Russian companies have made a qualitative leap in the use of large language models (LLMs). In the first half of 2026, the volume of foreign neural network consumption on the Yandex AI Studio platform reached 597 billion tokens—18 times more than in the same period a year earlier. This figure speaks volumes: AI has ceased to be an experimental toy and has become a working tool for business.
Chinese models are squeezing out competitors
Analyzing the demand structure, I see a clear trend: Russian business is betting on Chinese developments. Yandex's own models account for 54% of consumed tokens, but Qwen (21.2%) and DeepSeek (13.5%) firmly hold second and third places. American GPT-OSS from OpenAI only manages 7.5%, indicating a redistribution of preferences in favor of Asian vendors.
Even more telling is the dynamics on other cloud platforms. On the MWS GPT platform, consumption for the half-year grew 11 times relative to all of 2025, reaching 400 billion tokens. Here, Qwen is the absolute leader with 261.1 billion tokens, followed by GLM (91.2 billion) and Kimi (81.4 billion), which has noticeably displaced DeepSeek. I record a similar picture on Cloud.ru, where demand increased 18 times over eight months, with a significant share of requests coming specifically from Chinese models.
Market scale: estimates diverge
The LLM market in Russia is growing at an impressive pace. According to my calculations, based on player data, by the end of 2026 it will grow by about 35% and reach approximately 19.6 billion rubles. The cloud segment, including subscriptions and API access, is estimated at nearly 1.5 billion rubles. However, not all participants agree with such a conservative estimate. For example, Turbo Cloud (a Rostelecom structure) believes that the real commercial capacity of cloud LLM access will be at least 10–12 billion rubles.
This dispute fits into the broader picture: the country's entire AI market grew by 29.7% in 2025 to 316.1 billion rubles, and by 2030 the baseline scenario assumes growth to 830 billion. Businesses are increasingly renting capacity and ready-made solutions rather than purchasing equipment. Neural networks are already being integrated into HR processes (used by 49% of employers), and the largest banks are earning hundreds of billions of rubles from AI—Sber alone estimated the effect for 2025 at over 475 billion.
My comment: An 18-fold growth is not just statistics but a marker of market maturity. Russian companies have moved from point pilots to industrial AI deployment, and China has turned out to be the main beneficiary of this transition. The question is not whether the market will grow, but whether domestic vendors can hold their share in the fight against aggressively scaling Asian models. I believe that in the next two years we will see consolidation and intensified competition for corporate clients.