meta will spend millions Nvidia The sweeping new deal announced Tuesday gives Nvidia access to the company’s artificial intelligence data center chips, including new standalone CPUs and next-generation Vera Rubin systems.
Meta CEO Mark Zuckerberg said in a statement that the expanded partnership continues the company’s commitment to “bringing personal superintelligence to every person in the world,” a vision announced in July.
Financial terms of the deal were not disclosed.
Shares of Meta and Nvidia rose during extended trading Tuesday. advanced micro device The stock price fell about 4% on the news.
In January, Meta announced plans to spend up to $135 billion on AI in 2026. “This deal is definitely worth tens of billions of dollars,” said Ben Bajarin, a chip analyst at Creative Strategies. “We expect a significant portion of Meta’s capital expenditures to go toward this Nvidia addition.”
This partnership is not new, as Meta has been using Nvidia graphics processing units for at least a decade, but this deal represents a very broad technology partnership between the two Silicon Valley-based giants.
Standalone CPUs are the biggest new feature in the deal, making Meta the first company to bring Nvidia’s Grace central processing units to data centers as standalone chips, rather than being embedded with GPUs in servers. According to Nvidia, this is the first time the company has deployed Grace CPUs on a large scale.
“These are actually designed as companions to the Grace Blackwell/Vera Rubin rack to run inference and agent workloads,” Bajarin said. “For Meta to do this at scale is an affirmation of the soup-to-nuts strategy that Nvidia is deploying for both CPU and GPU infrastructure.”
Next-generation Vera CPUs are expected to be introduced by Meta in 2027.
The multi-year agreement is part of Meta’s overall commitment to spend $600 billion on data center and facility infrastructure in the U.S. by 2028.
Meta plans 30 data centers, 26 of which will be based in the United States. The company’s two largest AI data centers are currently under construction. One is the 1-gigawatt Prometheus site in New Albany, Ohio, and the other is the 5-gigawatt Hyperion site in Richland Parish, Louisiana.
The deal also includes Spectrum-X Ethernet switches, Nvidia’s networking technology used to link GPUs within large AI data centers. Meta also uses Nvidia’s security features as part of WhatsApp’s AI capabilities.
Social media giants don’t just rely on top chip makers. In November, Nvidia shares fell 4% after reports that Meta was considering using it. googleTensor processing units in data centers in 2027.
Meta also develops its own silicon processors and utilizes chips from AMD. AMD landed a notable deal with OpenAI in October as the AI giant sought a second source of supply to Nvidia amid limited supply.
Nvidia’s current Blackwell GPUs have been on backorder for several months, and its next-generation Rubin GPUs recently began production. With this deal, Meta secured a healthy supply of both.
Nvidia and Meta’s engineering teams will collaborate on “in-depth co-design to optimize and accelerate cutting-edge AI models” for the social media giant.
Meta is developing a new frontier model called Avocado as a successor to its Llama AI technology. The latest version, released last spring, failed to excite developers, CNBC previously reported.
Meta’s stock has been on a roller coaster in recent months, with the company’s AI strategy in particular perplexing Wall Street.
Shares hit a three-year low in October after the company announced ambitious AI investments, but jumped 10% in January after the company announced stronger-than-expected sales guidance.
CNBC’s Jonathan Vanian and Kristina Partinevelos contributed to this report.
