ByteDance is developing new artificial intelligence chips with InnoStar Semiconductor, according to reporting from The Information. The effort adds to a growing list of Chinese technology companies and chip designers working to build more of the hardware needed to train and run AI systems.

The reported collaboration comes as AI chip supply has become a strategic issue for companies around the world. Advanced processors are essential for powering large language models and other compute-heavy AI tools, but access to top-tier chips has been constrained by tight supply and, in some markets, export restrictions. For major technology firms, developing custom silicon can help reduce dependence on outside suppliers and improve performance for specific workloads.

ByteDance, the owner of TikTok and one of China’s most prominent internet companies, has already shown interest in expanding its AI capabilities. A move into chip development would fit a broader trend among large tech firms that want more control over the infrastructure behind their AI products. While the source material does not provide details on the design, specifications, or production timeline of the chips, the partnership suggests ByteDance is exploring a more direct role in the semiconductor stack.

InnoStar Semiconductor is known as a chipmaker, and the reported partnership indicates it may be contributing technical expertise or manufacturing support to the project. The source does not say how the work will be divided between the two companies, whether the chips will be designed for training or inference, or if they will be used internally by ByteDance or sold more broadly.

The report also compares the effort to chips associated with Groq, a company that has attracted attention for its AI-focused hardware and ties to Nvidia through investment activity. That comparison points to the kind of specialized accelerators companies are racing to build for AI inference and other demanding tasks. However, the source material does not say that ByteDance’s chips are intended to match Groq’s architecture or performance.

China’s technology industry has been investing heavily in domestic chip development as companies seek alternatives to foreign suppliers. That push has intensified as AI demand has surged, increasing interest in local semiconductor partnerships. A ByteDance and InnoStar collaboration would align with that broader national and commercial effort, even as the specific scope of the project remains unclear.

For now, the reporting suggests that ByteDance is taking another step beyond software and services into the infrastructure layer that supports AI development. If the effort progresses, it could give the company more flexibility in how it deploys AI systems and potentially help it manage costs and supply constraints.

Neither company has publicly detailed the project in the source material, and many questions remain about the chips themselves. Still, the reported partnership underscores how central custom AI hardware has become to the strategies of major tech firms competing in the next phase of artificial intelligence.