Chinese technology companies are rapidly shifting away from Nvidia chips toward homegrown alternatives, marking a significant turning point in the global AI hardware landscape. ByteDance, the parent company of TikTok, is in active negotiations to purchase AI chips from Shanghai-based Iluvatar CoreX and is also considering a deal with Baidu, according to sources familiar with the discussions. This move reflects a broader trend where Chinese chipmakers have captured nearly 41% of China's AI accelerator server market, effectively squeezing out the American chip giant that once dominated the region.

Why Is Nvidia Losing Ground in China?

Nvidia's market share in China has effectively fallen to zero, according to the company's CEO Jensen Huang. This dramatic decline stems from U.S. export controls on advanced chips, which have prompted Beijing to aggressively promote the use of locally developed semiconductors to improve technological self-reliance. Chinese chipmakers have seized this opportunity, developing alternatives that are becoming increasingly viable for real-world applications. The shift is not theoretical; it is happening now with major technology companies making concrete purchasing decisions.

ByteDance's potential deal with Iluvatar CoreX would represent a major commercial milestone for the Shanghai-based company. Iluvatar CoreX is expected to ship at least 50,000 chips to ByteDance this year, with most of them used for inference workloads, which involve answering user queries rather than training new AI models. Until now, Iluvatar CoreX has primarily supplied government procurement projects, so a deal with one of China's biggest technology companies would validate the company's technology at scale.

What Are the Key Players in China's AI Chip Ecosystem?

The competitive landscape in China's AI chip market includes several major players working to reduce dependence on foreign semiconductors:

• Iluvatar CoreX: A Hong Kong-listed GPU startup that reported 1 billion yuan (approximately $148 million) in 2025 revenue, with about 90% coming from GPU sales. The company's Tiangai series chips are designed for AI training, while its Zhikai series targets inference tasks.
• Huawei: Already serving as one of ByteDance's major domestic GPU suppliers, demonstrating that established tech giants are competing in this space.
• Cambricon: Another existing supplier to ByteDance, showing that multiple Chinese companies are building viable chip alternatives.
• Baidu: Developing Kunlunxin chips that are being considered by ByteDance and already used by Tencent, indicating broader adoption across the industry.

Iluvatar CoreX's growth trajectory is particularly striking. The company is projected to reach 3.04 billion yuan (approximately $449.8 million) in revenue this year, with total chip shipments expected to jump 139% to over 100,000 units. The Zhikai inference chips have an average selling price of about 12,000 yuan, or roughly $1,775 each, making them a cost-competitive option for companies deploying AI services at scale.

How Are Chinese Companies Expanding AI Chip Capacity?

The expansion of domestic chip production is accelerating rapidly. Tencent's Chief Strategy Officer James Mitchell noted in May that Chinese AI chips would become available in large quantities during the second half of this year. This timeline is critical because it suggests that Chinese companies will have sufficient supply to move away from dependence on foreign chips within months, not years. The shift from training chips to inference chips is particularly important because inference represents the ongoing operational cost of running AI services like ByteDance's Doubao chatbot, which serves millions of users daily.

ByteDance is also considering using Baidu's Kunlunxin chips, and Tencent is already a Kunlunxin customer, according to sources. This diversification strategy suggests that Chinese tech companies are intentionally building relationships with multiple domestic suppliers to reduce single-vendor risk and ensure stable supply chains independent of U.S. export restrictions.

What Does This Mean for the Global AI Hardware Market?

The shift toward domestic chips in China represents a fundamental restructuring of the global AI hardware market. For years, Nvidia's graphics processing units (GPUs) were the default choice for AI work because they offered superior performance and a mature software ecosystem. However, the combination of U.S. export controls and improving Chinese alternatives has created a situation where major technology companies can now meet their needs without relying on American chips. This is not a temporary workaround; it is a strategic pivot that will likely persist even if trade tensions ease.

The financial stakes are enormous. ByteDance is one of China's largest technology companies and a major spender on AI infrastructure. A successful deal with Iluvatar CoreX would validate Chinese chip technology for one of the world's most demanding use cases: running a global social media platform with hundreds of millions of users. Success here could accelerate adoption across other Chinese companies and potentially influence purchasing decisions in other countries seeking to reduce dependence on American technology.

The details of the potential deals are not final and remain subject to change, according to sources familiar with the negotiations. However, the direction of travel is clear: Chinese companies are building the infrastructure to operate independently of Nvidia, and that transition is happening faster than many observers expected.