Photo source: Visual China
The harder the big model battle got, the more fun AI arms dealer Nvidia laughed.
After the market on May 22, local time, Nvidia released its financial results report for the first quarter ended April 28, 2024. According to financial reports, Nvidia's quarterly revenue reached a record high of 26 billion US dollars, up 262% year over year, higher than analysts' expectations of 24.7 billion US dollars. Net profit surged 628% to US$14.88 billion. Non-GAAP diluted earnings per share were $6.12, up 461% year over year.
And this is the 20th consecutive quarter that Nvidia has surpassed analysts' expectations.
Nvidia founder and CEO Wong In-hoon said, “The next industrial revolution has begun. Companies and countries are collaborating with Nvidia to turn traditional data centers worth trillions of dollars into accelerated computing and build a new type of data center — AI factory — to produce a new commodity: artificial intelligence.”
Since Nvidia expects revenue for the second quarter of this year to be around 28 billion US dollars, higher than market expectations, Nvidia's stock price has also risen to a record high. In after-market trading, Nvidia's stock price once rose more than 7%, breaking through $1,000 per share for the first time, reaching a maximum of $1005 per share, which also boosted the rise of other chip and technology stocks.
Meanwhile, Nvidia also announced that it will split its shares by 1 split of 10 and raise the quarterly dividend of 150% to 10 cents per share. This could attract more retail investors and further boost Nvidia's share price.
Since this year, Nvidia's stock price has climbed nearly 92%. The company's market capitalization rose to $2.3 trillion, making it the third-largest company in the world by market capitalization, after Apple and Microsoft.
The data center business remains Nvidia's core revenue source, covering products such as AI chips and a range of software and systems required to run large-scale AI servers. Nvidia is considered to be the biggest beneficiary of the AI wave, accounting for more than 80% of the AI chip market. Although tech giants such as Microsoft, Google, and Amazon are stepping up their own chip development, they are still spending billions of dollars to buy Nvidia chips to train and run big language models.
According to financial reports, Nvidia Data Center's revenue for the first quarter was a record high of US$22.6 billion, a year-on-year increase of 427%, higher than analysts' expectations.
Nvidia's chief financial officer Colette Kress said this was due to the company's Hopper GPU shipments, including the snapped H100 GPU. She pointed out at the performance conference that one of the highlights of Q1 was Meta's announcement of the launch of a new open source big model, the Llama 3 model, which used 24,000 H100 GPUs.
Nvidia has also noticed that a small number of giant customers contribute more sales revenue, and large cloud vendors such as Amazon, Meta, Microsoft, and Google account for about 45% of data center sales. Nvidia also wants to diversify its business and not just rely on top players.
In the earnings report, Hwang In-hoon said that artificial intelligence is expanding to national sovereignty, consumer internet companies, automobile manufacturers, and healthcare customers. These opportunities may “create multiple vertical markets worth billions of dollars” outside of cloud vendors.
Regarding BlackWell, the highly anticipated next-generation AI chip, Hwang In-hoon said that the chip is being “produced at full capacity” and is expected to be launched in data centers in the fourth quarter of this year, which will bring more revenue growth. He also announced that new chips will be launched after Blackwell, speeding up the pace of chip architecture updates every two years to once a year.
Prior to the release of the earnings report, there were media reports that Amazon had suspended orders for Nvidia's Hopper chips to wait for the new, more powerful product, Blackwell, and worried that Nvidia would experience a decline in demand during the two product weeks. Amazon later denied the claim. Hwang In-hoon also responded that demand for Hopper chips has been showing a growing trend, and customers want GPUs to be put into use and data centers immediately, so demand is very strong.