Russian banking giant Sberbank is exploring the use of Chinese-manufactured microchips to support its flagship AI model, GigaChat, as Western sanctions continue to restrict access to advanced computing hardware. The bank’s CEO, German Gref, indicated that procuring chips from China is part of efforts to sustain Russia’s development of large-scale artificial intelligence systems.
Speaking during President Vladimir Putin’s visit to China, Gref said, “We are hoping that we will be able to use Chinese microchips for GigaChat,” highlighting Russia’s strategic pivot toward Eastern suppliers amid geopolitical constraints that have limited Russian access to leading Western semiconductors.
GigaChat is a generative AI model developed by Sberbank as part of broader initiatives to establish Russia’s presence in cutting-edge AI technology. However, Russia has lagged behind global leaders such as the United States and China in both AI research and the high-performance computing infrastructure required to train and deploy large models. The country’s heavy reliance on imported electronic components — particularly semiconductors — leaves it vulnerable to export controls and sanctions that target key technology sectors.
China, now Sberbank’s primary source of advanced electronics, has accelerated its own efforts to develop AI hardware, including semiconductors designed for machine learning workloads. Among the chips attracting interest is Huawei’s Ascend 950, which represents one of China’s most sophisticated AI processors. While the Ascend 950 is seen as a significant step forward in domestic chip development, it continues to trail behind the performance levels of Nvidia’s H200, a leading Western model widely used in AI research and industry applications.
Gref did not specify the exact chip models that Sberbank is pursuing, but emphasised the bank’s commitment to securing suitable hardware to support GigaChat’s computational needs. He noted that competition for Chinese chips is intense, with major Chinese internet companies — including ByteDance, Tencent and Alibaba — placing substantial orders for the same AI processors. This competition could add complexity to Russia’s efforts to source adequate AI hardware.
The broader context for Sberbank’s initiative reflects longstanding challenges facing the Russian technology sector. Sanctions imposed by Western governments have targeted Russia’s access to advanced semiconductors, cloud infrastructure and other foundational elements critical to AI development. In response, Russian entities have increasingly looked toward China as a partner for AI-related technology imports, although China’s domestic offerings remain behind the global cutting edge in some advanced computing capabilities.
By turning to Chinese chips, Sberbank aims to circumvent some of the limitations imposed by sanctions and sustain progress on GigaChat, a symbolic and strategic project in Russia’s AI ambitions. The success of these efforts will hinge not only on securing suitable hardware but also on navigating a competitive global market where demand for high-performance AI chips continues to accelerate.
