Nvidia’s Vera Aims at $200B Inference Market

Nvidia reported Q1 revenue of $81.62 billion and guided Q2 revenue to $91 billion. On the company’s earnings call, CEO Jensen Huang told analysts that the new Vera CPUs open a $200 billion market for inference and that he expects Vera revenue to reach $20 billion by the end of the fiscal year. He estimated Vera will become the company’s second-largest sales contributor.

Huang placed the Vera opportunity outside Nvidia’s earlier forecast that Blackwell and Rubin GPUs represent about $1 trillion in opportunity between 2025 and 2027. The company framed Vera as a response to rising demand for inference hardware, the processors used to run AI models in production to generate responses at scale and in real time.

Major cloud providers including Google, Amazon and Microsoft are developing custom chips for inference, and chipmakers Intel and AMD have promoted CPUs for those workloads. Inference workloads prioritize latency, power efficiency and cost per query over the raw floating-point performance that GPUs deliver for model training.

Nvidia said Vera incorporates technology licensed from Groq in a deal reported at roughly $17 billion. The full Vera Rubin offering, combining the Vera CPU with Rubin GPUs, is scheduled to ship later this year.

Huang addressed manufacturing limits on the call, warning that supply rather than demand will constrain Vera Rubin. “My sense is that we’ll be supply-constrained through the entire life of Vera Rubin,” he told analysts.

To expand production capacity, Nvidia disclosed that its supply commitments rose to $119 billion in Q1, up from $95.2 billion in the prior quarter. The company also announced an $80 billion share repurchase program and increased its quarterly cash dividend to $0.25 from $0.01.

Investor reaction was cautious after the release. Nvidia shares fell in extended trading following the report. Some market participants noted repeated quarterly beats are largely expected and questioned whether Nvidia can sustain growth as the industry shifts focus to inference silicon and in-house cloud solutions.

Analyst Jacob Bourne noted that while Nvidia has continued to beat estimates, investors remain focused on whether demand will be durable into 2027 and 2028 as competitors and hyperscalers develop their own inference chips.

Huang highlighted a growing group of AI-focused cloud customers outside the largest hyperscalers, saying their spending is roughly equal to hyperscale capex and is growing faster quarter over quarter. Nvidia projects Vera’s focus on inference efficiency and lower latency will appeal to customers seeking higher throughput and lower serving costs for deployed models.

Whether Vera reaches the $20 billion revenue target will depend on Nvidia expanding supply, securing customer design wins against in-house alternatives and delivering the integrated Vera Rubin platform when it ships later this year.