Nvidia (NVDA -0.06%) dominates the artificial intelligence (AI) graphics card market, with an estimated market share of 92%. This is why the company has grown at an incredible rate of 230% over the past year.
But the company faces potential threats from some tech giants. Let’s examine the threat before seeing how Nvidia plans to deal with it.
Custom AI chips pose a threat to Nvidia
On the other hand, the following companies microsoft, Amazon, meta platformand alphabet It’s no secret that Nvidia customers, companies that have spent billions of dollars on graphics processing units (GPUs), are developing custom AI chips to reduce their dependence on graphics card specialists. is.
For example, Alphabet has deployed custom AI accelerators known as tensor processing units (TPUs) on Google Cloud to “cost-effectively scale a wide range of AI workloads across training, fine-tuning, and inference.” Similarly, Meta Platforms is expected to introduce new custom AI chips this year to reduce its dependence on Nvidia.
Microsoft is building a custom AI chip for deployment in Azure data centers, which will be brought to market this year. Meanwhile, Amazon announced its own chip for training AI models in November, making it available to Amazon Web Services (AWS) cloud customers.
These tech giants are developing their own chips for two reasons. First, Nvidia has not been able to meet the huge demand for his AI GPUs. The wait period for his H100 AI graphics card, the company’s flagship product, could reportedly be up to a year.
Nvidia is doing its best to increase the supply of graphics cards with the help of its foundry partners, but customers may not be comfortable waiting so long to get their hands on these chips.
Second, Nvidia’s AI GPUs are very expensive. The H100 processor will reportedly cost between $30,000 and $40,000. However, investment banking companies raymond james Nvidia estimates that it costs just over $3,300 to manufacture one H100 GPU, demonstrating the immense pricing power the company enjoys in this market.
It’s no surprise, then, that the tech giant is looking to cut such huge expenses by developing custom chips in-house to tackle specific AI workloads where the H100 isn’t needed.
Officially known as application-specific integrated circuits (ASICs), these custom chips are completely specialized in performing specific operations quickly while increasing energy efficiency. Semiconductor research group SemiAnaracy reportedly estimates that successfully developing custom AI chips could save Nvidia customers hundreds of millions of dollars.
All of this explains why Nvidia is considering entering the custom AI chip market.
Nvidia won’t want to let go of this potential $55 billion revenue opportunity.
broadcom (AVGO -1.55%) and marvel technology (MRVL -1.70%) are two major manufacturers of ASICs, and both are seeing a surge in AI-related orders. Marvell, for example, could generate $1 billion in revenue this fiscal year from selling custom AI chips. Broadcom, meanwhile, is expected to sell $8 billion to $9 billion worth of custom AI chips in 2024, by one estimate. Together, these two companies control a 47% share of the ASIC market.
Investment banking firm Needham estimates the total custom chip market was worth an estimated $30 billion last year. Sales of high-end custom ASICs were reportedly between $13 billion and $18 billion last year, and AI is already a significant part of this space. morgan stanley predicts that ASICs could account for 30% of the $182 billion AI chip market by 2027, noting a $55 billion potential revenue opportunity in this space. .
A former Marvell executive is leading Nvidia’s custom chip division, and the GPU specialist is already in talks with Amazon, Microsoft, Meta, OpenAI and Google to build custom chips, according to a February 9 exclusive from Reuters. It is said that there is
If a Reuters report that Nvidia is entering the custom AI chip space is true, investors will have another solid reason to buy this fast-growing AI stock. The company currently trades at an attractive forward P/E ratio of 35x, which is lower than its average forward price/earnings ratio of 42x over the past five years.
Randi Zuckerberg is a former head of market development and spokesperson at Facebook, sister of Meta Platforms CEO Mark Zuckerberg, and a member of the Motley Fool’s board of directors. John Mackey, former CEO of Amazon subsidiary Whole Foods Market, is a member of the Motley Fool’s board of directors. Alphabet executive Suzanne Frye is a member of The Motley Fool’s board of directors. Harsh Chauhan has no position in any stocks mentioned. The Motley Fool has positions in and recommends Alphabet, Amazon, Meta Platforms, Microsoft, and Nvidia. The Motley Fool recommends Broadcom and Marvell Technology and recommends the following options: His long January 2026 $395 call on Microsoft and his short January 2026 $405 call on Microsoft. The Motley Fool has a disclosure policy.
