Mark Zuckerberg talks Llama 4, AI investment and AI glasses

Meta CEO Mark Zuckerberg shared insights into the company's AI strategy for 2025 during his latest quarterly report, focusing on the new Llama 4 language model and expanding the Meta AI assistant.

The smaller version of Meta's new language model Llama 4 has completed pre-training, with reasoning models and larger variants showing promise, Zuckerberg says. Sources say the company is closely watching recent model advances from Deepseek with V3 and R1.

Llama 4 is designed as a native multimodal model - or "omni-model" - with built-in agency capabilities to enable new use cases. While Llama 3 aimed to compete with closed models, Meta wants Llama 4 to lead in the open-source space. The company plans to share more details in the coming months.

Meta AI targets broad adoption

Zuckerberg also expects Meta AI Assistant to reach over a billion people this year, claiming it's already the most widely used AI assistant. The company focuses on personalization: "We believe that people don't all want to use the same AI -- people want their AI to be personalized to their context, their interests, their personality, their culture, and how they think about the world," Zuckerberg says.

The Ray-Ban Meta AI glasses are also gaining traction, according to Zuckerberg. He expects 2025 will show whether AI glasses can establish themselves as a new product category. "This will be a defining year that determines if we're on a path towards many hundreds of millions and eventually billions of AI glasses," the Meta CEO says, seeing them as the next computing platform and ideal format for many AI applications.

Major AI infrastructure expansion

Meta plans significant investments in AI infrastructure. According to Zuckerberg, the company will deploy nearly one gigawatt of computing capacity this year. Additionally, Meta is building an AI data center with two gigawatts of power that would cover a substantial portion of Manhattan.

For 2025, Zuckerberg anticipates another milestone: developing an AI engineering agent with the programming and problem-solving abilities of a skilled mid-level engineer. He believes the first company to develop this technology will gain a crucial advantage in advancing AI research.