During Nvidia's annual software developer conference, CEO Jensen Huang will address the growing pressure on major customers to reduce the costs associated with artificial intelligence, as Nvidia faces competition in the AI market. Following China's DeepSeek introducing a competitive chatbot requiring less computing power, Nvidia's stock declined due to its high-priced chips that contributed significantly to its revenue growth in recent years.

The upcoming conference is anticipated to feature the disclosure of Vera Rubin, a new chip system set for mass production later in the year. This follows the release of the previous chip, named after mathematician David Blackwell, which encountered production delays affecting Nvidia's profit margins. With AI markets transitioning from training to inference, Nvidia, dominating the training sector, encounters competition in inference where market share dynamics depend on how inference computing is executed.

Various players in Silicon Valley, including startups and traditional competitors like Advanced Micro Devices, are developing alternative chips to cater to the inference market at reduced costs, particularly in terms of electricity consumption. Nvidia's energy-intensive chips have prompted AI companies to explore more efficient power sources like nuclear reactors. The industry is witnessing a surge in companies like Untether AI challenging Nvidia's dominance in inference markets, leveraging innovative chip technologies to provide cost-effective solutions.