NEW YORK, Jan 20: On Thursday, Meta CEO Mark Zuckerberg announced a strategic move to consolidate the company's AI research team with the generative AI team, emphasizing an intensified effort to integrate AI technology into Meta's products. The social media giant is expanding its infrastructure to accommodate this initiative and aims to acquire approximately 350,000 H100 GPUs from Nvidia by the year-end. In addition to chips from other suppliers, Meta anticipates having a total of around 600,000 GPUs, making its system one of the largest in the tech industry when completed.

This substantial investment in computing power marks a significant shift for Meta, contrasting with its prior emphasis on AI research without a strong focus on integrating it into core social media products and AR/VR hardware. In response to the breakthrough success of OpenAI's ChatGPT in late 2022, Meta established a "GenAI" team last year to spearhead efforts to leverage generative AI across its platforms.

Recent developments include the launch of a commercial version of the Llama large language model, advertising tools capable of generating image backgrounds from text prompts, and a "Meta AI" chatbot accessible through Ray-Ban smart glasses. Zuckerberg also revealed ongoing efforts to train a third version of the Llama model.

Zuckerberg tied these AI investments to Meta's broader vision of the AR/VR-driven metaverse, emphasizing the need for new devices, such as glasses, for interacting with AI. This strategic push underscores Meta's commitment to stay at the forefront of AI technology integration, reflecting a departure from its historical research-focused approach.