US markets panicked on Monday for speculation that Deepseek AI models would crush GPU’s demand, NVIDIA’s shares lowering almost 20%.
But Meta does not back down, his CEO Mark Zuckerberg committing that the technology giant would invest “very strongly” in AI – even “hundreds of billions of dollars” – in the long term, he said during Calling Meta’s first quarter results during META’s results on META’s results on META’s results on META’s results on META’s results on META’s results on META results on the results of Meta on Meta’s results. Wednesday.
Zuckerberg already announced last week that Meta would spend more than $ 60 billion in 2025 for capital expenses, mainly in data centers.
In response to the question of an analyst on the impact of Deepseek on the expenses of the Meta AI, Zuckerberg said that the expenses strongly for the infrastructure of IA will continue to be a “strategic advantage” for Meta .
Meta considers that Deepseek is a new competitor and learns, but it is “too early” to say if the demand for fleas will stop increasing because they remain crucial for inference, said Zuckerberg, noting that Meta has billions of ‘Users.
“At this stage, I would bet that the ability to develop this type of infrastructure will be a major advantage for the quality of the service and the possibility of serving the scale we want,” said Zuckerberg.
Meta’s goal with its next model, Llama 4, is to make it the most competitive world in the world, or even compared to closed models (like Chatgpt), said Zuckerberg. He added that he expects him to have agentic capacities – something Openai and Anthropic has moved – as well as multimodal capacities.
“Our goal with Llama 3 was to make the open source competitive with closed models,” he said. “And our goal for Llama 4 is to lead.”
