HomeAI News
Nvidia's profit soared 843%, and its stock price exceeded $500! H100 becomes AI gold, orders are scheduled until next year

Nvidia's profit soared 843%, and its stock price exceeded $500! H100 becomes AI gold, orders are scheduled until next year

Hayo News
Hayo News
August 24th, 2023
View OriginalTranslated by Google
Nvidia announced the most explosive financial report in history, with a profit increase of 843% and a new high in market value. It is expected to continue to grow strongly, and it may become the wave of AI in the future!

As soon as Nvidia announced its second-quarter financial report after the market closed on Wednesday, its stock price rose 6.6% after hours, once crossing the $500 mark, reaching a peak of $517.

If calculated according to 500 US dollars, Nvidia's market value will exceed 1.2 trillion, a new record high!

Nvidia's financial report is released

Nvidia's revenue has hit $13.5 billion since May.

Among them, data center revenue accounted for a record US$10.32 billion in revenue, more than doubling in just one quarter, and Nvidia earned a profit of US$6.188 billion, a year-on-year increase of 843%!

Although the gaming industry's revenue is $1 billion short of its peak during the epidemic, its revenue actually increased by 22% year-on-year to $2.48 billion.

The bigger bright spot is the performance forecast for the third fiscal quarter as of the end of October. Nvidia's median revenue forecast range is $16 billion, which is much higher than analysts' consensus forecast of $12.6 billion.

Almost all the financial data in this financial report have achieved double-digit growth compared with the first quarter, and even achieved several times the growth compared with the same period last year!

H100 Becomes the Most Profitable Commodity in the AI ​​Wave

The most direct factor driving such a strong growth in Nvidia's performance is the H100, an AI computing chip it has just launched a few months ago.

After the wave of generative AI swept the world, AI technology stood on the most central stage of the technology industry.

From major technology giants to AI start-ups with dozens of people, they have bet on AI, making the computing power behind AI technology the most scarce resource in the market.

Nvidia's AI chip H100, which sells for more than $40,000 per chip, has been in short supply as the most powerful AI computing chip on the market.

According to industry insiders, the production of H100 chips in 2023 has already been sold out, and if you pay for an order now, you will not be able to get the goods until at least mid-2024.

And how much the AI ​​industry needs H100GPU, you can feel it from the picture summarized by an industry insider.

According to the content shown in the figure:

  • GPT-4 was probably trained on about 10,000-25,000 A100 sheets
  • Meta about 21000 A100- Tesla about 7000 A100
  • Stability AI about 5000 A100- Falcon-40B trained on 384 A100
  • Inflection used 3500 H100s to train a model comparable to GPT-3.5

In addition, according to Musk, GPT-5 may need 30,000-50,000 H100.

Previously, Morgan Stanley had stated that GPT-5 used 25,000 GPUs and had started training since February, but Sam Altman later clarified that GPT-5 had started training.

And not only technology companies lined up to buy H100, but sovereign countries like Saudi Arabia and the United Arab Emirates bought thousands of H100 GPUs at once.

Moreover, the H100 is not only easy to sell, but also has a frighteningly high profit margin.

Industry experts have said that the profit margin of Nvidia H100 is close to 1000%.

In other words, the cost of Boss Huang's AI chip is almost only 1/10 of the selling price, and customers have to queue up to beg him to buy it, and he is not the most powerful company or person in the world, so he is not qualified to be his customer at all. .

If he doesn't make money, who makes money? !

According to foreign media, Nvidia will directly increase the output of the H100 GPU, which is now difficult to find for one card, from about 500,000 this year to 1.5-2 million in 2024.

It can be expected that Nvidia's revenue and profits will continue to double next year. The question is only how many times it can be doubled.

There is also an even more outrageous thing. Some time ago, Coreweave, a company invested by Nvidia, directly mortgaged its Nvidia GPU to the bank and obtained a loan of more than 2.3 billion US dollars.

The company is said to have gone on to buy more Nvidia GPUs with the cash it borrowed from the bank!

It can be seen that NVIDIA GPU is now equivalent to gold in the AI ​​industry, and NVIDIA is the company that directly produces gold!

Profits are the present, Nvidia's AI dominance is the future

Not long ago, Cheng Yiran, a professor of computer science at Duke University, posted a Weibo, saying that Inflection AI, which had just raised more than 1.3 billion US dollars a few days ago, bought 1.1 billion NVIDIA GPUs after changing hands.

And the current AI industry has become an asset-heavy industry because of its extreme reliance on Nvidia's GPUs!

As the largest financing project in the AI ​​industry in the last month, Inflection AI has also become the company with the second highest financing amount in the AI ​​industry with a total career financing of 1.525 billion US dollars, second only to OpenAI's 11.3 billion.

But this 1.3 billion financing changed hands and became a 1.1 billion Nvidia GPU.

In an industry with unlimited potential like AI, the project with the second largest financing amount must have many institutions chasing the project to grab investment opportunities.

Many venture capital institutions that have been in business for decades may not necessarily have the opportunity to invest.

Nvidia, as one of the investors in this financing, can turn left and right without spending a penny (and even making a lot of money).

You can get a stake in a giant unicorn for nothing just by taking out some of your own company's products.

If the company can grow into an industry giant in the future, the initial investor's income may reach dozens or even hundreds of times.

This business is more profitable than selling graphics cards directly.

Nvidia has transformed itself from a giant in the GPU industry to one of the most powerful investment institutions in the world.

Nvidia has become the most powerful venture capital institution in the AI ​​circle

If OpenAI is not counted, in the first half of 2023, 4 AI unicorn companies with a valuation of more than 1.5 billion US dollars have raised financing.

Nvidia invested in 3 of them!

OpenAI and Anthropic are backed by Google and Microsoft respectively, and both can solve computing power problems through in-depth cooperation with their respective backers.

The other three AI unicorns have all accepted Nvidia's investment.

Because there is Nvidia as a shareholder, at least the computing power issue is not too much to worry about.

According to this trend, almost all AI start-ups that rank well will need to beg Nvidia to participate in the shares, and they can exchange shares directly with H100 without Nvidia giving money.

According to the current industry situation, unless the products of other manufacturers can form a positive competition with Nvidia's GPU, it is possible to break the current pattern.

However, judging from the performance in the MLPerf benchmark, apart from Nvidia's solution, no other company can beat it.

As long as Nvidia does not make big mistakes, it will be difficult for latecomers, let alone catch up, even if they are only close to Nvidia’s dominance.

Someone said before that Nvidia made a lot of money by taking advantage of the AI ​​wave.

Maybe in a few years, Nvidia will become the wave of AI.

Because at that time, Inventec's products will still be the most important basic resource for the development of the entire industry.

Nvidia may still hold most of the shares of the leading companies in the AI ​​​​industry, and more fully share the dividends of the growth of the entire AI industry.

Investors' high valuation of Nvidia is due to the fact that it is continuously creating amazing profits, and more importantly, it also monopolizes the future of the AI ​​industry, and even the entire technology industry.



Reprinted from 新智元 润View Original


no dataCoffee time! Feel free to comment