Translation. Region: Russian Federation –
Source: Phosagro – Phosagro – An important disclaimer is at the bottom of this article.
Moscow. February 26, 2026. Denis Novikov, CEO of Apatit JSC (part of the PhosAgro Group), discussed PhosAgro's best practices in developing its own artificial intelligence infrastructure at the Russian Union of Industrialists and Entrepreneurs' Russian Business Week Digital Forum.
The discussion was attended by the President of the Russian Union of Industrialists and Entrepreneurs (RSPP) Alexander Shokhin, Co-Chair of the RSPP Coordination Council for Technological Leadership and Industrial Policy, First Deputy General Director of the non-governmental development institute Innopraktika Natalia Popova, Chair of the RSPP Electric Power Commission, General Director of RUSENERGOSBYT LLC Mikhail Andronov, Commercial Director of YADRO (part of ICS Holding) Alexander Bakulin, General Director of BASIS LLC David Martirosov, Senior Managing Director – Director of the Infrastructure Solutions Department of Sberbank PJSC Karen Karapetyan, and Director of the Federal State Autonomous Institution Digital Industrial Technologies Eduard Shantayev.
In his opening remarks, RSPP President Alexander Shokhin noted that artificial intelligence is changing the economy and social sphere faster than institutions, from education to the labor market, can adapt. He added that just five to seven years ago, it was difficult to imagine the scale of AI's penetration into the economy and everyday life.
"We're not yet fully prepared for all the challenges we're already facing and will face in the coming years. Currently, personnel needs are forecasted for seven years based on current trends, but it's truly difficult to estimate the scale of AI penetration in the future. Therefore, we need to make our personnel training system as flexible as possible to enable rapid adaptation," the RSPP president stated.
Alexander Shokhin also outlined the infrastructure constraints that are currently determining the development of AI. These include, first and foremost, the growing demand for electricity and the need to expand data center capacity. He noted the discussion about the possible shift of generation and data center infrastructure to the Urals and Siberia, noting that such decisions require consideration of technological and network limitations, as well as an economic model, as data center investments must be cost-effective and recoup their costs.
The forum was moderated by Mikhail Oseevsky, a member of the RSPP Board of Directors, co-chair of the RSPP Coordination Council on Digitalization, Chairman of the RSPP Committee on Digital Economy, and President of PJSC Rostelecom. He added that artificial intelligence is becoming one of the main consumers of electricity worldwide, so Russia also needs to consider building specialized data centers for AI and ensuring their reliable power supply. He noted that the forecasts for growing electricity demand under discussion reach tens of gigawatts. Even if some of these estimates are overstated, the scale of future needs remains extremely significant. He also noted the trend of placing AI infrastructure closer to the points of data generation and processing: "Infrastructure is gravitating toward the locations where information is generated and processed—toward megacities where large corporations are headquartered and consumers live, and this trend is likely to intensify."
Denis Novikov, CEO of Apatit JSC (part of the PhosAgro Group), discussed PhosAgro's best practices as it implements its own AI transformation strategy, actively integrating technology into key production, service, and corporate-wide business processes.
"We initially chose to build infrastructure and develop our own AI solutions. This ensures the complete security of commercial, production, and internal information, and also allows us to be independent of foreign software while accumulating our own data. This data is then used to further train large language models, making this work even more efficient," Denis Novikov emphasized.
According to the CEO, the company utilizes several AI-based technologies, including machine learning, video analytics, and generative artificial intelligence. Large language models are deployed within the company's internal IT infrastructure, including Sber's flagship domestic model, GigaChat Max. This already allows for resource savings, increased efficiency and productivity, and enhanced employee safety.
PhosAgro's computing capacity is concentrated in two of its own data centers—in Kirovsk and Cherepovets. A third data center is being built in Volkhov in collaboration with Rostelecom. A key advantage of this approach is flexibility: the modular architecture of the data centers allows for capacity expansion in line with PhosAgro's growing digital needs. Furthermore, virtually all of the company's production sites use on-site electricity, which also meets the needs of the data centers.
"Our goal is to make AI a core technology for all employees and integrate it into all business processes. We are integrating AI solutions, from predictive analytics to AI agents, into all import-substituting information systems," noted Denis Novikov. "Our company's successes are being watched with interest not only by our colleagues in the agrochemical industry but also by representatives from other sectors of the economy. I am confident that through close collaboration and partnership, which is developing at the RSPP sites and within industrial competence centers, we will be able to more quickly scale such solutions and reduce development costs."
Summing up the discussion, forum participants agreed on the need for cooperation among leading technology players to achieve synergies, as well as the formation of a constructive alliance between equipment manufacturers and software developers that will accelerate compatibility and the commercialization of digital solutions.
Please note: This information is raw content obtained directly from the source. It represents an accurate account of the source's assertions and does not necessarily reflect the position of MIL-OSI or its clients.