Nvidia The fourth-quarter financial results are scheduled to be reported on Wednesday after the bell rings.
This is expected to end one of the most outstanding years of a large company ever. Analysts surveyed by FACTSET expect sales to end in January to be $38 billion, up 72% per year.
The January quarter will end its second fiscal year, with NVIDIA’s sales doubled. It’s a breathtaking connection, which is the essential hardware for NVIDIA’s data center graphics processing unit (GPU) that is the key to building and deploying AI services like Openai’s Chatgpt. NVIDIA shares have risen 478% over the past two years, making it the most valuable company in the United States with a market value of over $3 trillion.
However, NVIDIA’s stock has slowed down in recent months as investors question it can start from here.
It trades at the same price as last October, and investors are wary of any signs that Nvidia’s most important customers may have tightened their belts after years of large capital expenditures. This is particularly worrying after the recent AI breakthrough in China.
Most of NVIDIA’s sales are used to build large server farms for a few companies, usually renting them to other companies. These cloud companies are often referred to as “high standards.” Last February, NVIDIA said a client accounted for 19% of its total revenue in fiscal 2024.
Morgan Stanley analysts estimate this month Microsoft In 2025, NVIDIA’s latest AI chip, Blackwell, will account for nearly 35% of spending in 2025. Google at 32.2%, Oracle 7.4% and Amazon 6.2%.
That’s why any indication that Microsoft or its competitors may back off spending plans can shake NVIDIA shares.
Last week, TD Cowen analysts said they learned that Microsoft canceled leases with private data center operators and slowed down its negotiation process to enter new leases and adjusted plans to spend on international data centers. plans to support U.S. facilities.
The report raises concerns about the sustainability of AI infrastructure growth. This may mean that there is less demand for Nvidia chips. Michael Elias of TD Cowen said his team found metrics for Microsoft’s “potential oversupply position.” NVIDIA shares fell 4% on Friday.
Microsoft returned to Malaysia on Monday, saying it still plans to spend $80 billion on infrastructure in 2025.
A spokesperson told CNBC: “While we can do strategic pace or adjust infrastructure in certain regions, we will continue to grow strongly in all regions. This allows us to invest in and allocate resources to growth for the future of the future and allocate resources to growth,” a spokesperson told CNBC. field.”
Last monthmost of NVIDIA’s major clients touted large investments. Letters as target $75 billion In this year’s capital expenditure Yuan Will spend as much as possible $65 billion Amazon’s goal is to flower $100 billion.
Analysts say about half of AI infrastructure capital expenditures ultimately accompany Nvidia. Many large-scale trek into AMD’s GPUs and are developing their own AI chips to reduce their dependence on NVIDIA, but the company occupies most of the market for cutting-edge AI chips.
So far, these chips have been primarily used to train cutting-edge AI models, a process that can cost hundreds of millions of dollars. After AI was developed by companies from OpenAI, Google and Anthropic, the repository of the Nvidia GPUs had to provide these models for customers. That’s why NVIDIA expects its revenue to continue to grow.
Another challenge for NVIDIA is the emergence of Chinese startup DeepSeek last month, which unleashed efficient and effective efficiency.Distillation“AI model. Its performance is high, showing that billions of dollars of NVIDIA GPUs are not needed to train and use cutting-edge AI. Temporarily dropped Nvidia’s stock, causing the company to lose nearly $600 billion in market capitalization.
NVIDIA CEO Jensen Huang will have a chance to explain on Wednesday why AI will continue to require more GPU capacity even after a massive construction last year.
Recently, Huang talked about the “expansion method” observe Starting with Openai in 2020, AI models use more data and computation when creating them.
Huang said DeepSeek’s R1 model points to new wrinkles in what Nvidia calls the scaling method “Test time zoom. Huang has been arguing that the next major way to AI improvement is to apply more GPUs to the process of deploying AI or inference. This allows chatbots to “cause” or generate large amounts of data in the process of thinking through the problem.
AI models have been trained only a few times to create and fine-tune them. But AI models can be called millions of times a month, so using more inference computing will require more deployment to customers’ NVIDIA chips.
Huang said: “The market’s response to R1, ‘Oh my god, AI is done, AI no longer needs to do calculations.” Last week’s interview. “The opposite is true.”
