Markets in the U.S. were in chaos this Monday. The speculation that DeepSeek’s AI would drastically reduce GPU demand caused Nvidia’s shares to tumble close to 20%.
Nevertheless, Meta isn’t stepping back. As a show of commitment, CEO Mark Zuckerberg announced the company’s plans to invest massively in AI during Meta’s Q1 earnings call on Wednesday. He hinted the investment could be in the “hundreds of billions of dollars” extent in the long run.
Last week, Zuckerberg confirmed that Meta would expend over $60 billion on capital investments in 2025, with the prime focus on data centers.
When probed about the impact of DeepSeek on Meta’s AI expenditure, Zuckerberg emphasized the company’s resolve to continue significant spending on AI infrastructure, viewing it as a strategic advantage.
Though DeepSeek is now seen as a contender by Meta, Zuckerberg stated it’s too premature to ascertain if the demand for chips would cease to rise, given their critical need for inference applications.
“Right now, I’d wager that the ability to delineate that kind of infrastructure will be a significant benefit for service quality and handle the scale we aspire to,” jotted Zuckerberg.
With its upcoming model, Llama 4, Meta aims to outclass other models globally. Zuckerberg hinted at the model having agentic and multimodal capabilities.
“Goal with Llama 3 was to make the open-source at par with closed models”, he said. “And with Llama 4, we intend to lead.”
Original source: Read the full article on TechCrunch
