Meta Overhauls AI Division to Catch Up With Google, OpenAI

Meta is overhauling its AI operations in a major strategic move to close the gap with Google and OpenAI, both of which have taken a commanding lead in the pursuit of artificial general intelligence (AGI).

The restructuring, announced internally and first reported by Axios, involves a split of Meta’s core AI efforts into two focused teams—one building foundational AGI capabilities, and the other applying them directly to Meta’s consumer products.

This shift is more than an internal shuffle; it reflects mounting pressure on Meta to deliver meaningful breakthroughs in AI amid growing dominance from rivals. OpenAI’s GPT-4.5 and Google’s Gemini 1.5 Pro models have raised the bar for what AI assistants can achieve, and Meta is eager to prove its Llama model family can compete at that level.

According to internal communications, Meta’s Chief Product Officer Chris Cox outlined the formation of two new AI units. One team, known as AI Foundations, will concentrate on long-term research and AGI-enabling technologies like reasoning, multimodal understanding, and Meta’s open-source Llama language models.

The other team, AI Products, will focus on integrating AI into Meta’s suite of platforms, including Facebook, Instagram, WhatsApp, and its Meta AI assistant.

This bifurcated structure is designed to streamline development and reduce internal dependencies, allowing the company to move faster and compete more directly with OpenAI’s product rollouts and Google’s integration of Gemini across Search, Docs, and Android.

Meta’s pivot comes amid concerns about its slipping position in the AI race. The company recently lost a number of high-profile researchers, including many of the original architects behind Llama. Some of them reportedly joined French startup Mistral, while others were poached by competitors.

Meanwhile, Meta’s next-gen model, Llama 4, originally expected in early summer 2025, has been delayed. Sources suggest that internal dissatisfaction with model quality and leadership has pushed the release to fall or beyond—giving rivals more time to widen the lead.