NVIDIA: license requirement for chips…
August 26 Nvidia (“Broad Moat”) has announced that the US government has imposed a new license requirement for the sale of certain data center graphics processing units (“GPUs”), namely the current generation A100 and next generation H100 exported to China. Russia.
It should be noted that Nvidia has not sold products to its customers in Russia since the start of the war in Ukraine.
Among Nvidia’s top cloud customers are Chinese companiesAli Baba, Tencent and Baidu for a variety of applications such as natural language processing, image recognition, and deep recommendation engines.
The new license agreement aims to address risks associated with the military use of Nvidia’s AI-related chips amid rising tensions between the US and China.
In a filing with the SEC, the U.S. stock market watchdog, Nvidia pointed out that its revenue forecast for the fiscal third quarter included about $400 million in potential revenue that could be subject to a new licensing requirement for China.
We believe that many of the end-use applications for Nvidia’s data center GPUs among Chinese customers are consumer-centric, and therefore we believe that the company should be able to license to those customers.
Stock market decline
After this news, the company’s share price fell by about 4% in the post-market.
While the stock is trading at a $200 discount to our fair value estimate, we believe market volatility could create a more attractive entry point.
Notably, Chinese customer demand has been weak in recent quarters due to milder economic conditions related to the COVID-19 lockdown in China.
We reiterate our expectation of at least 40% growth in Nvidia’s data center segment this year, but we will probably revise our growth assumptions for this activity if the company is unable to ship data center processors to Chinese customers for an extended period of time.
We think long-term investors should start to find Nvidia attractive, as its data center business will be more resilient against macro challenges, even if the evolving situation with Chinese cloud customers is worrisome.
Even accounting for weak demand in China, we expect Nvidia’s data center segment to grow as customers consistently every quarter in 2022. Microsoft, Amazon and Alphabet Invest in Nvidia’s A100 GPUs for internal and external workloads such as natural language processing and recommendation engines, and Meta spends heavily to develop the infrastructure needed for its metaverse offering.
The upcoming H100 GPU family (Hopper) is expected to provide performance gains over the A100 family, particularly in natural language and computer vision models, and contribute to significant Q4 revenue.
© Morningstar, 2022 – The information contained herein is for educational purposes and is provided for informational purposes ONLY. It is not intended to be, and should not be construed as, an invitation or inducement to buy or sell listed securities. Any comment is the opinion of its author and should not be taken as an individual recommendation. The information in this document should not be the only source for making an investment decision. Be sure to consult a financial advisor or financial professional before making any investment decisions.
