Wallstreetcn
2023.08.26 10:36
portai
I'm PortAI, I can summarize articles.

NVIDIA says data center upgrade spending will reach $1 trillion in the next four years, but the question is: Who will foot the bill?

Huang Renxun predicts that there will be approximately $1 trillion in demand for AI data centers in the next four years. Is there really that much?

NVIDIA's explosive earnings report this week has injected a strong stimulant into the technology sector. However, there are also many doubts, with the biggest focus being whether the demand for NVIDIA's "expensive" GPU can truly be sustained.

NVIDIA CEO Jensen Huang stated during a conference call on Wednesday that there will be approximately $1 trillion allocated for AI data center upgrades in the next four years, including GPU upgrades. He even provided specific figures for annual expenses.

He said, "There is approximately $1 trillion worth of global data center demand, with an annual capital expenditure of around $250 billion."

According to Huang, most of the expenses will be covered by leading cloud service providers and other large tech companies, who are actively participating in the competition for generative artificial intelligence.

These giants mainly include companies such as Amazon, Microsoft, Google, and Meta. The latter is actively advancing generative artificial intelligence, including models like Llama2.

So, the question is, how much cash do they have? Is it enough to support NVIDIA?

As of June 30th, Amazon, Microsoft, Google, and Meta had approximately $334 billion in cash and cash equivalents. Amazon holds around $41 billion in cash, while Meta holds around $64 billion. Even if we consider the new cash flow generated each year, it still falls short of NVIDIA's estimated $1 trillion over four years.

This concern was also raised during NVIDIA's conference call on Wednesday.

Bank of America Merrill Lynch analyst Vivek Arya asked Huang, "We see that overall spending by cloud service providers has not increased that much. So, how confident are you that they can gain a larger share of generative artificial intelligence? Please explain to us the sustainability of this demand in the next one to two years."

Huang did not directly answer this question. Here is his response:

"There is approximately $1 trillion worth of global data centers, with an annual capital expenditure of around $250 billion. You will see that data centers worldwide are investing this capital expenditure into the two most important computing trends today, which are accelerated computing and generative artificial intelligence. So, I believe this is not a short-term event. It is a long-term industry transformation, and we are witnessing both of these transformations happening simultaneously."

Finally, let's take a look at the recent comments from Meta, one of NVIDIA's biggest potential customers, regarding this issue.

Meta CFO Susan Li stated that the company expects capital expenditures this year to be between $27 billion and $30 billion. By 2024, this number will increase as the company builds newly designed data centers and invests more in servers to support all of its core AI ranking, recommendation work, and next-generation AI projects.