ByteDance , the parent company of TikTok and China’s Douyin , appears to be accelerating its efforts to build a more independent artificial intelligence infrastructure , increasingly turning to domestic semiconductor makers . According to information cited by Reuters, the tech giant is in advanced talks with Iluvatar CoreX , a growing Chinese GPU company based in Shanghai, to supply tens of thousands of artificial intelligence chips .
The potential deal highlights a broader shift underway in the Chinese tech market, where major companies are attempting to reduce their reliance on American technologies, especially after export restrictions imposed by the United States on Nvidia.
Iluvatar CoreX enters the spotlight
Until now, Iluvatar CoreX has been known primarily for its sales to government agencies and state projects. However, a potential partnership with ByteDance could radically change its business profile.
See also: Zhang Yiming: ByteDance founder becomes China's richest man
Market sources say ByteDance is considering sourcing at least 50,000 GPUs by 2026, which would be one of the largest commercial orders in Iluvatar’s history. For a company that reported revenue of around 1 billion yuan (about $148m) last year, a deal of this size could act as a catalyst for further growth.
At the same time, the successful implementation of such a demanding project will also be a significant test for the company's technology, as operating artificial intelligence services on a scale of hundreds of millions of users requires extremely high levels of reliability and performance.
Doubao's role in growing demand
A central role in this strategy appears to be played by Doubao, ByteDance's artificial intelligence chatbot that is experiencing rapid growth in the Chinese market.
As the number of users and queries to the system increases, the need for more computing power becomes increasingly urgent. The information indicates that most of the chips in question will be used for inference, that is, for generating answers from already trained AI models.
This process is less demanding on computing resources than training new models, which allows domestic manufacturers to compete more effectively with Nvidia's solutions.
An ecosystem of domestic suppliers
Iluvatar CoreX isn't the only company on ByteDance's radar. The company is already working with Huawei and Cambricon to supply AI accelerators, and is also looking at Baidu's Kunlun chips .
See also: ByteDance's Jimeng AI available on Android and iOS
This strategy reveals a conscious effort to diversify suppliers and reduce risks arising from potential geopolitical developments or supply chain.
At the same time, the creation of a multidimensional ecosystem of Chinese manufacturers could enhance the competitiveness of the domestic semiconductor industry, which in recent years has been the focus of significant state investments.
The challenge against Nvidia
Despite the significant advances made by Chinese companies, Nvidia still maintains a technological lead, particularly in the more demanding tasks of training large language models.
series is presented as a competitor to Nvidia's A100 and A800, but the real test will come when these systems are called upon to operate in high-scale production environments.
Experts point out that the theoretical performance of a chip is not enough. Stability, energy efficiency, software compatibility and the ability to support complex workloads are equally critical factors for success.
See also: ByteDance: Lays off hundreds of employees from its gaming division
The new era of Chinese artificial intelligence
The potential deal between ByteDance and Iluvatar CoreX reflects a deeper shift taking place in the global AI landscape. China is no longer just seeking to use advanced technologies, but also to develop its own alternatives at every level of the value chain.
If the talks are successful, ByteDance will gain even greater flexibility in managing its AI infrastructure, while Iluvatar will gain a landmark customer that could accelerate its market penetration. In any case, the direction is now clear: Chinese tech giants are investing aggressively in building a self-sufficient AI ecosystem, gradually reducing their dependence on foreign suppliers and shaping a new balance of power in global technology competition.
