Nvidia CEO Jensen Huang told attendees at the Goldman Sachs Communacopia conference that the company expects revenue to grow 70 percent year over year through the end of next year, projecting approximately $680 billion in annual revenue on top of a current fiscal year heading toward $400 billion.
The confidence stems from Nvidia's position as the foundational compute platform for the entire AI industry. Huang said the company runs every major model from Anthropic, OpenAI, and Google, as well as open-weight offerings, and tracks every gigawatt of data center land, power, and construction worldwide through its partner network of neoclouds, OEMs, and AI-native companies.
One product line alone — a system combining 36 Grace CPUs with 72 Blackwell GPUs — is experiencing 27 percent month-to-month sales growth. Huang redefined what a GPU means in the AI era: one unit now costs $8.5 million, contains 2 million parts, and draws 250,000 kilowatts, shipped by the thousands.
On Nvidia's controversial circular investment deals, where the company invests in firms that then buy its hardware, Huang was characteristically blunt: we put in $1 and $100 comes back. He insisted every investment is backed by real customer contracts totaling over $100 billion in verified revenue.
The outlook raises the familiar question of whether any AI infrastructure dominance can persist long-term. For now, Nvidia's embedded position across every layer of the AI stack — from suppliers to data centers to the models themselves — gives it an unmatched vantage point on the industry's trajectory.
