Meta Secures Millions of NVIDIA Chips for AI Training and Deployment
According to the Economic Desk of Webangah News Agency, Meta has entered into an agreement to integrate millions of NVIDIA’s specialized processors into its infrastructure, primarily for training and executing sophisticated artificial intelligence models. While NVIDIA did not disclose the exact financial value of the contract, the company confirmed that Meta will deploy these chips in both its proprietary data centers and through the NVIDIA Cloud Collaboration Program.
The NVIDIA Cloud Collaboration Program encompasses third-party hosting providers such as CoreWeave and Crusoe, which rent out access to NVIDIA hardware to other corporations. Jensen Huang, Chief Executive Officer of NVIDIA, commented on the partnership, stating that no other company is implementing artificial intelligence at Meta’s scale, emphasizing that this collaboration merges cutting-edge research with massive industrial infrastructure.
Huang further detailed that the partnership will equip Meta’s researchers and engineers with NVIDIA’s complete platform, including deep-code design across CPUs, GPUs, networking, and software, enabling them to build the foundation for the next frontier of AI. NVIDIA announced that in addition to graphic processing units, Meta is launching its first large-scale servers based on the Grace CPU, which operate without discrete GPUs. Furthermore, Meta intends to deploy the Vera systems by 2027.
These Grace-based servers present a potential challenge to established players like Intel and AMD, as they are structurally similar to Meta’s more traditional servers powered by Intel chips and notably lack GPUs. Beyond processing power, Meta will also incorporate NVIDIA’s Confidential Computing technology into its WhatsApp messenger service to ensure private data processing.
The recent news arrives as shares in several AI-related companies face downward pressure. Meta’s stock has declined by 3.3 percent, and Microsoft shares have dropped over 17 percent since the start of the year. NVIDIA stock has seen a slight decline of over one percent year-to-date, while AMD has fallen more than five percent.
Yahoo Finance reported that key NVIDIA clients, including Amazon, Google, and Microsoft, have recently unveiled new versions of their proprietary AI chips. Reports also suggest Meta has been in discussions regarding the use of Google’s TPU chips. Despite these competitive developments, analysts maintain that NVIDIA is unlikely to relinquish its leading position soon, given the extensive deployment of its GPUs across diverse segments of the artificial intelligence landscape.
