Per editor: Du Yu
On May 25, local time, the three major stock indexes of U.S. stocks were mixed, and as of the close, the Dow reported 32764.65 points, down 0.11%; The S&P 500 index was at 4151.28 points, up 0.88%; The Nasdaq closed its command at 12,698.09, up 1.71%.
NVIDIA jumps 24%, leading the technology sector to "carnival alone"
Big tech stocks rose widely, with Nvidia (NVDA) closing up 24.37% at $379.8 and a market value of $939.29 billion as of Thursday's close, with shares hitting a record high. Its market value soared by $184 billion overnight.
It is worth noting that Intel's latest market value is $114.29 billion, which means that NVIDIA rose nearly "two Intels" in one day.
While NVIDIA soared, AMD and TSMC also rose by more than 10%, and Microsoft and Google also rose at the same time. Tesla and Apple rose slightly.
Data center revenue reached a record $4.28 billion
After the US stock market on Wednesday, May 24, Eastern time, NVIDIA announced that in the first fiscal quarter of fiscal year 2024 ended April 30, 2023 in the Gregorian calendar: first-quarter revenue of $7.19 billion, analysts expected $6.52 billion; expected second-quarter revenue of $11 billion fluctuated by 2%. Analysts expected $7.18 billion; Q1 adjusted earnings per share were $1.09 versus $0.92; first-quarter gaming revenue was $2.24 billion versus analysts' expectations of $1.98 billion; first-quarter data center revenue was $4.28 billion versus analysts' expectations of $3.91 billion; and first-quarter gross margin was 66.8% vs. analysts' expectations of 66.6%.
Image source: NVIDIA official website
As another of the two pillars of NVIDIA's revenue, the game business reported revenue of $2.24 billion in the first quarter, down 38% year-on-year, but up 22% from the previous quarter. In the automotive business, NVIDIA recorded revenue of $296 million in the first quarter, up 1% sequentially and also a record high.
During the earnings call, Nvidia told analysts that many cloud companies are competing to deploy AI chips. Demand for graphics processing unit (GPU) terminals in personal computers (PCs) was "solid" in the first quarter. Growth is coming from the data center business. The company has locked in a significant increase in data center chips and plans to significantly increase supply in the second half of the year.
As a one-stop solution provider, NVIDIA provides GPUs that can help AI products train large amounts of text, images and videos, and holds the "life gate" of supplying computing power in the AI large model competition, thus becoming a big winner in this year's AI concept boom.
It is worth noting that NVIDIA's share price has more than doubled since the beginning of the year.
The industry expects that with the growing demand for training of artificial intelligence large models, it will also promote NVIDIA's AI chip sales.
Nvidia's AI chips have been speculated to more than $40,000
According to Interface News reported on April 17, as the computing power basis of the entire AI tide, NVIDIA's A100 and H100 GPUs (graphics processing units) have become sought-after goods, especially the price of the new flagship H100 overseas has also been hyped.
John Carmack, CEO of Keen Technologies and a well-known game developer, was the first to notice this phenomenon, saying on Twitter that the Nvidia AI flagship chip H100 sold for more than $40,000 in multiple stores. At that time, there were at least 6 stores selling H100 on the overseas e-commerce platform eBay. Generally speaking, NVIDIA's A100, H100 and other AI chips are mainly shipped by enterprises, and are rarely retailed in the open market, and there is some room for speculation.
Image source: Visual China
In the context of the surge in demand for computing power, NVIDIA's GPUs have almost become the only "hard currency". Musk also said earlier: "It seems that everyone around me is buying GPUs. Musk himself was exposed to Nvidia buying thousands of GPU chips that provide computing power support for artificial intelligence, and this news also pushed Nvidia's stock price up.
Developers are using the H100 chip to build more powerful Large Language Models (LLMs). Training AI models, especially large models like GPT, often requires hundreds of high-end NVIDIA AI chips to work together. For example, to build ChatGPT, Microsoft spent hundreds of millions of dollars to buy tens of thousands of NVIDIA A100 chips.
Earlier this year, Nvidia also announced a new service that allows some businesses to rent DGX, a supercomputer powered by its AI chip A100, for $37,000 a month.
Nvidia said the H100 is the first chip optimized for a specific AI architecture that underpins many of the latest advances in AI (transformers). Industry experts say that building models that are larger and more data-intensive than those currently available will require more powerful chips.
Daily economic news comprehensive NVIDIA official website, interface news, public information
Daily economic news