Nvidia Makes a $1T Prediction

Nvidia just upped its already eye-popping AI ambitions, with CEO Jensen Huang now projecting $1 trillion in revenue through 2027 from its AI chips—double last year's forecast. Speaking at the company's GPU Technology Conference, Huang noted, "a trillion dollars is an enormous amount of infrastructure; you have to have complete confidence that the trillion dollars you're putting down will be utilized, would be performant, would be incredibly cost-effective and have useful life for as long as you could see." He argued that so-called "agentic AI" will make that happen. As AI agents mature, he said, the balance of money shifts from training models to running them—an inference-heavy future he traces back to tools like Anthropic's Claude Code. He said every Nvidia software engineer is now assisted with coding by such AI agents.

Huang cast agentic AI as "the new computer" and predicted every software as a service or "SaaS" firm will morph into an AgaaS or "agentic as a service" business. As CNN puts it, this essentially means Nvidia believes "AI agents are going to be everywhere." Nvidia is trying to own that shift: rolling out its first Groq inference chips, moving deeper into CPUs, and teaming with OpenClaw, an open-source AI agent platform he likened to Windows for agentic machines. OpenClaw's need for full device access has triggered security warnings, so Nvidia also introduced NemoClaw as a supposedly safer enterprise option. The company is pushing space-based AI data centers and robotaxis with global automakers, but investor enthusiasm looks cooler: Nvidia shares dipped slightly after the keynote. As Axios notes, Nvidia "is operating at the center of the AI universe"; its processing infrastructure powers Claude as well as OpenAI's ChatGPT.