Companies

Meta's Ambitious Llama 4 Project with NVIDIA GPUs

Published November 1, 2024

Mark Zuckerberg, CEO of Meta, has recently provided insights into the company's development of its new Llama 4 model. This model is being trained on an unprecedented scale, utilizing a massive cluster of over 100,000 NVIDIA H100 AI GPUs. Zuckerberg described the setup as "bigger than anything I've seen," highlighting its significance in the AI landscape.

This remarkable AI infrastructure is expected to cost over $2 billion, primarily due to the expense of the H100 GPU chips. During a recent earnings call with investors and analysts, Zuckerberg confirmed that the initial launch of Llama 4 is anticipated later this year. He mentioned, "We're training the Llama 4 models on a cluster that is bigger than 100,000 H100s, or bigger than anything that I've seen reported for what others are doing. I expect that the smaller Llama 4 models will be ready first." This statement indicates Meta's strong commitment to advancing its AI capabilities.

The substantial investment in the AI supercomputer reflects Meta's strategy to compete effectively in the rapidly evolving field of artificial intelligence. Currently, Meta's advanced cluster is nearing completion, with expectations that it will be operational by October or November.

Moreover, the race for AI dominance is fierce. For instance, Elon Musk's xAI is also escalating its AI capabilities. Musk's company plans to double its Colossus AI supercomputer, which currently operates with 100,000 NVIDIA Hopper AI GPUs, to a total of 200,000 GPUs. This competition among tech giants to build the largest and most powerful AI superclusters is an exciting development in the technology sector.

Zuckerberg's discussions with NVIDIA CEO Jensen Huang further reinforce Meta's position as a key customer for NVIDIA, as Zuckerberg noted that Meta owns over 600,000 H100 GPUs. This large-scale uptake illustrates Meta's aggressive push in the AI domain, showing their intent to leverage these advanced resources to develop innovative models.

As the industry evolves, the interest in these high-powered AI systems continues to grow, making it essential to monitor advancements and their implications on the market and technological progress.

Meta, Zuckerberg, AI