Meta, popularly known as the rebranded tech giant Facebook, is ramping up its play in the artificial intelligence (AI) arena. In a major move to expand its data center capabilities, Meta and Nvidia have entered a strategic, multi-year agreement. What’s so special about this deal? It’s all about Nvidia’s advanced hardware. We’re talking Grace and Vera CPUs and their Blackwell and Rubin GPUs, which will be integrated into Meta’s infrastructure on a large scale.
Meta has always relied on Nvidia’s technology for its AI products, but this deal proves different. It essentially marks the first massive deployment of Nvidia’s Grace-only tech—a momentous leap ahead. Nvidia is pretty confident about this move, pointing out that the deployment will substantially boost performance-per-watt inside Meta’s data centers. In layman’s terms, there’s potential for serious improvements in efficiency and, crucially, sustainability.
Now, let’s cast our gaze a little further into the future. As part of this collaboration, Meta has big plans to usher Nvidia’s next-generation Vera CPUs into its data centers by 2027. It’s this forward-thinking strategy that underscores Meta’s dedication to technology. The goal? To ensure their infrastructure remains strong enough to deal with the soaring demands of AI-driven applications in the era of rapid digital advancements.
While courting Nvidia’s expertise, Meta isn’t giving up on internal growth either. It’s also churning its wheels to develop its own in-house chips, designed specifically for running AI models. All is not smooth sailing here, though. As the Financial Times reported, the company has hit a few tech hurdles and rollout delays on its journey to these proprietary chips. Regardless, Meta remains resolute in its long-term vision, vying for optimum control over its AI infrastructure.
Keen to dive deeper into this story? Check out the full details on The Verge.
This website uses cookies.