Nvidia Doubles AI Chip Opportunity to $1 Trillion, Launches Vera CPU
CEO Jensen Huang raised Nvidia's addressable AI chip opportunity to $1 trillion through 2027, doubling last month's $500 billion forecast, and unveiled a new Vera CPU alongside an AI system leveraging $17 billion in Groq-licensed technology. The company outlined an inference architecture splitting prefill to Vera Rubin chips and decode to Groq hardware.
1. Market Opportunity Revision
Nvidia raised its addressable AI chip market to at least $1 trillion through 2027, up from last month's $500 billion forecast for Blackwell and Rubin series. This revision reflects the company's focus shift from training workloads toward the more commercially urgent inference phase, which it expects to drive significant revenue growth.
2. New Hardware and Partnerships
CEO Jensen Huang unveiled the Vera CPU, marking Nvidia's entry into central processor design alongside its established GPU lineup. The announcement also introduced a new AI system leveraging technology licensed from Groq in a $17 billion deal, targeting optimized inference performance.
3. Inference Architecture Strategy
Nvidia detailed an inference architecture splitting tasks into two stages: prefill tasks handled by Vera Rubin chips and decode tasks performed by Groq hardware. This division aims to address distinct computational demands—parallel processing for prefill and sequential, latency-sensitive operations for decode—clarifying Nvidia's strategy to dominate inference.