In a bold move to advance artificial intelligence, Elon Musk revealed plans for a state-of-the-art supercomputer at his AI startup, xAI. The ambitious project aims to drive the next iteration of xAI's chatbot, Grok, showcasing Musk's ongoing commitment to pushing the boundaries of AI technology.
During a recent presentation to investors, as reported by The Information, Musk outlined his vision to have the supercomputer operational by fall 2025. This advanced system is set to potentially collaborate with Oracle, a major player in the tech industry, to bring the massive computing infrastructure to life.
The proposed supercomputer will leverage Nvidia's flagship H100 graphics processing units (GPUs), renowned for their dominance in the AI data center market. According to Musk, the interconnected H100 chips will create a system at least four times larger than the current largest GPU clusters, representing a significant leap in computational power.
With the increasing demand for powerful GPUs, securing enough H100 units poses a challenge. Musk's previous projects underscore this need; earlier this year, he mentioned that training the Grok 2 model required about 20,000 H100 GPUs, with future models like Grok 3 anticipated to need up to 100,000 chips.
Founded last year, xAI positions itself as a competitor to established AI giants like Microsoft-backed OpenAI and Alphabet's Google. Musk's dual role as a founder of both xAI and OpenAI highlights his influential presence in the AI landscape, driving innovation and competition in the field.
Reference(s):
cgtn.com
