As per the report of The Information, OpenAI is currently in discussions with system-on-a-chip designers, including Broadcom to develop a new artificial intelligence chip. This project’s primary goal is to address the lack of expensive graphics processing units (GPUs) required to build AI models, including ChatGPT, GPT-4, and DALL-E3.
These components make the creation and fine-tuning of deep learning models time-consuming and expensive; thus, to decrease its reliance on Nvidia GPUs, OpenAI is even thinking of developing its AI chips. The firm has been recruiting Google engineers who designed TPU and is now coming up with an AI server chip, insiders reveal.
The representative of OpenAI noted that the company continues negotiations with the market and governmental actors to enhance the necessary infrastructural conditions so that AI benefits can be provided to the public.
News from Bloomberg News revealed that OpenAI’s CEO, Sam Altman said he intends to source for billions of dollars to start constructing a network of factories that will produce semiconductors. Possible collaborators in the development of this initiative are Intel, Taiwan Semiconductor Manufacturing Co (TSMC), and Samsung Electronics.