NVIDIA CEO says that artificial intelligence models are more advanced will maintain a chip, and the growth center growth is continuing

Amnesty International Bubble? What is the artificial intelligence bubble? If you ask the CEO of Nvidia Jensen Huang, we are in a “new industrial revolution”.

Huang Company, of course Amnesty International Gold Rush, has become the largest work in the world by taking advantage of the growth of artificial intelligence, bubble or not. Speaking on Wednesday during a profit call, his company reported about revenues of 46.7 billion dollars in the past quarter, did not indicate any sign that the amazing growth of Trucifier intelligence The industry will slow down.

“I think the next few years, certainly, during the contract, we see great growth opportunities in the future,” Huang said.

Compare that with Recent comments From Openai Sam Altman, who said he believed that investors are currently “excessive about artificial intelligence.” (Al -Taman also acknowledged that he still believes artificial intelligence is “the most important thing that happens in a very long time.”)

Huang said that his company has “very great expectations” of demand for more chips and computers that manage artificial intelligence, indicating that the rush of more data centers does not stop soon. It is speculated that the spending on Amnesty International’s infrastructure may reach $ 3 trillion to 4 trillion dollars at the end of the contract. (The US GDP is about $ 30 trillion.)

This means a lot of data centers, which take a lot of lands Use a large amount of water and energy. These AI factories have become larger and large American electrical network stress. The growth of various artificial intelligence tools that require more energy can make this demand larger.

---

Do not miss any non -biased technology content and laboratory -based reviews. Add cnet As a favorite Google source on Chrome.

---

More powerful and demanding models

No one is always a one -router. The source of increased demand for arithmetic energy is that the latest artificial intelligence models that use “thinking” techniques use a lot of strength for one question. Huang said: “It is called long thinking, and whenever I think often, it produces better answers,” Huang said.

This technology provides the artificial intelligence model to search on different websites, try a question several times to get better answers and put out different information together in one response.

Some artificial intelligence companies offer thinking as a separate model or as an option that bears a sign of something like “deep thinking”. Openai work correctly In GPT-5 versionWith a program to guide whether it has been dealt with through a lighter and direct model or the most intense thinking model.

Huang said that the thinking model can require 100 times of computing power or more than the response of the traditional large language model. These models, along with Agent systems The task models and robots models that can deal with perception and operation in the material world can perform, and the demand for chips, energy and data center data remains high.

“With every generation, the demand grows only,” said Huang.