The chip solution: How is the modern AI race unfolding?

The world powers are reaching new levels in competition in the field of artificial intelligence. The large car manufacturing corporation Stellantis is integrating the French Mistral AI into its production chain. For the company, this means optimizing processes and speeding up workflow. For Europe, it is an opportunity to abandon AI solutions from the United States. At the same time, the American Open AI in October became the most expensive startup in the world — due to the sale of employee shares, it was valued at $ 500 billion. China is switching to its own hardware and adapting its AI systems to it in the context of a trade war with the United States. Russia also continues to develop its models of generative intelligence. Because of the sanctions, Moscow has to seek cooperation with Beijing and the BRICS countries. What determines the victory in the AI race and how Americans slow down their competitors — in the Izvestia article.

European sovereignty in the field of AI

In early October, the multinational car manufacturing corporation Stellantis, whose concern includes Peugeot, Citroën, Fiat and Jeep brands, announced the introduction of an artificial intelligence model from the French company Mistral AI into its production structure.

This is a remarkable stage in the AI race, which will only grow in the coming years. For the average person, artificial intelligence is about writing texts and advice on caring for animals, but for modern governments, it's about the pace of growth and expansion of power. Whoever is the first to be able to fully integrate AI into the country's economy will be able to increase productivity, reduce costs, create new products, manage infrastructure, and do much more.

For example, in the case of the French Mistral AI, artificial intelligence will be engaged in sales of equipment, engineering and tracking different stages of production. AI will be able to reduce the processing time of requests, automate the workflow in the company and accelerate internal engineering chains. Everyone benefits from this.: Stellantis directly optimizes processes and, therefore, incurs lower costs, and Mistral AI runs its model of generative intelligence in production processes. For Europe, this means one thing: a serious combination of a local AI developer and a large corporation has appeared in the region.

A month earlier, Mistral AI had agreed on another strategic partnership, but this time with the Dutch manufacturer of microelectronics installations ASML. The Dutch invested about €1.3 billion, effectively becoming a key shareholder in the French company. It is important to understand that modern chips play a major role in training artificial intelligence models. They act as the core of microelectronics and allow AI models to perform billions of similar calculations. The faster these operations take place and the cheaper they cost the company, the faster the model learns.

Before that, the Europeans relied on the AI solutions of the Americans. Now Mistral has the opportunity to develop its model on world-class tasks with Stellantis and have a direct industrial partner in ASML. As a result, the European Union can gradually assemble its own AI chain, from financial resources and necessary technologies to implementations in the real manufacturing sector. In the future, the EU will be able to depend less on American capacities and, consequently, strengthen its position in the AI race.

The confrontation between the United States and China

At the same time, the United States remains the absolute leader in the field of artificial intelligence. Private investments in AI alone in the country in 2024 amounted to about $109 billion, which is more than China and Europe combined. Large numbers of data centers are being built in the country, on the basis of which AI models are being trained. Microsoft, Amazon and Alphabet have raised investments in the industry to historical highs.

Another factor in the dominance of the United States is computing power, including graphics accelerators and microelectronics, on which models are trained. American NVIDIA controls the lion's share of the GPU market for data centers, their batches are bought in advance, and the main fleet of such accelerators is concentrated in the United States. The core of AI developers is also located there. Recently, OpenAI was valued at about $500 billion following the sale of employee shares, making it the most expensive startup in the world. This is an indicator of the scale of capital available in the United States.

China has traditionally been the main competitor of the United States in this industry. And Washington is using all available tools to contain it. Beijing is developing its own range of artificial intelligence models — Alibaba Group's Qwen, Baidu's ERNIE, and DeepSeek.

China is very seriously focused on developing its own infrastructure, component base and computing power, Mikhail Komarov, Professor of the Department of Business Informatics at the Higher School of Business of the National Research University Higher School of Economics, told Izvestia. In his opinion, this leads to the development of their own artificial intelligence solutions.

— China has significant amounts of its own data for the development of AI technologies in various fields, for example, for unmanned vehicles, both land, air and surface. All this leads to taking a leading position in the global technological arena," the expert emphasized.

At the same time, it is vital for China not only to switch to its own hardware, but also to make it powerful enough. Therefore, Washington has restricted access to top-end chips for Beijing since 2022. Only weakened accelerator models go to China, and then with interruptions in supplies. This does not allow China to properly scale the training of large AI models, which significantly slows down their speed.

