ByteDance, the Chinese tech giant behind TikTok, is exploring a significant shift in how it powers its artificial intelligence systems — turning to domestic chipmakers rather than relying on foreign suppliers like Nvidia.
According to reports cited by Seeking Alpha via MSN, ByteDance is in active talks with Shanghai-based Iluvatar CoreX to purchase AI chips for inference workloads — the computationally intensive process of running AI models to generate responses for users. The company is also evaluating a similar deal with Kunlunxin, the chip unit spun out of Chinese internet giant Baidu.
The move reflects a broader push by ByteDance to expand what CXO Digitalpulse describes as its "domestic sourcing strategy" — a deliberate effort to source critical AI hardware from within China rather than depending on chips that have become increasingly difficult to obtain due to U.S. export restrictions.
Inference chips are distinct from the training chips used to build AI models in the first place. They need to be fast, efficient, and available at scale, since they handle every query a user sends to an AI product. For a company the size of ByteDance — which runs AI features across TikTok, Douyin, and its Doubao AI assistant — the volume of inference demand is enormous.
Iluvatar CoreX and Kunlunxin represent China's growing but still maturing domestic chip ecosystem, which Beijing has prioritized heavily amid escalating U.S. tech restrictions. Whether these chips can match the performance of restricted Nvidia hardware remains an open question — but ByteDance's willingness to evaluate them signals that Chinese firms are increasingly treating domestic alternatives as a viable path forward, not a last resort.