The leader of a major US high -tech company at the forefront of artificial intelligence (AI) encourages the popular new model of China's Startup DeepSeek, despite the possibility of overturning the industry.
Openai CEO's Sam Altman, Meta CEO Mark Zuckerberg, and Microsoft's CEO Satya Nadella seemed to be almost indifferent to the recent new AI models after sending Tech Stocks earlier this week.
DeepSeek argues that the new R1 model has the same performance as Openai and is only $ 5.6 million to train. This is a small amount compared to the billions of dollars built an infrastructure to develop AI by a US high -tech company.
Altman acknowledged on Thursday that he did “really great things,” but he neglected the overall impact and called it “very exaggerated.”
“This is a model of the ability level I had long ago,” he said at the Washington DC event, “I felt what I needed to execute a highly cost -efficient model.”
“I don't want to reduce work,” he added. “I think this is a great job. In a sense, Openai is a surprising or new competitors for a long time.”
On the fourth quarter of meta, Zuckerberg was confident in the company's plan to continue investing in AI infrastructure with the advent of DeepSeek.
Facebook and Instagram's parent company Meta announced last week that it plans to spend $ 600 billion to $ 65 billion in 2025 in 2025 in 2025.
“It's probably too early to have a really strong opinion about what this means for infrastructure, maintenance, and such trajectory,” Zuckerberg said. “There are many trends that are happening here.”
He emphasizes that even if DeepSeek's appearance shows that there are few computer infrastructure necessary for training, other factors such as increasing AI use may require this level of investment. I did.
“I keep thinking that I will invest a very large amount of Capex and INFRA.[structure] Zuckerberg added. “If not at a certain point, you may learn, but I think it's too early to call it.”
“And at this point, I think the ability to build such infrastructure will be a great advantage for both the quality of the service and the ability to provide the scale we want.”
Microsoft's Nadella has made a positive tone when discussing DeepSeek in the company's revenue call on Wednesday. Earlier this month, his company announced a plan to spend $ 80 billion of construction data center in 2025.
“I think DeepSeek has some real innovation,” he said. And the big beneficiary of such a software cycle is the customer, right? “
Deepseek became popular and at the top of the Apple's App Store, and some public high -tech jirates were defeated on Monday. NVIDIA, the key to the power to supply power to the AI boom, lost $ 600 billion on Monday alone, one of the most difficult hits.
President Trump said a Chinese AI model was regarded as a “morning call” for high -tech companies in the United States on Monday night, saying that “we must focus on lasers by competing to win.” But he also suggested that it could be a “positive development” for the industry.
“Instead of spending billions of dollars, you will spend a little more expenses, and if you do well, you'll come up with the same solution,” he said at a national dralal resort on the outskirts of Miami, and Republican members are the annual policy retreat. He said he was gathering.
