Nvidia Bets Big on Agentic AI with New Vera CPU
Nvidia founder and CEO Jensen Huang has identified what he calls a "brand new $200 billion TAM" — total addressable market — for his company, centered on a newly introduced CPU product called Vera. The announcement came during Nvidia's latest earnings call, following another record-breaking quarter in which the company posted $81.6 billion in revenue and forecast $91 billion for the next quarter. The scale of those figures alone underscores why the industry pays close attention when Huang makes bold claims about new market opportunities.
The Details: What Is Vera and Why Does It Matter?
Vera is Nvidia's new CPU, introduced in March. Unlike its GPU products — which have made Nvidia the dominant force in AI model training and inference — Vera is specifically designed with agentic AI workloads in mind. Huang describes it as "the world's first CPU, purpose-built for agentic AI."
The fundamental distinction Huang draws is architectural. Traditional cloud CPUs are built around "cores" — designed to run multiple application instances simultaneously as efficiently as possible. Vera, by contrast, is engineered to process tokens as fast as possible. This makes it better suited to the way AI agents operate: executing tasks sequentially and at scale, rather than managing parallel compute loads in the conventional cloud sense.
Vera is available both as a standalone product and bundled with Nvidia's Rubin GPU. According to Huang, the company has already sold $20 billion worth of standalone Vera CPUs this year, suggesting early commercial traction is substantial. He also noted that every major hyperscaler and system maker is already partnering with Nvidia to deploy Vera.
The Agentic AI Opportunity: A New Computing Paradigm
Huang's core thesis is straightforward: while the "thinking" part of an AI model runs on GPUs, AI agents primarily run on CPUs. As agentic AI systems proliferate — executing tasks, using tools, and operating with degrees of autonomy — the demand for purpose-built CPU infrastructure will grow dramatically.
Huang made the scale of his vision explicit on the call: "The world has a billion users, human users. My sense is that the world is going to have billions of agents... and those billions of agents will all use tools. And those tools are going to be like PCs, just like us humans using PCs today. We're going to need a lot more CPUs."
In this framing, agents are not just software features running inside existing platforms — they are entities that will require dedicated computing infrastructure, much as human users today rely on personal computers. This positions the CPU market as a long-term, high-volume growth opportunity that Nvidia has historically not targeted as a core business.
Competitive Context: Why This Is a Contested Market
Nvidia's move into purpose-built AI CPUs does not go uncontested. The CPU market has historically been dominated by Intel and AMD. More recently, major cloud providers have invested heavily in developing their own AI chips. Amazon Web Services, for instance, recently announced a large contract with Meta for its homegrown AI CPUs, and AWS CEO Andy Jassy has stated publicly that he believes AWS can match or exceed Nvidia's capabilities in both GPU and CPU AI chip development.
The entry of hyperscalers into chip development represents a structural competitive risk that Wall Street has flagged repeatedly. Huang's response, at least implicitly, is that Nvidia's head start in purpose-built agentic CPU design — combined with its existing relationships with every major hyperscaler and system maker — provides a durable advantage. Whether that proves true at scale remains to be seen.
Relevance for E-Commerce and Shopware Operators
For shop operators, e-commerce managers, and developers working with Shopware 6, the broader shift toward agentic AI infrastructure has direct practical implications. Agentic AI systems — capable of autonomously executing multi-step tasks such as product data enrichment, customer service workflows, inventory management, or dynamic content generation — are moving from experimental to production-grade deployments.
The infrastructure investments Nvidia is making with Vera signal that the compute layer underpinning these capabilities is maturing rapidly. As purpose-built CPU architectures lower latency and increase throughput for agent workloads, the operational performance of AI-powered e-commerce tools — including plugins like SOS Newsdesk Autopilot — will benefit from an increasingly capable and cost-efficient infrastructure layer.
Practical Considerations for Teams Evaluating AI Tooling
- Agentic AI is becoming infrastructure-grade: The scale of investment from Nvidia signals that agentic AI is no longer a niche capability. Shop operators should evaluate which of their workflows are candidates for agent-based automation.
- CPU-driven agents will handle task execution: Understanding the distinction between GPU-based model inference and CPU-based agent task execution helps teams make more informed decisions when evaluating AI vendors and their underlying infrastructure.
- Hyperscaler competition may drive down costs: As AWS, Nvidia, and others compete in the AI CPU space, the cost of running agentic workloads at scale may decrease over time — improving the economics of AI-powered content and operational automation.
Outlook: The World Is Rebuilding Computing for Agentic AI
Huang's summary of the moment is direct: "The world is rebuilding computing for agentic AI and robotic physical AI. Nvidia sits at the center of these transitions." Whether Vera ultimately captures the $200 billion market Huang envisions will depend on how the competitive landscape evolves — particularly as hyperscalers continue developing proprietary silicon.
What is already clear is that the infrastructure layer for agentic AI is being built now, at significant scale and speed. For e-commerce teams planning their AI tooling roadmap, the direction of travel is unambiguous: agent-based automation is moving to the center of enterprise computing, and the hardware ecosystem is being designed to support it.