Huawei is taking its competition with Nvidia in the global AI market to a new level by deciding to expand its Ascend series of advanced AI chips beyond China. The company will officially launch its Ascend 950 chips in the South Korean market starting in 2026. This move directly targets international companies that are struggling to access Nvidia products or are seeking alternative infrastructure due to US technology restrictions.
Huawei is taking its chips out of China
Huawei Korea CEO Balian Wang announced at an official event in Seoul that they will begin offering AI computing boards and data center solutions to South Korean customers next year.
Wang emphasized that Huawei’s strategy is not just about selling individual components, but rather providing “end-to-end” solutions that combine networking, storage, and high-performance computing. The company positions these integrated cluster systems as a strong alternative to Nvidia’s market dominance.
The Ascend 950 series, to be launched in the South Korean market, comes in two different variants depending on the intended use. The 950PR model is optimized for computationally intensive tasks such as recommendation systems and data preprocessing, while the 950DT model focuses on training and inference processes of large language models thanks to its high memory bandwidth.
Huawei promises data transfer speeds reaching 4 TB/s in these new generation chips using its own developed high-bandwidth memory (HBM) technology. Furthermore, the chips are offered in massive server clusters called Atlas SuperPods, increasing the system’s scalability.
This expansion strategy is not limited to hardware. Huawei also plans to make its self-developed, now open-source, HarmonyOS operating system available to businesses in Korea.
Although there is no sign of a turnaround in the smartphone market, the goal is to build a new ecosystem through industrial applications and data center software. Huawei’s aggressive move in the global market is being evaluated in the technology world as a “new AI-focused roadmap,” but whether the company’s production capacity can meet global demand remains the most critical question for the coming period.
