Russia’s largest bank, Sberbank, is planning to work with Chinese researchers on artificial intelligence (AI) projects, a top executive told Reuters. This comes as China’s DeepSeek disrupts the AI industry by developing advanced models at a fraction of the cost of U.S. competitors.
Russia and China have long discussed AI cooperation, including in military technology, but details remain unclear. Under CEO German Gref, Sberbank has transformed from an old-style Soviet bank into a leading AI player, launching its GigaChat model in 2023.
“Sberbank has many scientists, and we plan to conduct joint AI research with Chinese experts,” said Alexander Vedyakhin, Sberbank’s First Deputy CEO. He did not disclose which Chinese institutions would be involved.
DeepSeek, a startup based in Hangzhou, recently shook the global AI market, with investors betting its low-cost models could challenge U.S. tech leaders like Nvidia. A potential AI partnership between Russia and China—nations seen by Washington as key rivals—could further disrupt the industry as the AI race between China and the U.S. intensifies.
Presidents Vladimir Putin and Xi Jinping share a common view of the West as declining, while China challenges U.S. dominance in AI, quantum computing, and military power. Western sanctions on Russia following the Ukraine war have driven Moscow and Beijing closer, with Putin recently calling China an “ally.”
Vedyakhin emphasized that scientific collaboration between the two nations could strengthen their AI advancements. Putin has instructed Sberbank, which faces Western sanctions, to boost AI partnerships with China and other BRICS nations to counter U.S. influence. However, details on Russia’s AI progress remain unclear, as some projects are classified.
Both Russia and China face challenges in AI development due to restrictions on advanced Western technology imports. Russia ranks 31st in the Global AI Index, far behind the U.S., China, and even fellow BRICS members India and Brazil.
Since DeepSeek launched its models, Sberbank has compared them to its own GigaChat MAX. The results showed DeepSeek performed better in scientific applications, while GigaChat remained competitive in banking. Vedyakhin noted that DeepSeek’s success aligns with Sberbank’s focus on cost-effective AI solutions, rather than massive investments in computing infrastructure.
“DeepSeek has shown that even with limited resources, AI models can rival American technology. This challenges the need for enormous spending on AI megaprojects,” Vedyakhin said.
Like DeepSeek, Sberbank has made many of its AI tools publicly available, including its text-to-image generator, Kandinsky, and its base AI model, GigaChat Lite. Vedyakhin pointed out that DeepSeek’s openness contrasts with the secrecy of OpenAI, which has helped attract a strong developer community.
As Russia and China deepen their AI collaboration, the global AI competition is set to become even more intense.
