Meta expands cooperation with Nvidia, to use millions of AI chips in data center construction

Source: Global Market Report

A large-scale new agreement announced on Tuesday local time shows that Meta will use millions of Nvidia chips in its artificial intelligence data centers, including Nvidia's brand-new standalone CPU and the next-generation Vera Rubin system.

Meta CEO Mark Zuckerberg stated in a press release that the expanded cooperation will continue to advance the company’s goal of “providing personal superintelligence for everyone in the world,” a vision officially proposed in July last year.

The financial terms of the agreement have not been disclosed.

In January this year, Meta announced that its AI-related investment could reach up to $135 billion by 2026. Ben Bajarin, a chip industry analyst at the creative strategy consulting company, stated: “The scale of this deal is undoubtedly in the tens of billions of dollars. It is expected that a significant portion of Meta’s capital expenditure will be invested in this partnership with Nvidia.”

This is not the first collaboration between the two parties. Meta has used Nvidia GPUs for at least a decade, but this agreement marks a significant expansion in the scope of technical cooperation between these two Silicon Valley giants.

The independent CPU is the biggest highlight of this cooperation: Meta will become the first company to deploy Nvidia Grace CPUs as standalone chips in its data centers, rather than integrating them with GPUs in servers. Nvidia says this is the first large-scale independent deployment of Grace CPUs.

Bajarin stated: “These chips are specifically designed to run inference tasks and intelligent agent tasks, as supporting components for the Grace Blackwell/Vera Rubin cabinets. Meta’s large-scale adoption of this move validates Nvidia’s full-stack strategic layout in both CPU and GPU infrastructures.”

The next-generation Vera CPU is planned to be deployed and launched by Meta in 2027.

This multi-year agreement is part of Meta’s overall plan: the company has committed to investing $600 billion in the United States for data centers and supporting infrastructure construction by 2028.

Meta plans to build 30 data centers, 26 of which will be located in the United States. Currently, two of the largest AI data centers are under construction: the 1 GW Prometheus project in New Albany, Ohio, and the 5 GW Hyperion project in Richland Parish, Louisiana.

The agreement also includes Nvidia’s networking technology Spectrum-X Ethernet switches, used to achieve high-speed GPU interconnection within large AI data centers. Meta will also use Nvidia’s security capabilities to support AI features for WhatsApp.

This social media giant does not rely solely on this leading chip manufacturer. In November last year, it was reported that Meta was considering using Google Tensor Processing Units (TPUs) in its data centers by 2027, causing Nvidia’s stock price to drop 4% on the day the news broke.

Meta is also developing its own chips and uses AMD products. In October last year, AMD reached a major cooperation with OpenAI, as AI giants sought a second supply channel amid Nvidia’s supply constraints.

Nvidia’s current Blackwell GPUs have been out of stock for several months, and the next-generation Rubin GPU has only recently entered mass production. Through this agreement, Meta has secured sufficient supply.

Nvidia and Meta engineering teams will conduct in-depth joint design to optimize and accelerate the training and inference of cutting-edge AI models for Meta.

Meta is developing a new cutting-edge large model codenamed Avocado, as the successor to its Llama series of AI technologies. According to previous reports, the latest version released last spring failed to satisfy developers.

In recent months, Meta’s stock price has fluctuated sharply, and its AI strategy in particular has puzzled Wall Street.

In October last year, after Meta announced its ambitious AI investment plan, the stock price recorded its largest single-day drop in three years; in January this year, after the company issued revenue guidance exceeding expectations, the stock rose by 10% in a single day.