
It will be produced with 3nm technology of TSMC
OpenAI will thus not have to be connected to the Nvidia chips to train and run artificial intelligence models. According to Reuters, the new chip will have a high -bandwidth memory (HBM) and comprehensive network capabilities and will be produced with TSMC’s 3nm technology.
OpenAI will initially produce a limited number of chips and often use artificial intelligence models to run. The company is also planning more advanced chips for the future.
First of all, it was reported that OpenAI had developed a special chip by going to cooperation with Broadcom. OpenAI’nin for this purpose former Google TPU engineer Richard Ho’yu design team, and in recent months, the team has increased from 20 people to 40 people said.