Mark Zuckerberg, CEO of Meta Platforms Inc. arrives. , Meta Connect event at Menlo Park, California, on September 25, 2024.
David Paul Morris Bloomberg Gety pictures
Dead CEO Mark Zuckerberg said on Wednesday that it is too early to know any effect on the Deepseek model, which will be made by the artificial intelligence model on his company and the broader technology industry.
However, Zuckerberg poured cold water on the idea that META's comprehensive Amnesty spending will decrease soon after Deepseek has proven that advanced artificial intelligence requires much less money and calculation than previously thought.
Zuckerberg said: “It is possible that it is too early to have a strong opinion about what this means to the path about infrastructure and structure,” Zuckerberg said. “There is a set of trends that occur here once.”
the Technology markets They were reeling this week on the Amnesty International model created by Depsik, which is a AI Lab document to Hangzhou Associated with the Chinese hedge box. Dibsic claims that it took only two months and less than 6 million dollars to develop it Great language model R1. This news caused NVIDIA shares tank 17 % on Monday due to the implications of the companies that companies do not need the largest number of chipping maker graphics processing units for their artificial projects.
Zuckerberg Deepseek discussed a call with analysts on Wednesday as part of Meta's Fourth quarter profits. The company stated that sales increased by 21 % year on an annual basis to 48.39 billion dollars, before Wall Street's expectations.
Meta shares increased by approximately 2 % on Monday, indicating that some investors believe that the cost efficiency in Deepseek may mean that the social media company may be able to create strong AI systems for less money. Mita announced last week to plan for He invested between 60 billion dollars and 65 billion dollars This year as part of pushing artificial intelligence. The company said on Wednesday that its total expenditures 2025 will range between 114 billion dollars to 119 billion dollars.
Zuckerberg said Meta is still digesting some Deepseek and his team ultimately hopes to implement some of these developments for its artificial intelligence projects.
“This is part of the nature of how this works, whether it is a Chinese competitor,” Zuckerberg said.
Although it is possible that companies ultimately need less computerized resources to train these strong models, Zuckerberg said that having a ton of server energy may be necessary when the program operates and actually procedures, which leads to a “higher level of intelligence and higher quality Service.
“It will be useful for us to serve all of these people because we serve many people,” said Zuckerberg, whose company said that his company has 3.35 billion people per day in the fourth quarter.
Zuckerberg said that Mita will continue to learn from technologies such as Deepseek's free and free Llama Ai models that can pay the industry as well.
“I still think that investment in Capex and Infra will be a strategic advantage over time,” he said. “It is possible that we will learn otherwise at some point, but I think it is too early to name it, and at this point, I bet that the ability to build this type of infrastructure will be a great advantage.”
Zuckerberg also said that the appearance of Dibsic is valid for a dead commitment to an open source approach to artificial intelligence.
“There will be an open source standard worldwide,” he said. “For our national features, it is important to be an American standard.”
He watches: Dead below and bottom.
