The Russian artificial intelligence market is on the verge of tectonic changes. The government is actively working on a draft decree that could radically change the rules of the game for all market participants — from large corporations to state structures. The focus is on introducing a mandatory priority for sovereign, that is, fully Russian, AI models over foreign counterparts.
The essence of the legislative initiative
According to the document, which is currently at the discussion stage, organizations will be obliged to choose domestic AI solutions if their cost does not exceed foreign analogues by more than 1.5 times. Notably, the initiative comes from representatives of Sber, who are actively promoting this idea in working groups on AI regulation under the government.
Special emphasis in the draft is placed on the public sector. Sovereign models are proposed for use in strategic planning, development of state programs and national projects, as well as in the creation of regulatory acts. In addition, the list of priority areas includes industries with increased reliability requirements: tax and customs services, critical information infrastructure, education, healthcare, and public services.
Market reaction and systemic contradictions
The industry met the initiative with mixed feelings. The Ministry of Digital Development confirmed the fact of discussion, while Sber insists that the document is consolidated and reflects the position of the entire working group, not just the bank. Yandex, in turn, emphasizes the critical importance of maintaining competition in the market — without it, the quality of domestic developments could seriously suffer.
Interestingly, the measure fits into the state's overall logic of stimulating technological sovereignty. In parallel, authorities are preparing tax incentives for developers of national models, including an investment deduction and accounting for expenses at triple the amount. However, there is a fundamental contradiction here: on the one hand, the market is to be closed off from foreign solutions, and on the other, active work is underway on compatibility standards so that developers do not create closed ecosystems.
Hardware foundation
Without affordable equipment, any requirements will remain mere declarations. In this context, the cooperation between MWS AI and Rubytech is telling — they have already tested AI systems on Chinese GPUs as an alternative to Nvidia solutions. The commercial launch of such systems is planned for 2027 — it is then that it will become clear how realistic it is to fulfill import substitution requirements in practice.
Analyst's comment: The initiative looks like a logical continuation of the course toward technological sovereignty, but it raises serious concerns about the quality and competitiveness of domestic models. In world practice, forced import substitution in high-tech areas rarely leads to the creation of innovative products — usually it stimulates stagnation and rising costs. The key success factor will be how quickly Russian developers can offer truly competitive solutions, rather than merely "meeting the need" formally.