In addition, Chinese microelectronics has not yet caught up with American microelectronics. Beijing was able to reach the 7-nanometer process technology, which significantly affects the speed of the chip, but the United States has been working on 5 nm and 3 nm for a long time. Similar Chinese chips are not manufactured on a large scale, are more expensive, and are not as efficient.

In the context of the technological race between the two powers, Taiwan is notable, and tension has been building around it for years. Formally, the island is the territory of China, but in fact it considers itself an independent state. It is the Taiwanese company TSMC that dominates the production of the thinnest chips — they account for about 60% of the global market. This is a significant technological gap in the current reality, which the United States is trying to win over by involving TSMC in the construction of its factories in Arizona.

The role of Russia and BRICS in the development of AI

Russia is also participating in the AI race. At the state level, Moscow has a strategy for the development of artificial intelligence until 2030, and at the end of 2023, a decree was issued with a course on personnel training, the introduction of AI in public services and industry, as well as in the development of banks and other areas.

Sberbank has its own GigaChat. According to the company's management, it is already used by about 15 thousand organizations in Russia. A new version with improved logic and reasoning is expected to be released in 2025. Yandex is also not far behind — it has its own YandexGPT and AI-Studio products. At the same time, the model branches are updated fairly quickly.

In addition, in Russia, all developers of large language models (LLM) have examples of integrations in industrial companies, Sergey Dutov, Director of Corporate Innovation at the Skolkovo Foundation and ProGenAI project leader, told Izvestia.

— R-Pharm, together with Sber and AIRI, accelerate the process of developing the structure of molecules for new drugs. And KEAZ is experimenting with Russian neural networks YandexGPT, GIGAChat, and Chinese opensource models DeepSeek and Qwen to create local small language models that learn from data from the same workshop, without an Internet connection," he said.

At the same time, the development of the domestic AI industry is hampered by sanctions and export controls from the United States. Rapid development is simply impossible without advanced accelerators and components. This forces the Russian Federation to focus on an out-of-system coalition in the development of AI. Back in January 2025, Russian President Vladimir Putin instructed the government and the National Security Council to develop cooperation with China in the field of artificial intelligence.

Dutov noted that there are only three vendors of large language models in Russia, on the basis of which local and secure solutions can be developed. And only Sber and Yandex continue to develop their own models, while others have switched to affordable Chinese opensource.

The expert suggested continuing to stimulate fundamental research through grants to the country's leading AI centers, supporting domestic LLM vendors, stimulating demand for their products and developing international cooperation. As an example, he cited the agreement between the BEAC and Tsinghua University on the joint development of AI projects.

Cooperation can also develop within the framework of BRICS. In 2024, Putin announced the launch of the AI Development Alliance. Moscow's initiative was worked out at the level of ministries responsible for digitalization and communications. In this case, BRICS acts as a platform where countries agree on their own development rules, bypassing the Western ones.

Currently, the alliance includes national associations and institutes in the field of AI not only from BRICS member or partner countries, but also from other countries, the Russian Foreign Ministry told Izvestia. There are 17 organizations from 14 countries in total.

As Belarus integrates into the BRICS, its cooperation with Russia in the field of technological sovereignty can develop even faster. Minsk is committed to new initiatives and proposals for further cooperation.

"In general, we certainly see the possibility of expanding the format of cooperation as the association is ready to include new members, continuing a constructive dialogue in the interests of both our country and the entire BRICS," the Belarusian foreign ministry told Izvestia.

The general view on the development of the AI industry is reflected in the agreements between the two countries. The common position is based on ensuring "digital" sovereignty, neutralizing risks and threats emanating from the malicious use of ICT, including artificial intelligence systems, the Russian Foreign Ministry said.

The United States is definitely leading the AI race at the moment. They dominate investment, technology, and export controls. Perhaps now everything looks quite unambiguous: Europe is trying to assemble its own AI development unit, and China is moving to technical facilities of its own production, the Russian Federation is relying on its talents and investing in cooperation with partners, while Washington is conquering new heights. However, winning the race for perfect artificial intelligence and its integration into production and other processes will depend not only on technology, but also on who is more successful in imposing their own rules of the game in an ever-changing political environment.