Nvidia’s CEO Projects $1 Trillion In Orders For Blackwell And Rubin Chips

Nvidia CEO Forecasts Demand Surge For New Generations Of AI Processors Amid Production Ramp Plans

At the company’s annual GTC conference in San Jose on Monday, Nvidia Chief Executive Jensen Huang said he expects demand for the firm’s Blackwell and the newer Rubin processors to reach roughly $1 trillion through 2027 — about double the level the company cited for the same product family through 2026. The projection, delivered during Huang’s keynote, underscores how Nvidia sees its data-center chips fueling a rapid expansion of the market for AI hardware.

Huang told the audience that Nvidia had tracked roughly half a trillion dollars in orders for Blackwell and Rubin products through 2026, and that current indicators point to at least twice that amount by the end of next year. His comments framed the company’s roadmap for Rubin, the successor architecture announced in 2024, as a major driver of the anticipated spending surge from cloud providers, enterprises and research organizations.

Rubin, which Nvidia began producing earlier this year, is positioned as a higher-performance follow-up to Blackwell. Company statements from January describe Rubin as capable of about three-and-a-half times the training throughput of Blackwell and roughly five times the speed on inference workloads, with peak performance quoted up to 50 petaflops. Nvidia has said it plans to scale manufacturing in the second half of the year to meet rising orders.

Industry analysts welcomed the estimate but urged caution. “The math starts to work when hyperscalers and big enterprises accelerate AI deployments, but supply and procurement cycles can slow a rapid climb,” said Maria Lopez, a chip-market analyst at TechFoundry, who spoke after the keynote. She added that the timing of factory ramps and customer delivery schedules will be critical to realizing that revenue picture.

For investors and customers, the announcement serves as a reminder of how central specialized AI processors have become to computing strategies. Nvidia’s forecast sets a high bar for the market: if orders approach the level Huang described, it would reshape spending plans for cloud operators and enterprises buying large-scale model training and inference capacity. But turning a projection into sustained deliveries will depend on production execution and how quickly buyers shift budgets toward the newest architectures.