Why trust Greenbot
Meta CEO Mark Zuckerberg reaffirmed the company’s massive artificial intelligence (AI) investment plans on Wednesday, even as Chinese startup DeepSeek demonstrated similar AI capabilities at a fraction of the cost.
During Meta’s fourth-quarter earnings call, Zuckerberg defended the company’s planned $60-65 billion AI infrastructure spending for 2025. He argues that computing power remains crucial for deployment at scale.
“Investing very heavily in capital expenditure and infrastructure is going to be a strategic advantage over time,” Zuckerberg said.
The company reported $48.39 billion in fourth−quarter revenue, up 21% year-on-year. Zuckerberg also revealed plans to launch Llama 4. However, investor attention was set on Meta’s response to the recent industry panic.
DeepSeek recently unveiled its R1 AI model, which reportedly matches the performance of OpenAI’s flagship systems at just 6 million in training costs—far below the 100 million-plus price tag for similar models. Market concerns sent shares of AI chip maker Nvidia tumbling 17% on fears that companies might need fewer processors.
“It’s probably too early to really have a strong opinion on what this means for the trajectory around infrastructure and CapEx,” Zuckerberg told analysts. “There are a bunch of trends that are happening here all at once.”
Meta’s stock rose 2% after the call, suggesting investors support the company’s aggressive AI strategy.
Zuckerberg explained that while training models might become cheaper, running AI services for Meta’s massive user base still requires colossal computing power. “It’s going to be expensive for us to serve all of these people because we are serving a lot of people,” he said, pointing to Meta’s 3.35 billion daily active users.
The CEO also sees DeepSeek’s emergence as validation of Meta’s open-source AI approach. “There’s going to be an open-source standard globally,” Zuckerberg said. “For our own national advantage, it’s important that it’s an American standard.”
Meta plans to incorporate DeepSeek’s advances into its upcoming Llama 4 AI model. This integration aligns with Zuckerberg’s prediction that Meta’s AI assistant will reach one billion users in 2025. “That’s part of the nature of how this works, whether it’s a Chinese competitor or not, ” he said.
CFO Susan Li reinforced that Meta appears “more insulated” from market reactions to DeepSeek because it doesn’t rely on selling AI models for revenue. Zuckerberg said that Meta has a “strong business model” as an advantage compared to other companies.
Meta’s total expenses are projected to reach between $114 billion and $119 billion in 2025, up from $95 billion in 2024. This reflects the company’s commitment to maintaining its AI infrastructure advantage despite emerging cost-efficient alternatives.
