OpenAI is collaborating with Broadcom to create its own AI chip, which could be released as early as 2026, writes TechCrunch.
The company is already working with Broadcom to develop this specialized chip, intended to run AI models at a lower cost. The decision to produce chips in-house comes amid high prices and delays caused by global demand for Nvidia GPUs, until now the main resources used by OpenAI.
Instead of investing in a network of its own fabs, OpenAI will focus on chip design in partnership with TSMC and Broadcom. In the short term, the company will continue to use AMD chips through Microsoft’s Azure cloud platform to train AI models, expanding the range of vendors beyond Nvidia.
Moreover, the move represents an important step for OpenAI in reducing reliance on traditional vendors and controlling operating costs as demand for AI resources continues to grow.
Source: www.descopera.ro
