Russian Businesses Turn to Chinese AI Models as Demand Surges

Russian businesses are rapidly increasing their use of Chinese artificial intelligence models, drawn by their lower cost and ability to run on domestic infrastructure as Moscow seeks to develop a homegrown “sovereign AI” industry, the Kommersant newspaper reported Thursday.

Customers of Russia’s largest cloud platforms have sharply increased their use of models including Qwen, DeepSeek, GLM and Kimi in 2026, according to data provided to the newspaper.

The trend highlights Russia’s growing technological reliance on China as Western sanctions and security concerns limit access to U.S.-developed technology. It also exposes the difficulties facing Moscow’s effort to build competitive AI systems of its own.

MTS Web Services Cloud said customers used 400 billion tokens generated by Chinese models in the first half of 2026, 11 times more than in all of 2025.

Qwen was the most widely used Chinese model on the platform, accounting for 261.1 billion tokens. GLM and Kimi generated 91.2 billion and 81.4 billion tokens, respectively.

Yandex AI Studio said Qwen was also the most popular foreign model on its platform, with a 21.2% share of token consumption. DeepSeek ranked second with 13.5%, while U.S.-developed GPT-OSS models accounted for 7.5%.

Cloud provider Cloud.ru recorded an 18-fold increase in AI model use during the first eight months of 2026 compared with the same period last year. The company did not disclose absolute figures but said Chinese models accounted for a significant share of demand.

Artur Samigullin, head of Yandex AI Studio, told Kommersant that consumption of Chinese models had increased more than fivefold over the past year.

Chinese models are generally cheaper and are regarded as safer for Russian companies because they can be deployed on domestic infrastructure, allowing businesses to retain control over sensitive data.

“The main format is deploying the model on Russian infrastructure,” Alexander Tugov, director of Selectel’s AI division, told Kommersant. “In this format, the data remain within the company’s systems, and the choice of model is purely technical and does not depend on security requirements.”

That arrangement is particularly important for companies handling sensitive information. Dmitry Cheklov, founder of the Hybrid advertising technology ecosystem, said many businesses were guided by operating costs, where Chinese models offered a clear advantage.

The rise of Chinese AI comes as Russian developers seek to create a domestic market for what officials call sovereign AI. Under proposed rules, Russian models would be assessed for compliance with the country’s officially defined “traditional spiritual and moral values.”

Only models that pass the review would be eligible for designation as national or sovereign. That status would grant access to government procurement contracts, significant IT projects and data held in state information systems.

Sberbank CEO German Gref said last month that Russian tech giant Yandex was further training Qwen as the basis for its AI development, though Yandex later denied that claim.

Natalya Kaspersky, head of cybersecurity company InfoWatch, has said Russia would struggle to catch up with the United States and China in developing large foundation models because such systems require vast financial, computing and energy resources.

“When you are walking along the same tracks on which trains are traveling, you have essentially no chance of catching up with that train,” Kaspersky told the RTVI television network in July. “I do not think we have any realistic chance of creating something groundbreaking in large foundation models.”