Nvidia’s deal with Meta signals a new era in computing power

Ask anyone what Nvidia makesand they’ll likely say “GPUs” first. For decades, a chipmaker has been defined by advanced parallel computing, and the emergence of generative AI and the resulting increase in demand for GPUs has been a problem. A blessing for the company.

But Nvidia’s recent moves suggest it’s looking to attract more customers at the less compute-intensive end of the AI ​​market — customers who don’t necessarily need the best and smartest. Powerful graphics processing units To train AI models, but instead they are looking for the most efficient ways to run AI agent programs. Nvidia recently spent billions to license technology from a chip startup focused on low-latency AI computing, and has also begun selling standalone CPUs as part of its latest superchip ecosystem.

And yesterday, Nvidia and Meta Announce The social media giant has agreed to buy billions of dollars worth of Nvidia chips to provide computing power for the social media giant’s massive infrastructure projects – with Nvidia CPUs as part of the deal.

The multi-year deal is an expansion of the ongoing, cozy partnership between the two companies. Meta previously estimated that by the end of 2024, it would have been bought out 350,000 H100 chipset From Nvidia, and that by the end of 2025 the company will have access to Total 1.3 million GPUs (Although it wasn’t clear if these were all Nvidia chips.)

As part of the latest announcement, Nvidia said Meta will “build large-scale data centers optimized for both training and inference to support the company’s long-term AI infrastructure roadmap.” This includes the “large-scale deployment” of Nvidia CPUs and “millions of Nvidia Blackwell and Rubin GPUs.”

Notably, Meta is the first tech giant to announce that it will make a large-scale purchase of Nvidia’s Grace CPU as a standalone chip, something Nvidia said would be an option when it revealed the full specifications of its new Vera Rubin superchip in January. Nvidia also emphasized that it is offering technology that connects different chips, as part of its “integrated approach” to power computing, in the words of one analyst.

The move signals that Nvidia recognizes that a growing array of AI software now needs to run on CPUs, in the same way traditional cloud applications do, says Ben Bajarin, CEO and principal analyst at technology market research firm Creative Strategies. “The reason the industry is bullish on CPUs inside data centers right now is agent AI, which is putting new demands on general-purpose CPU architectures,” he says.

A A recent report from the Semiana Analysis chip newsletter He emphasized this point. Analysts note that CPU usage is accelerating to support AI training and inference, citing one of Microsoft’s data centers for OpenAI as an example, where “tens of thousands of CPUs are now needed to process and manage petabytes of data generated by GPUs, a use case that would not be required without AI.”

However, Bajarin points out that CPUs are still just one component of more advanced AI hardware systems. The number of GPUs Meta buys from Nvidia still outnumbers the number of CPUs.

“If you’re one of the hyperscalers, you’re not going to run everyone “Your inferential computing is on CPUs,” Bajarin says. “You just need whatever software you’re running to be fast enough on the CPU to interact with the GPU architecture that’s actually the driving force of that computing. Otherwise the CPU becomes a bottleneck.”

Meta declined to comment on its expanded deal with Nvidia. During a recent earnings call, the social media giant said it plans to significantly increase its AI infrastructure spending this year to between $115 billion and $135 billion, compared to $72.2 billion last year.