Nvidia targets Intel and AMD dominance as Jensen Huang bets big on CPUs for AI agents

Nvidia Shifts Strategy as AI Workloads Evolve

Nvidia, the company that built its fortune on graphics processing units, is now positioning itself to become a major force in the CPU market traditionally dominated by Intel and AMD. CEO Jensen Huang revealed during the company's fourth-quarter earnings call that Nvidia is ready to compete aggressively in the CPU arena, declaring "we love CPUs as well as GPUs" 1

1

. The shift comes as evolving AI workloads increasingly favor central processing units, particularly as companies move from training AI models to deploying them in real-world applications.

Huang has long noted that computing ratios flipped in recent years, with 90% of work moving to GPUs from the traditional CPU-dominated landscape 2

2

. But the pendulum is swinging back. At the Consumer Electronics Show in January, Huang predicted that Nvidia could become "one of the largest CPU makers in the world" as high-performance Nvidia CPUs proliferate in data centers 1

1

.

AI Agents Drive Computing Back to CPUs

The resurgence of CPU importance stems from the rise of AI agents that independently execute tasks such as writing code, analyzing documents, and generating research reports. According to Ben Bajarin, an analyst at Creative Strategies, this type of agentic computing "is happening more and more, and sometimes primarily, on the CPU" 2

2

. This represents a fundamental shift in AI model deployment strategies.

Nvidia's current flagship AI server, the NVL72, contains 36 CPUs paired with 72 GPUs. However, Bajarin suggests this ratio could shift to 1-to-1 for agentic work, or GPUs might be bypassed entirely for certain tasks 1

1

. This architectural evolution underscores how AI workloads are diversifying beyond the training phase that made Nvidia's GPUs indispensable.

Jensen Huang CPU Strategy Centers on Data Processing

Nvidia's approach to data center CPUs differs fundamentally from the Nvidia vs Intel and AMD playbook. Huang explained that Nvidia minimized the chiplet approach—breaking chips into smaller components—that competitors favor. Instead, Nvidia's processors are designed for "very high data processing capabilities" with superior memory access 2

2

. "Most of the computing problems that we're interested in are data driven—artificial intelligence being one," Huang stated during the analyst call.

The company recently secured a significant deal with Meta Platforms, which will deploy large volumes of Nvidia Grace and Vera CPUs on a standalone basis—a departure from Nvidia's typical configuration where CPUs accompany multiple GPUs 1

1

. While Meta hasn't abandoned AMD, which also announced a major CPU deal with the social media giant, the partnership signals growing confidence in Nvidia's processor offerings.

What This Means for the Competitive Landscape

Dave Altavilla, principal analyst at HotTech Vision and Analysis, suggests Nvidia aims to prove that Intel's traditional CPU dominance "is no longer the assumed default foundation of modern compute infrastructure. Instead, it becomes just one architectural option among several" 2

2

. This repositioning could reshape data center economics as customers evaluate whether specialized, data-optimized processors deliver better performance for AI workloads than general-purpose alternatives.

Nvidia first released its data center CPU offerings in 2023, making this a relatively young product line compared to Intel and AMD's decades of processor development. Yet the company's dominance in AI infrastructure—and its deep relationships with hyperscalers and AI companies—provides a formidable distribution advantage. Huang promised more CPU disclosures at Nvidia's annual developer conference in Silicon Valley next month 1

1

, suggesting the company has additional announcements prepared to solidify its position in this expanding market segment.