The US Market fell almost 20 % on Monday, presuming that DeepSeek’s AI models crush the demand for GPUs, and fell almost 20 %.
However, the meta has not retreated. CEO’s Mark Zuckerberg has pledged that a technology giant invests “very large” for a long time and “hundreds of millions” in “hundreds of millions”. on wednesday.
Zuckerberg announced last week that Meta would spend more than $ 600 billion for more than $ 600 billion in capital spending alone in 2025.
Zuckerberg responded to Meta’s analyst questions on Deepseek’s effect on AI spending, and stated that spending on AI infrastructure will continue to be a “strategic advantage” for meta.
META regards DeepSeek as a new competitors and is learning from there, but since it is important for the purpose of inference, it is too early to stop increasing the demand for chips. He pointed out that there are billions of users.
“At this point, the ability to build such infrastructure will be a great advantage for both the quality of the service and the fact that we can provide the scale we want,” Zuckerberg says.
According to Zuckerberg, the following model, LLAMA 4, is the most competitive in the world, compared to closed models (such as Chatgpt). He added that along with the abilities of the multimodal, both open eyes and mankind have the ability to have moved agents.
“The goal of LLAMA 3 was to make the closed model and open source competitive,” he said. “And the goal of Lama 4 is to lead.”
TechCrunch has a newsletter focusing on AI! Sign up here and get it on the reception tray every Wednesday.