The landscape of China’s artificial intelligence server market is undergoing a structural transformation that would have seemed implausible just three years ago. Domestic tech giants Huawei Technologies and Cambricon are on track to capture a combined 80% share of the market by the end of 2026, effectively marginalizing US chipmaker Nvidia, whose presence is projected to shrink significantly. The shift is the most visible evidence yet that US export controls, rather than simply slowing China’s AI ambitions, have catalyzed the emergence of a parallel domestic hardware ecosystem capable of sustaining the country’s AI build-out on its own terms.
According to a new analysis reported by Huawei Central, the combined market share of foreign chip suppliers, including Nvidia and Advanced Micro Devices, is expected to drop from 34% in 2025 to 21% in 2026. The decline is a direct consequence of the progressive tightening of US export restrictions, which have made it increasingly difficult and expensive for Chinese companies to acquire the most capable foreign AI accelerators.
The Domestic Champions Step Up
In stark contrast to the retreat of foreign suppliers, Huawei and Cambricon are experiencing rapid market-share expansion. Their combined share is projected to surge from 46% in 2025 to 56% in the near term, ultimately reaching an estimated 80% by year-end as the full impact of recent export control measures ripples through procurement pipelines.
Huawei’s Ascend series of AI processors has been the primary beneficiary of this shift. The Ascend 910C, which the company has been shipping in volume since late 2024, has demonstrated competitive performance across a range of AI training and inference workloads, offering Chinese cloud providers and enterprise customers a credible domestic alternative. Cambricon, a Beijing-based AI chip designer that listed on the Shanghai STAR Market in 2020, has similarly seen demand for its MLU series of processors surge as customers seek to diversify away from Nvidia dependency.
The Chinese government’s unwavering support for the domestic semiconductor industry has created a favorable environment for both companies. Substantial state investment in research and development, combined with procurement policies that strongly encourage the adoption of homegrown technologies, has accelerated the maturation of China’s AI hardware ecosystem at a pace that few outside observers anticipated.
The Role of Internet Giants
The growth of China’s AI server market is also being driven by the country’s largest internet companies. Firms like Alibaba Group Holding, ByteDance, and Baidu are not only major consumers of AI infrastructure but are also increasingly developing their own application-specific integrated circuits (ASICs). These in-house chip development efforts, which include Alibaba’s T-Head unit and Baidu’s Kunlun processors, are expected to capture a 23% share of the market, up from 20% in 2025.
This trend mirrors the strategies of global hyperscalers like Google and Amazon, which have long recognized the benefits of custom silicon for optimizing performance and reducing costs. For Chinese internet giants, the additional motivation of supply chain security, insulating themselves from the risk of future export control escalations, makes the in-house chip imperative even more pressing.
“These Chinese tech giants are not only the biggest buyers of servers and AI infrastructure, but we also expect them to become the most promising players in in-house chip development going forward,” noted industry analyst Frank Kung.
The Impact of US Export Controls
The dramatic decline in Nvidia’s market share is a clear indication that US export controls are having a tangible impact on the competitive dynamics of the Chinese AI market. While these restrictions were intended to slow China’s progress in artificial intelligence, they appear to have inadvertently accelerated the country’s drive for self-sufficiency. The black market for smuggled Nvidia hardware has simultaneously exploded, with prices for the flagship DGX B300 server doubling to over $1.1 million, a testament to the undiminished demand for Nvidia’s products even as domestic alternatives mature.
By cutting off access to the most advanced foreign chips, the US has forced Chinese companies to invest heavily in domestic alternatives. This has created a massive opportunity for companies like Huawei and Cambricon, which are now stepping up to fill the void. The long-term consequences of this decoupling remain to be seen, but it is increasingly clear that the global AI hardware market is fracturing along geopolitical lines.
Challenges and Future Outlook
Despite the rapid progress made by domestic companies, significant challenges remain. Developing cutting-edge AI chips requires immense technical expertise and access to advanced manufacturing capabilities. While China has made strides in chip design, it still faces constraints in semiconductor manufacturing, particularly in producing the most advanced process nodes. The software ecosystem surrounding domestic AI chips is also still evolving. Nvidia’s CUDA platform has long been the industry standard, and convincing developers to transition to alternative platforms requires significant effort and investment.
Nevertheless, the trajectory is clear. China is determined to build a self-reliant AI infrastructure, and the projected 80% market share for Huawei and Cambricon is a testament to the resilience and ingenuity of the country’s tech sector. As the geopolitical landscape continues to evolve, the battle for supremacy in the AI server market will undoubtedly intensify, and the outcome will shape the global AI landscape for years to come.