Nvidia’s Vera Chip Targets $200 Billion AI Inference Market

Nvidia is sharpening its focus on the fast-growing AI inference market with the introduction of its Vera chip, as outlined in its recent earnings call. While Nvidia’s record Q1 revenues once again surpassed expectations—with $81.62 billion against estimates of $78.86 billion—the company’s leadership emphasized a more strategic development amid ongoing sector shifts.

Chief Executive Jensen Huang revealed that the company’s new Vera central processors are positioned to tap a $200 billion market opportunity, distinct from the $1 trillion in projected revenue from Nvidia’s established Blackwell and Rubin AI GPU lines, expected between 2025 and 2027. Huang forecasts that Vera could generate $20 billion in revenue by the end of this fiscal year and described the chip as Nvidia’s “second largest” sales driver moving forward.

The growing need for AI inference—deploying and serving machine learning models in real time—has shifted the industry’s focus. As cloud giants such as Google, Amazon, and Microsoft collectively plan to invest over $700 billion in AI infrastructure this year, many are also developing their own custom chips to optimize AI workloads. Established chipmakers, including Intel and AMD, have also expanded their CPU offerings for inference, challenging Nvidia’s GPU dominance.

Historically, Nvidia has led the critical model training phase of AI development, but the competition is intensifying around inference, where speed and efficiency at scale are paramount. Google with its TPU series, Amazon with Trainium, and newer entrants like Groq—all emphasize rapid, cost-effective inference solutions.

Nvidia’s Vera chip is designed specifically for this task. Partly developed using technology from inference-focused startup Groq, which Nvidia reportedly licensed for $17 billion, the Vera chip forms the basis of the Vera Rubin platform. This upcoming release combines Vera CPUs and Rubin GPUs to support complex, large-scale AI workloads. The full launch is expected later this year.

However, Huang cautioned about a potential bottleneck: "My sense is that we’ll be supply constrained through the entire life of Vera Rubin." To address this, Nvidia has increased its supply commitments to $119 billion in Q1—up substantially from $95.2 billion the previous quarter—underlining both strong demand and industry concerns about global memory chip availability.

Financially, Nvidia announced an $80 billion share repurchase programme and raised its quarterly cash dividend. Yet, after the results, Nvidia’s shares dipped 1.6% in after-hours trading. This reflects recurring analyst concerns about whether Nvidia can maintain its growth as the AI sector’s focus broadens beyond training into inference, facing custom silicon competition from major tech and semiconductor players.

Huang countered by pointing to rapid growth among AI-centric cloud customers, whose spending now rivals—yet outpaces—the traditional hyperscalers quarter over quarter. He stressed that the Vera chip is crucial to capitalizing on this trend, even as supply chain obstacles persist.

artificialintelligence-news.com