Openai is reportedly accelerating efforts to reduce dependence on NVIDIA by developing its own artificial intelligence (AI) silicon and plans to complete the design of its first in-house chip in the coming months. The Chatgpt Maker is preparing to use its design for manufacturing, according to sources familiar with the matter. According to a source familiar with the matter.
The process of submitting production chip designs (called “tapes”) is reportedly marked Openai’s ambition to mass-produce its AI chips by 2026. Given that the initial tape could cost tens of millions of dollars and it would take about six months to determine, the first iteration will work as expected. Any problem may require another cycle of troubleshooting and re-extraction, delaying progress.
The initiative is reportedly seen as a strategic move to improve the negotiation leverage between OpenAI and its leading chip suppliers. Following this initial design, OpenAI plans to develop increasingly complex processors to expand their capabilities rather than continuous iterations.
If successful, the new AI chip could provide alternatives to NVIDIA’s main products, which currently is estimated to have an estimated 80% market share. Openai design is growing faster than many other chip designers, indicating that the company has made great strides in hardware innovation. Technology giants like Microsoft and Meta have spent years trying to develop their own AI chips, and the results are different.
TSMC will reportedly use its cutting-edge 3-nanometer process technology to manufacture OpenAI’s AI chips. The design combines a shrink array architecture, high bandwidth memory (HBM) and a wide range of network capabilities, similar to NVIDIA’s existing chips.
While OpenAI’s first-generation chips may be able to train and run AI models simultaneously, it is expected to initially deploy it on a limited scale, focusing primarily on model reasoning rather than full-scale training.
(with Reuters input)
