ByteDance is reportedly developing a new AI chip aimed at running artificial intelligence models and expanding its AI infrastructure. According to Odaily, the chip's architecture is inspired by the language processing unit (LPU) of U.S. chip design company Groq, with a focus on cost-effective AI model operation.
Additionally, ByteDance is closely collaborating with InnoStar Semiconductor to explore deep integration of the chip with its AI platform. Industry insiders suggest this move indicates an acceleration in ByteDance's independent hardware strategy, aiming to improve the efficiency of large model operations and reduce operational costs.