In a strategic move to bolster its AI capabilities, Meta has secured the expertise of Trapit Bansal, a prominent researcher from OpenAI, to join its newly formed AI superintelligence unit. Bansal, who has been with OpenAI since 2022 and played a pivotal role in the company's reinforcement learning initiatives alongside co-founder Ilya Sutskever, is listed as a foundational contributor to OpenAI's first AI reasoning model, o1. His departure from OpenAI, confirmed by spokesperson Kayla Wood, marks a significant addition to Meta's ranks.

Bansal's move to Meta could be a game-changer for the company's AI superintelligence lab, which is already home to industry leaders such as former Scale AI CEO Alexandr Wang and is eyeing the addition of former GitHub CEO Nat Friedman and Safe Superintelligence co-founder Daniel Gross. Meta aims to develop a cutting-edge AI reasoning model that can compete with the likes of OpenAI's o3 and DeepSeek's R1, as it currently does not offer a publicly available AI reasoning model.

Mark Zuckerberg has been actively expanding Meta's AI team, offering lucrative compensation packages worth $100 million to top researchers. The recent addition of Bansal to the team, along with other former OpenAI researchers Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai, as well as former Google DeepMind researcher Jack Rae and former machine learning leader at Sesame, Johan Schalkwyk, underscores Zuckerberg's success in attracting top-tier AI research talent.

Meta's pursuit of AI excellence has also involved attempts to acquire startups with robust AI research labs, such as Safe Superintelligence, Thinking Machines Labs, and Perplexity. Although these acquisition talks did not advance to a final stage, Meta's commitment to AI development remains evident.

OpenAI CEO Sam Altman has acknowledged Meta's efforts to recruit top talent from his startup but maintains that none of OpenAI's best people have accepted Meta's offers. Despite this, Meta's AI superintelligence team is poised to become a critical internal group, potentially powering products across the company, much like Google's DeepMind unit.

Meta's ambition to build AI agents for business, under the leadership of former Salesforce CEO of AI, Clara Shih, requires the development of advanced AI reasoning models. With the addition of Bansal and other key AI researchers, Meta is positioning itself to lead in the AI race. However, the upcoming release of an open AI reasoning model by OpenAI in the coming weeks could add pressure to Meta's AI offerings, making the competition in this space even more intense.