When one of the most influential minds in artificial intelligence steps away from a tech giant, the move is never quiet. Yann LeCun, Meta’s Chief AI Scientist and one of the three “godfathers of deep learning,” is preparing to leave the company at the end of 2025. For more than a decade, LeCun shaped Meta’s research culture, built the FAIR lab into a respected global institution, and helped cement the company’s reputation as a serious AI powerhouse. His exit is more than a change in leadership, It is a shift in the long-term direction of AI research itself.
Also read: Inside Snapdragon X2 Elite’s Oryon: The CPU challenging Intel, AMD and Apple
LeCun’s entry into Meta began in 2013, when he joined the company – then still Facebook – to build an ambitious research division focused on advancing foundational AI. FAIR was designed to operate like an academic lab inside a technology company, free from the constant pressure of product deadlines. That environment helped Meta produce breakthroughs in machine learning that influenced the entire industry.
But over the last few years, Meta has transformed. With the explosive rise of generative AI, the company has pivoted sharply toward rapid product development and massive consumer-scale deployments. Everything now moves faster, from model training to feature rollouts, as the company competes with OpenAI, Google, and others. LeCun’s approach to AI, however, has always been rooted in long-horizon research. As Meta accelerated toward short-cycle development, a natural tension grew between the company’s priorities and his own.
According to reporting from Bloomberg and Reuters, that divergence is one of the reasons LeCun decided to step outside Meta and pursue his vision independently. Yet the move is not a hostile departure. Meta is expected to partner with the new venture, an unusual but significant detail that suggests both sides still benefit from staying close. Meta gains access to long-term innovation without slowing its own pace, and LeCun gains freedom from corporate timelines while maintaining a valuable ally.
Also read: Gemini 3 vs Gemini 2.5: A Complete Breakdown of Google’s Next-Gen Upgrade
At the core of his new startup is a concept he calls Advanced Machine Intelligence. While today’s AI world is dominated by large language models that generate text and mimic conversation, LeCun is focused on something far more fundamental. He wants to build AI systems that can understand the physical world, form internal representations of reality, learn continuously from experience, and plan complex actions. In his view, current AI, however impressive, lacks true reasoning and grounding. It operates on patterns, not understanding.
LeCun has spent years arguing that the future of AI lies in giving machines the ability to predict, navigate, and adapt, much like animals or humans. His new venture is designed to pursue that vision without compromise. If successful, it would move the field beyond the constraints of language-first systems and toward a form of intelligence capable of real-world autonomy.
LeCun’s departure from Meta represents a broader shift taking shape across the AI landscape. The field is increasingly split between companies chasing fast, productised AI and researchers pushing toward deeper scientific breakthroughs. His exit reinforces the idea that the next major leaps may come from smaller, focused labs rather than corporate giants optimising for quarterly goals.
With Meta remaining a close partner and LeCun free to pursue a more ambitious scientific agenda, his new company instantly becomes one of the most anticipated ventures in the AI world. The industry now waits to see whether his vision of Advanced Machine Intelligence can push the field beyond its current limits.
Also read: Google Antigravity IDE explained: Free tool for code development
