- Статьи
- Internet and technology
- The Intellectual Revolution: how Russian developers are changing the information technology market
The Intellectual Revolution: how Russian developers are changing the information technology market
Russian developers are transforming the IT technology market. New services have been introduced in the last few days. This is Pavel Durov's Cocoon AI computing network and the Alice AI neural network developed by Yandex. Details can be found in the Izvestia article.
Infiltrate the cocoon
A few days ago, Telegram founder Pavel Durov announced the launch of a network for cheap AI computing. He presented a project called Cocoon at the Blockchain Life 2025 forum, held in Dubai.
The new network, as the entrepreneur noted, will allow developers to use artificial intelligence without investing significant amounts in their infrastructure. Thus, users will be able to run neural networks without their own servers. The computing power will be provided by the owners of the video cards participating in the project.
All data, according to Durov, will be encrypted, so that no one can see what specific information is being processed by computers. This is how Cocoon differs from other AI models, where user data is stored on servers and can be provided on request.
The name of the network itself is an abbreviation and stands for Confidential Compute Open Network. This translates to "Confidential Computing Open Network". Durov is convinced that the creation of decentralized technologies is the only way to preserve "digital freedoms."
Telegram is expected to be the first major client of the network. The messenger will use the system for confidential requests to artificial intelligence.
The launch of the Cocoon project is scheduled for November 2025. The exact date, however, was not disclosed. It is only known that this will happen after processing the applications received from the volunteers.
Alice in Wonderland
On October 28, Yandex held a conference dedicated to the new product, presenting the Alice AI neural network. It is positioned as a universal neural network capable of solving any problem using a new response format — pictures, videos and information from Yandex Maps.
The presentation was mainly in the format of an announcement of new products. It is expected that in the near future, the updated Alice will be able to remember the user's business, remind them of important plans and run errands with the help of AI agents.
Despite the fact that most of the features are only at the development stage, users are enthusiastic about the new service. The Alice AI app has moved from eighth place to first in the Russian AppStore in less than a day. It was ranked third on Google Play.
We have noticed the new neural network even abroad. Aman Chadha, an AI developer at Apple and Amazon, expressed his opinion about Alice. He congratulated "Yandex" with the launch of the project and expressed the opinion that the company now makes sense to add in the number who prepares the next frontier in the development of artificial intelligence, along with OpenAI, Google and DeepSeek.
"Exchange" of capacities
The Cocoon project can reduce the cost of renting computing power, as it will combine free computer resources around the world, creating an "exchange" for them, believes Sergey Nazarenko, head of Reksoft's data management and machine learning technology development department. This, he expects, will stimulate competition and give companies more technological independence.
— As a result, in the future, this may push the IT industry towards a model where it is possible to flexibly combine local clouds for storing important information and cheaper distributed capacities for complex computing. This reduces costs and accelerates the creation of new digital products using generative technologies, since companies do not need to invest heavily in their own infrastructure from the very beginning, the expert believes.
Cocoon democratizes access to computing power through a distributed GPU network (a server equipped with video cards that can handle high loads, which makes it possible to process large amounts of data faster), notes Stanislav Yezhov, Director of AI at Astra Group. This is especially important for startups and small businesses, as the cost of infrastructure in this case is reduced to the level of operating costs.
A decentralized network will provide access to resources for small independent developers who do not have the finances to buy and rent their own computing power, confirms Dmitry Ovchinnikov, information security architect and vCISO UserGate.
— Theoretically, the computing power that is currently in the hands of users and is not used in daily work is comparable to the capacity of large data centers. Such an approach will not only dramatically expand the availability of computing power, but will also create the necessary reserve in terms of capacity," the source points out, adding that all this will have a very positive impact on the development of AI models and the entire field.
However, it is necessary to wait for the first real performance and cost indicators, since Chinese models today often turn out to be cheaper due to optimization, draws attention to Sergey Stepanishchev, head of Getmobit business process automation.
Penetration into everyday life
Alice AI, trained on real user scenarios, shows the growth of Russian expertise in NLP (a general field of artificial intelligence and mathematical linguistics that studies the problems of computer analysis and synthesis of texts in natural languages) and multimodal models, Stanislav Yezhov points out. This is proof of scalability for the mass segment.
More than half a million downloads of Alice AI show that the Russian consumer is ready for competitive AI products, the expert is convinced. The AI software market in Russia, he expects, will reach 50 billion rubles in 2025 with an average annual growth of 32.5%.
"Alice AI" in this context rather affects the user segment, believes Sergey Stepanishchev, since it is a massive AI assistant for everyday tasks. It shows that neural networks continue to "dive" into a person's personal space.
"Given that there is an Alice speaker in almost every home, it can be concluded that users no longer think about how much AI has entered their lives," the Izvestia interlocutor believes.
The rise of Alice AI may spur massive demand for agent-based scenarios and multimodality out of the box, which will accelerate the implementation of AI functions in end-user services, Sergey Nazarenko expects.
— However, the latest AI services have yet to be tested for sustainability and economic relevance. Warming up mass demand will force other companies to actively implement AI in their services in order to keep up. This will boost the development of assistants in IT solutions for businesses, too," he is confident.
Task with an asterisk
However, there are also difficulties that will have to be faced when implementing services. Dmitry Ovchinnikov believes that a serious problem in the case of Cocoon is obtaining a credit of trust both to TON and to the computing network itself. Since January 1, 2025, the cryptocurrency has almost tripled in price, he recalls.
—If there is no trust and ease of use, then a good technical organization will not be able to attract a sufficient number of participants," the expert is convinced.
At the same time, as the number of consumers of computing power grows, the number of those who rent out these capacities should also grow, he clarifies. Otherwise, consumers may face a shortage of resources and return to more conservative ways of training AI models.
Another problem that Cocoon has is the immaturity of decentralized networks, which can lead to performance problems when scaling, Stanislav Yezhov draws attention to.
In general, if the service is implemented according to the general principles of the API (a set of rules and protocols that allow different software applications to interact with each other), the technical difficulties of integration should be minimal, but it will take time to assess real performance and stability, Stepanishchev points out.
In the case of Alice AI, the security of artificial intelligence plays a special role, Yezhov emphasizes. There is a possibility of confidential data leakage, lack of end-to-end encryption, and configuration vulnerabilities.
—Critical systems require auditing at all levels of the stack, from data processing to tokenization, which requires careful selection of the vendor for the AI solution," the source said.
It is also important to take into account that Alice AI has serious competitors and limitations, and for large-scale implementation in a corporate environment, it is necessary to understand what advantages the integration of this AI assistant will bring to companies, in addition to user scenarios, Stepanishchev notes.
In general, it's too early to talk about difficulties, Ilya Petukhov believes. It is necessary to see how all these products will take root in the market.
— But what will definitely happen is the need to comply with regulatory requirements and safety standards. And also take into account the difficulties of integrating new technologies into existing systems and ensuring compatibility with various devices and operating systems," the specialist summarizes.
Переведено сервисом «Яндекс Переводчик»
