are the only two U.S. companies heavily engaged in this business."

As the "only two" in the U.S., the competition between Broadcom and Marvell has begun.

In early December last year, Broadcom CEO Hock Tan hyped the potential of the custom AI chip market, claiming that the company's top three clients, including Google, could spend $60 billion to $90 billion annually on custom chips over the next two years.
However, analysts believe that although Broadcom currently holds about 70% of the custom AI chip market, its share may gradually be taken by Marvell, potentially dropping to around 50%. Specifically, Marvell has several advantages:
Support from key clients: Marvell has won favor from major clients like Amazon and possibly Microsoft. Broadcom, on the other hand, has alienated clients like Amazon due to its aggressive negotiation tactics.
Favor from cloud service providers: Analysts believe the primary demand for custom AI chips will come from cloud service providers like Amazon, Microsoft, and Google, where Marvell holds a stronger position.
Greater growth potential: Due to its smaller current size, Marvell has more room to grow as the AI chip market expands. By 2028, Marvell's annual AI chip revenue could grow to over $20 billion, with overall revenue quadrupling. Additionally, Marvell's valuation is more attractive.

NVIDIA will maintain its dominance.

Although Broadcom and Marvell are growing rapidly in the custom AI chip space, analysts believe NVIDIA's dominance in the broader AI chip market will be hard to shake in the short term.
Kevin Krewell, principal analyst at tech research firm Tirias, noted that compared to the AI chips developed by Broadcom and Marvell, NVIDIA's GPUs are more versatile, "user-friendly," and support a wider range of workloads. Custom chips remain a "niche" business compared to NVIDIA's general-purpose chips.
Moreover, NVIDIA's technological expertise and ecosystem development in AI chips far outpace its competitors.

The copyright of this article belongs to the original author/organization.

The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.