OpenAI plans to build its own AI chips on TSMC's forthcoming

OpenAI is moving forward with plans to develop its own AI chips, utilizing TSMC's upcoming 1.6 nm A16 process node. Reports indicate that OpenAI is collaborating with TSMC for production, shifting from an earlier plan to use the N5 process node. The A16 node will incorporate advanced technologies, including backside power delivery. OpenAI is reportedly in discussions with Broadcom and Marvell for chip design, although these companies lack extensive experience with TSMC's latest nodes. There are speculations about a potential partnership with Apple, which could influence the project, especially as Apple currently relies on Google’s AI servers. OpenAI's decision to create its own chips stems from the high operational costs associated with using Nvidia's AI servers, which dominate the market. While developing proprietary chips could reduce ongoing expenses, the project faces uncertainties, particularly if OpenAI is acquired by major tech firms like Microsoft or Nvidia, which may not prioritize the chip initiative. If successful, the chips are expected to excel in specific applications, particularly for OpenAI's needs, but may not attract interest from other companies due to their specialized nature.

- OpenAI is developing AI chips using TSMC's 1.6 nm A16 process node.

- The project may involve partnerships with Broadcom, Marvell, and potentially Apple.

- The shift from Nvidia's servers aims to reduce operational costs for OpenAI.

- The future of the chip project is uncertain amid potential acquisitions by major tech companies.

- The chips are expected to be highly specialized, primarily benefiting OpenAI's applications.