Jensen Huang says he’s found a ‘brand new’ $200B market for Nvidia

Nvidia Targets $200B CPU Market for AI Agents

Nvidia CEO Jensen Huang has identified what he calls a “brand new $200 billion” market for his company, centered on its new Vera CPU. Announced during an earnings call that reported another record-breaking quarter, Huang’s proclamation signals a deliberate and aggressive push into the CPU territory historically dominated by Intel and AMD. This move aims to expand Nvidia's reach beyond its core GPU business by creating processors specifically for the emerging field of 'agentic AI'.

Vera: A Purpose-Built Processor

Huang positioned the Vera CPU as a foundational component for the next wave of computing, where AI agents will perform tasks autonomously. He explained that while GPUs handle an AI model's 'thinking,' the agents themselves will primarily run on CPUs to execute their assigned functions. To meet this need, Vera is designed differently from conventional CPUs.

• Primary Function: Optimized for processing tokens as quickly as possible, which is crucial for AI agents.
• Architecture: Diverges from classic cloud CPUs that are built with 'cores' to run multiple applications simultaneously.
• Market Validation: Huang stated that Nvidia has already sold $20 billion worth of standalone Vera CPUs this year.
• Product Strategy: Vera is sold as a standalone processor and is also bundled with the company's Rubin GPU.

This strategy confronts growing anxiety on Wall Street about competition from hyperscalers like Amazon Web Services (AWS), which is developing its own AI chips for customers like Meta. By creating a purpose-built CPU for agents, Huang argues that Nvidia is establishing a new growth driver, predicting a future where billions of AI agents will require their own form of CPU-driven PCs, vastly increasing the demand for this type of hardware.

By introducing the Vera CPU for agentic AI, Nvidia is not just entering a new hardware market; it's attempting to define and own the entire compute stack for a future where AI agents require their own specialized processors, moving beyond its GPU-centric identity to become a full-spectrum AI systems provider.