With 22,000 pieces of Nvidia H100 GPU,Inflection AI is developing a supercomputer

CTOnews.com, July 5 (Xinhua) artificial intelligence startup Inflection AI recently announced that it is developing a supercomputer with 22,000 pieces of Nvidia H100 GPU to meet the needs of the development of generative AI.

Inflection AI, founded by Google DeepMind co-founder Mustafa Suleyman and recruitment platform LinkedIn co-founder Reid Hoffman, focuses on developing consumer-oriented artificial intelligence products and is considered OpenAI's main competitor.

CTOnews.com previously reported that Inflection AI recently received $1.3 billion in financing, and investors such as Microsoft and Nvidia continued to participate.

Inflection AI says the supercomputer consumes an astonishing 31 megawatts with 22000 H100 GPU and nearly 700 four-node racks of Intel Xeon CPU.

The company just launched Pi, a chat robot, last month, which uses generative artificial intelligence technology similar to ChatGPT to interact with users through conversations, allowing people to ask questions and share feedback. Inflection AI says it hopes to create a personalized artificial intelligence product that helps people plan, schedule, gather information and perform other tasks.

▲ chat robot Pi, image source Inflection AI official website