Nvidia is projecting a staggering $1 trillion in revenue from its artificial intelligence chips by 2027, as the company moves to cement its leadership in the rapidly growing AI market. CEO and co-founder Jensen Huang revealed the forecast at the company’s annual GTC developer conference in San Jose, California, unveiling a new central processor and an AI system built on technology licensed from chip start-up Groq for $17 billion last December. The announcement highlights Nvidia’s strategy to dominate inference computing, where AI systems answer queries in real time, an area that is becoming increasingly competitive with CPUs and custom processors from companies like Google.
Huang described the “inference inflection” as a major turning point, signaling that demand for AI chips continues to accelerate. Nvidia has long been dominant in AI model training, but the new push into inference represents a broader market opportunity, reflecting the shift from experimental AI deployments to large-scale, real-time applications. Nvidia’s Vera Rubin chips will handle the initial “prefill” stage, converting user requests into AI-readable tokens, while Groq’s chips will process the “decode” stage to deliver answers. This two-step system is designed to meet the surging demand from companies like OpenAI, Anthropic, and Meta Platforms, which are serving hundreds of millions of users worldwide.
The $1 trillion revenue projection doubles Nvidia’s previous forecast of $500 billion through 2026 for its Blackwell and Vera Rubin AI chips, illustrating the company’s confidence in sustained growth despite investor skepticism. Shares briefly spiked following the announcement before settling with a 1.2 percent gain, as some market watchers remain cautious about the company’s aggressive reinvestment strategy in the AI ecosystem. Analysts, however, see Huang’s forecast as evidence that Nvidia’s AI infrastructure remains in high demand and that the broader AI industry is moving from early experimentation to large-scale deployment.
Huang also emphasized Nvidia’s expansion into CPUs with the new Vera CPU, noting the increasing role of traditional processors in inference computing. “We are selling a lot of CPU standalone,” Huang said, forecasting the CPU business alone to become a multibillion-dollar segment. As AI adoption accelerates globally, Nvidia’s dual focus on GPUs for training and CPUs for inference positions the company to capture a dominant share of the AI chip market over the next several years.
