Today, the world is experiencing a true “gold rush” in artificial intelligence. Models are catching up with one another faster than we can update apps or get acquainted with newly created AIs. The market has become so aggressive that even giants cannot always keep up with the pace. And while OpenAI’s ambitious Sora is phasing out its old lineup and the product in its current form, Anthropic’s Claude is already performing some computer-based tasks on your behalf.
All of this is pushing governments to consider whether they should create their own AI and secure at least a degree of digital independence. Ukraine is no exception. A few days ago, a vote concluded in the Diia app to name the Ukrainian AI. Ukrainians chose the name “Syaivo” (“Glow”).
However, is Ukraine truly ready to create its own AI right now? How critical is this for a country at war? And could this ambitious project turn into yet another costly PR move?
Timeline of the Ukrainian AI Development
Before answering these questions, it is worth recalling how the Ukrainian large language model (LLM) “Syaivo” has been developed and what stage it is currently at.
The creation of the Ukrainian LLM began in 2025. On June 17, 2025, the Ministry of Digital Transformation and Kyivstar announced the start of joint development of a national LLM. At this stage, the foundation of the partnership was laid: no state funds are being used, while Kyivstar became the strategic partner and operational executor of the development.
The next step was forming the computational base. On August 7, 2025, the AI Factory initiative was introduced. This is a separate state project aimed at building its own infrastructure for AI services. Key AI solutions developed by the WINWIN AI Center of Excellence are expected to run on this infrastructure (primarily an AI assistant in Diia and an AI tutor in Mriia). Later, it will also be used for AI projects in defense, medicine, and science. As reported, AI Factory is an entire technological facility with GPU clusters, server halls with liquid cooling, data storage systems, and more, which is expected to operate in Ukraine.
Moreover, in November 2025, the Ministry of Digital Transformation announced cooperation with NVIDIA. Thanks to this, the AI Factory infrastructure will be based on NVIDIA accelerated computing and will become part of a broader strategy for building sovereign AI in Ukraine.
On December 1, 2025, the developers decided on the “brains” of the system. Gemma 3 — an open model by Google — was chosen as the base for training. This made it possible not to build a foundational model from scratch but to focus on adapting the Gemma 3 family to the needs of the state and Ukrainian users. Thus, at the beginning of 2026, the team started preparing data and adapting the model to the Ukrainian language, context, and use cases.
At the end of March 2026, a vote was held in Diia, where more than 22,000 Ukrainians chose the name “Syaivo” for the national AI.
The Ministry of Digital Transformation reported that the system’s beta launch is scheduled for late spring. However, it is not yet specified whether this will be a full launch of the neural network or only the first limited stage of beta testing.
Expert Opinion on Ukrainian AI
UA.NEWS spoke with digital technology expert Oleksandr Hlushchenko about Ukrainian AI. He believes that creating a national model for Ukraine is a logical and necessary step:
“Every country that looks to the future strives to build its own model. This allows it to tailor the system to its needs, train it on its own data, and be confident that this data will not go beyond internal networks. Moreover, nothing in this world is truly free — especially in artificial intelligence. If we use external services instead of our own autonomous system, we always pay for it: for electricity, tokens, processing resources, and the margins of someone else’s business. And we even train it at our own expense.”
The key advantage of having a domestic model is full autonomy. Unlike global services, the Ukrainian model will operate on its own hardware within the country. The expert noted that the Ministry of Digital Transformation conducted the necessary tenders for equipment procurement last autumn.
Additionally, Hlushchenko emphasizes the issue of data protection:
“The global situation shows that digital infrastructure, like data, must be physically protected. No one will do this better in Ukraine than we ourselves. We probably have the richest experience on the planet in surviving and maintaining connectivity during a full-scale war. The Ukrainian experience is a highly valuable consulting service for all civilized countries that have unstable neighbors.”
Concerns about energy shortages, according to him, are exaggerated. AI does not consume more energy than a typical factory, so “there is no need to build a nuclear power plant for it.” The solution lies in hybrid systems and batteries:
“For a data center, there are several types of power supply: the general grid, its own batteries, and alternatives such as solar or diesel. In Ukraine, the energy storage business is currently developing rapidly, where powerful batteries are placed in specialized ‘sarcophagi.’ Energy is stored there when it is cheap or self-generated, and later used or sold when needed. Therefore, from an energy security perspective, the most effective solution for Ukraine will be chosen.”
Hlushchenko also notes that the value of “Syaivo” for Ukrainians will lie in its narrow specialization. A model integrated with government services and citizens’ data may not be interesting globally, but for Ukraine it is key to efficiency. For example, it would allow working with documents in Diia with maximum accuracy, without the need to constantly rewrite prompts, or enable additional services.
We also asked the expert whether the state could become an AI monopolist in the country. Hlushchenko is skeptical about this:
“This is a myth. A monopoly cannot exist in this field because each model is tailored to a specific task. A model that can do everything well does not exist and never will. It is extremely expensive. Any ‘all-in-one machine’ always performs worse than a specialized tool.”
According to him, AI represents a scientific and technological revolution comparable to the emergence of the internet. However, he warns against excessive trust in AI, noting that neural networks often “hallucinate,” and one should not forget about “data intoxication.” The expert predicts that after a rapid peak in popularity, there will be a slight decline both in the number of market players and in attitudes toward models, which will push the world to rely more on humans:
“Those who know how to work with information will be in demand. They will be able to correctly formulate tasks for systems and verify the results so as not to look foolish. Humans remain the main filter that checks what the machine produces.”
The most painful issue for Ukraine’s AI sector today is how to retain developers when companies like Apple or OpenAI offer enormous salaries. Hlushchenko says that in many cases, a non-market but effective factor is at play:
“I won’t reveal a secret. Many talented specialists in Ukraine remain in certain, not highly paid positions for one reason — draft exemption. The problem is that a specialist with 20–30 years of experience does not come out of nowhere; they are extremely valuable. Human capital is the most valuable resource — it cannot be reproduced, replaced, or found in a matter of minutes. Scientists, specialists, innovators, and experts are now being competed for by all global corporations. Even in Ukraine, the shortage of personnel is a negative phenomenon across all sectors without exception.”
However, draft exemption is only a temporary solution and will not solve the global problem of talent outflow after the war and a just peace for Ukraine. Hlushchenko adds that this must be addressed now — strategically, not just tactically.
“We may be able to build a kind of ‘galley’ that exports code. But this is a flawed solution, especially in the context of neural network development, which shows its inefficiency. The goal must be global — to create a product or service in Ukraine, by Ukrainians, that can be sold worldwide. The state’s task is to create conditions for building and developing ecosystems,” the expert emphasizes.
Hlushchenko is convinced that patriotism alone will not be enough to retain specialists in Ukraine. What is needed are simple and transparent tax conditions, a functioning — not manual — judicial system, and protection of property and business. Only then will global giants like Amazon, Microsoft, Google, or Netflix enter Ukraine and establish full-fledged development centers. This will allow Ukrainian developers to create world-class products while staying at home. As a country, Ukraine competes with others for money, investments, and resources. Therefore, its strategy must span at least 10–20 years, not just the term of current officials who may not care about what happens afterward.
After the statements of the Ministry of Digital Transformation and expert forecasts, one thing is clear: the launch of Ukrainian AI is unlikely to take years. The question now is whether the state will have enough resources, political will, and consistency to give the project a long life.
