Nvidia CEO Jensen Huang is set to take the stage this week at the company’s annual software developer conference, where he will defend the nearly $3 trillion chipmaker’s dominance as mounting pressure forces its biggest customers to rein in artificial intelligence costs.

The conference follows market turbulence sparked by China’s DeepSeek, which recently unveiled a competitive chatbot that allegedly requires less computing power than its rivals. Nvidia’s stock took a hit, as the company’s soaring revenue—quadrupling over the past three years to $130.5 billion—has been fueled by selling high-performance AI chips that cost tens of thousands of dollars each.

At the event, Nvidia is expected to unveil details about its upcoming chip system, Vera Rubin, named after the astronomer who contributed to the discovery of dark matter. The system is slated for mass production later this year. However, it follows the David Blackwell chip, announced at last year’s conference, which has seen production delays that have affected Nvidia’s profit margins.

Nvidia also faces growing competition as the AI market shifts from “training”—the process of feeding AI models large datasets to improve their capabilities—to “inference,” where models generate responses for users. While Nvidia dominates the training sector with over 90% market share, its grip on inference computing is less certain, with competitors eyeing a larger slice of the market.