Meta's 'LLaMA 3.1' Launch Announcement Sparks Buzz with Jensen Huang and Zuckerberg Swapping Jackets
Strengthening AI GPU and LLaMA 3.1 Mutual Supply Cooperation
Also Applied to Foundry Services
Two leading global big tech companies spearheading the era of artificial intelligence (AI), Nvidia and Meta, are gradually concretizing their collaboration.
Mark Zuckerberg, CEO of Meta, and Jensen Huang, CEO of NVIDIA, are taking a commemorative photo wearing each other's shirts. Photo by Mark Zuckerberg Instagram
On the 25th (local time), Nvidia announced that it will provide the Nvidia AI Foundry (semiconductor foundry) service and Nvidia NIM inference microservices to enhance generative AI for companies worldwide through Meta's open-source AI model collection, "Llama 3.1."
This news came just one day after Meta announced the release of Llama 3.1. Accordingly, Meta has secured a path to widely distribute Llama 3.1 simultaneously with its launch through various services provided by Nvidia. The two companies have accelerated their cooperation after revealing a photo of CEOs Jensen Huang and Mark Zuckerberg swapping jackets and publicly shaking hands. Meta disclosed that it used 16,000 of Nvidia's latest GPUs, the "H100," to train Llama 3.1, and CEO Zuckerberg set a goal earlier this year to purchase 350,000 H100 units by the end of the year.
Through the collaboration of the two companies, customers using Nvidia AI Foundry will be able to create customized "supermodels" tailored to domain-specific industrial use cases by leveraging Llama 3.1 along with Nvidia's software, computing power, and expertise. These companies can train such models using their own data as well as synthetic data generated from Llama 3.1 405B and Nvidia's Nemotron reward model.
Nvidia AI Foundry is powered by the Nvidia DGX Cloud AI platform, co-designed with the world's leading public clouds. This provides companies with vast computing resources that can be easily scaled according to changes in AI demand.
The service supports not only enterprises but also the construction of customized large language models (LLMs) for countries developing their own AI strategies. These countries aim to build customized LLMs equipped with domain-specific knowledge for generative AI applications that reflect their unique businesses or cultures, and Nvidia's service can assist in this effort.
Nvidia CEO Jensen Huang said, "Meta's open-source model, Llama 3.1, marks a pivotal moment for the adoption of generative AI by companies worldwide. Llama 3.1 has opened the door for all companies and industries to build cutting-edge generative AI applications. Nvidia AI Foundry has fully integrated Llama 3.1 and is ready to support companies in building and deploying customized Llama supermodels."
Meta CEO Mark Zuckerberg also emphasized, "The new Llama 3.1 model is a very important step for open-source AI. Companies can easily create, customize, and deploy state-of-the-art AI services that people want through Nvidia AI Foundry and Nvidia NIM. We are excited to be able to provide these services to people."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
