Nvidia’s CEO, Jensen Huang, recently unveiled Blackwell, the company’s latest artificial intelligence graphics engine, with an estimated price range of ,000 to ,000 per unit. Huang highlighted the innovative technology developed to bring Blackwell to market, with Nvidia pouring nearly billion into research and development.
The pricing of the Blackwell chip suggests it will be on par with its predecessor, the H100 (Hopper), which cost between ,000 and ,000 per chip. Analysts anticipate high demand for AI software like ChatGPT, driving interest in Nvidia’s AI processors. The release of the Hopper generation in 2022 saw a significant increase in costs compared to previous models.
During an interview with CNBC, Huang clarified that the price of Blackwell includes infrastructure costs associated with building data centers and integrating them with other businesses. Nvidia typically releases new AI processor generations every two years, with each iteration boasting improved performance and energy efficiency. Blackwell surpasses its predecessors in physical size and consists of two chips.
Since the launch of OpenAI’s ChatGPT in 2022, Nvidia’s quarterly sales have tripled, fueled by the popularity of its AI processors. Leading AI developers have been utilizing Nvidia’s H100 for training their AI models over the past year.
Nvidia does not disclose the list price of its chips, which come in various configurations. The final price for end-users, such as Microsoft or Meta, varies based on factors like chip volume, direct purchases from Nvidia, or through vendors like Supermicro, Dell, HP, who offer AI servers equipped with up to eight AI GPUs.
