Generative artificial intelligence can bring the Russian economy from 6 to 21 trillion rubles of accumulated effect by 2030, according to a BEAC study. However, to realize this potential, it is not enough for a business to buy access to neural networks: it will have to rebuild work processes, train employees and ensure data security. While banks and technology companies are already making money from GenAI, many businesses are only trying out its capabilities and assessing the risks. What are they?
The introduction of generative artificial intelligence (GenAI) by Russian businesses can provide the country's economy with an additional increase of 2.8 to 9.8% of GDP by 2030. This forecast is made by Sberbank's Center for Macroeconomic Research (CMI) in its study "Economics of GenAI".
The calculation is based on Russia's nominal GDP for 2025 in the amount of 214.3 trillion rubles, and includes the cumulative effect (added value and impact on related industries). We are talking about the share of the automation potential of work tasks, and not about the number of companies that will start using generative AI.
According to Alexander Isakov, head of the Central Research Institute of Sberbank, the range of estimates depends on the pace of technology spread and how quickly companies can rebuild their work processes. The lower limit assumes the implementation of about 30% of the technically possible automation of work tasks, the upper one – about 70%.
Generative AI is models that create new: text, program code, images, speech, video. They are given a task in ordinary language, and in response they receive a draft of the contract, a letter to the client, a piece of the program or a squeeze from three hundred pages of documentation, explains Roman Koposov, deputy director of the strategic consulting company ARB Pro.
"Banks and telecom companies put GenAI in contact centers, write and check the code and documents with it, and give employees internal assistants. So far, the industry is mainly looking for regulations and technical documentation, incidents are being investigated, instructions are being prepared, tenders and contracts are being handled.
The industry has its own history, which is little talked about. Experienced engineers and technologists retire, and their knowledge goes with them. If we translate this knowledge into the corporate database now, the young engineer will ask the question in his own words and receive an answer with a link to the required point in the regulations," says Koposov.
Generative AI is used, for example, by Sibur and Norilsk Nickel. "In e-commerce, generative AI is used to prepare product descriptions and images, communicate with customers, and help them choose purchases. For example, Wildberries provides sellers with the opportunity to create images of clothes on virtual models, and Yandex uses generative AI to select and compare products based on customer requests," says Kirill Chernovol, a researcher at the Gaidar Institute's Laboratory for the Analysis of Best International Practices.
The development of this can give a significant increase to Russia's GDP – from 2.8% to 9.8% by 2030. At the same time, the implementation of the most optimistic scenario depends on a number of factors, the dynamics of which is still difficult to predict.
"The figures of 2.8% and 9.8% are not the share of AI in the economy and not an increase in one year. This is the effect accumulated by 2030: from 6 to 21 trillion rubles, if we count from the GDP of 2025. In terms of a year, this results in about half a percentage point to two points of additional growth," explains Koposov. The upper limit, in his opinion, looks bold, because the effect appears only when all the work is adapted to the technology.
"The spread is due to the difference between the two scenarios. In the first case, companies will be cautious and implement GenAI slowly, in the second case they will act quickly and really widely.
To reach 10%, companies need not just to buy licenses, but to rebuild their work processes, train their staff, and integrate GenAI into legacy systems. The main barriers here are the lack of qualified personnel, resistance to changes within,
the lack of ready–made industry solutions for specific tasks and issues related to data quality," adds Dmitry Demidov, Head of the Norbit Innovation Laboratory (LANIT Group of Companies).
So far, the implementation of GenAI in Russia has been uneven: some companies are already actively using its capabilities, while others are only eyeing the technology.
"As usual, large technology companies, banks, individual industrial holdings are in the lead, they have resources, tasks, and data. They use GenAI in development, analytics, and customer service. E-retail is actively experimenting with content generation and chatbots. Medium and small businesses, on the contrary, have neither the budgets nor the expertise to implement pilots. The public sector is moving cautiously, with a focus on security," Demidov says.
Koposov divides companies into three groups. The first one is already making money on GenAI: banks, telecom, IT, e-commerce. They have a lot of digital data, a lot of similar operations, and strong development teams. Sberbank, Yandex, T-Bank, and MTS are developing their own language models, and marketplaces allow sellers to generate product descriptions.
"The second group of companies is piloting and eyeing: retail, logistics, pharma, large oil and gas and energy companies. There is money and data, but the processes are difficult, and the cost of error is high.",
– the expert believes.
According to him, industry, construction, agriculture, most of the public sector, medium and small businesses are at the beginning of the road. "If an industrial company uses AI, it is more often for document management and search according to regulations. Few people reach production management. In small businesses, AI is used by the owner and individual employees through public chatbots. It's useful, but it doesn't change the company's economy," Koposov continues.
As the use of generative AI increases, so will the importance of the risks associated with it. Among them, Demidov highlights the dependence on a particular provider and security threats: information leaks, abuse and the creation of malicious content. "An AI agent with access to mail, databases, and payments is becoming a new point of attack. You will have to protect yourself with engineering: access rights, contour delineation, control of the model's actions," Koposov warns.
"The impact on the labor market is also critical. Adapting to the rapid reduction of routine professions and the slow emergence of new ones will definitely not be easy," Demidov notes.
However, Koposov does not expect mass unemployment with the current shortage of personnel. "There won't be many layoffs. Companies will simply stop hiring new people in contact centers, content management, and junior positions in development and accounting. It's more likely to help an economy where there aren't enough hands," he says.
"And perhaps the main fear is related to the fact that even the creators of AI models cannot always explain the solutions of artificial intelligence.",
– Demidov thinks.
"The model makes mistakes confidently and plausibly. While the AI helps one employee, the error is caught by a human. When AI is embedded in the process, the error has time to spread through thousands of documents before anyone notices it," says Koposov.
According to him, the main risk for an individual company is to be late. "You can buy a model, but you can't buy a redesigned process. Whoever rebuilds the work in the next two years will receive a cost and speed by 2030 that competitors will not achieve by buying the same model," the source concludes.
Olga Samofalova
