Nvidia CEO Jensen Huang used the company's latest earnings call, following another record-breaking quarter with $81.6 billion in revenue and a forecast of $91 billion for the next, to pitch Vera, its new CPU product, as the entry point into a market the company has never previously addressed. Huang described Vera as the world's first CPU purpose-built for agentic AI, arguing that while GPUs handle the thinking layer of AI models, AI agents primarily run on CPUs when executing their assigned tasks. Vera is specifically designed to process tokens as fast as possible, which Huang contrasted with traditional cloud architecture CPUs built around cores optimized for running multiple application instances simultaneously. He claimed Nvidia has already sold $20 billion worth of standalone Vera CPUs in 2026 alone, a figure he presented as evidence that the market opportunity is real and already materializing rather than speculative.

Huang's core argument is that the growth of AI agents will require a proportional expansion of CPU infrastructure in a way that current general-purpose chips are not optimized to serve. He projected that the world will eventually have billions of AI agents, each using tools that function similarly to how humans use PCs today, creating demand for a scale of purpose-built CPU deployment that no current product adequately addresses. Every major hyperscaler and system maker is reportedly already partnering with Nvidia to deploy Vera, giving the company early distribution momentum in a market where Amazon Web Services and other cloud providers are also actively developing their own AI chip alternatives. Whether Nvidia can hold that position against homegrown chip efforts from the hyperscalers remains the central competitive question, but Huang's track record of delivering on his projections has earned him at least the benefit of the doubt from investors watching the claim unfold.