Meta CEO, Mark Zuckerberg, isn’t backing down on AI investments. Many thought DeepSeek’s low-cost breakthrough might change Meta’s plans. But that’s not the case. Speaking during Meta’s Q4 earnings call, Zuckerberg reaffirmed the company’s focus on infrastructure and capital expenditure.
DeepSeek’s Disruptive AI Model Shakes the Industry
The AI community was stunned this week by DeepSeek’s announcement. The Hangzhou-based AI lab, backed by Chinese hedge fund High-Flyer, claimed it built its R1 large language model (LLM) in just two months for under $6 million. That’s a fraction of what tech giants like Meta, Google, and OpenAI spend on similar projects.
DeepSeek’s revelation sent shockwaves through the stock market. Nvidia’s shares dropped 17% on Monday. Investors feared companies might no longer need as many high-end GPUs for AI training, reducing demand for Nvidia’s chips.
What Zuckerberg Thinks: AI Costs Go Beyond Training
DeepSeek’s cost-efficient model has sparked debate over AI spending, especially with Trump’s recent $500 billion AI research initiative. But Zuckerberg remains skeptical about its impact on Meta’s strategy.
“It’s too early to tell how this affects infrastructure and CapEx,” he said. “There are many trends happening at once.”
Meta recently announced plans to invest $60 billion to $65 billion in AI infrastructure in 2025. Total expenses could reach $114 billion to $119 billion. Some analysts expected DeepSeek’s efficiency to lead to spending cuts, but Zuckerberg dismissed that idea.
While training AI may get cheaper, the real challenge is deployment. With 3.35 billion daily active users, Meta’s AI needs massive computing power to ensure real-time, high-quality responses.
“It’s expensive to serve billions of people,” Zuckerberg said.
Is Open-Source AI a Competitive Edge?
Despite the competition, Zuckerberg sees an opportunity. Meta has been a leader in open-source AI, recently releasing its Llama models. He suggested that Meta could learn from DeepSeek’s breakthroughs.
“That’s how AI works, whether the competitor is Chinese or not,” he said, stressing shared progress in AI research.
Zuckerberg also believes DeepSeek’s emergence validates Meta’s open-source approach. He argued that setting global open-source standards benefits the AI industry while keeping American companies ahead.
We read all the AI news and test the best tools so you don’t have to. Then we send 30,000+ profesionnals a weekly email showing how to leverage it all to: 📈 Increase their income 🚀 Get more done ⚡ Save time.
“There will be a global open-source standard,” he said. “For our national advantage, it should be an American one.”
Will AI Funding Stay High?
DeepSeek’s model suggests AI spending may shift, but Zuckerberg sees Meta’s investments as a long-term advantage. He acknowledged AI costs might drop but said it’s too soon to make conclusions.
“Heavy investment in CapEx and infrastructure will be a strategic advantage over time,” he said. “Maybe we’ll learn otherwise, but it’s too early to say.”
For now, Meta remains committed to building a powerful AI infrastructure. The company aims to keep leveraging AI innovations at scale.
