ByteDance is exploring AI chip deals with Iluvatar CoreX and Baidu as Chinese tech firms increasingly turn to domestic semiconductor suppliers amid U.S. export restrictions.
![]() |
| ByteDance's discussions with Iluvatar CoreX and Baidu signal growing confidence in China's AI chip industry as domestic suppliers gain market share. Image: CH |
SHANGHAI, China — June 15, 2026:
ByteDance is in talks to buy artificial intelligence chips from Shanghai-based Iluvatar CoreX and is also considering a similar arrangement involving Baidu's Kunlunxin chips, according to sources familiar with the discussions. While no agreement has been finalized, the negotiations point to a notable shift in China's rapidly evolving AI hardware market.
The company behind TikTok has become one of China's biggest spenders on AI infrastructure. Its growing investment in AI products, including the popular chatbot Doubao, has created rising demand for computing power, particularly for inference tasks that generate responses to user requests.
If a deal with Iluvatar CoreX moves forward, the startup would become ByteDance's third major domestic GPU supplier after Huawei and Cambricon. That would mark an important commercial breakthrough for Iluvatar CoreX, which has so far been known primarily for supplying government-linked projects.
The talks also suggest that Chinese technology companies are becoming increasingly comfortable relying on homegrown chipmakers. For years, Nvidia dominated China's AI accelerator market. However, U.S. export restrictions have limited access to some of Nvidia's most advanced products, creating opportunities for local competitors.
That opportunity appears to be growing. Chinese AI chip developers have steadily improved their products and expanded production capacity, allowing them to compete in areas where demand is rising quickly. Inference computing has become one of those areas, as AI applications move from development to large-scale deployment.
Sources said Iluvatar CoreX is expected to ship at least 50,000 chips to ByteDance this year, with most of them intended for inference workloads. Those workloads typically require less computing power than AI model training, making them a practical entry point for emerging chip suppliers.
ByteDance is also evaluating chips from Baidu's Kunlunxin unit. The consideration is significant because it shows that major Chinese technology companies are increasingly exploring multiple domestic suppliers rather than depending on a single vendor.
Tencent has already adopted Kunlunxin chips for some of its operations, according to one source. That trend indicates broader industry acceptance of locally developed AI hardware.
The potential agreements come as Beijing continues to encourage greater use of domestic semiconductor technology. Policymakers have repeatedly emphasized the importance of technological self-reliance, especially in strategic sectors such as artificial intelligence and advanced computing.
The shift is already affecting market dynamics. Chinese GPU and AI chip manufacturers captured a substantial share of the country's AI accelerator server market last year, gradually reducing Nvidia's influence in what was once one of its most important overseas markets.
For ByteDance, securing additional chip suppliers could help ensure stable access to computing resources as competition in AI intensifies. For companies like Iluvatar CoreX and Baidu's Kunlunxin, winning business from a technology giant of ByteDance's scale would provide both revenue and credibility.
While the discussions remain subject to change, they reflect a broader transformation underway in China's technology sector. Domestic chipmakers are no longer being viewed simply as alternatives. Increasingly, they are becoming central players in the country's AI ambitions.
