Ukraine announced at the Ukraine Recovery Conference in Gdansk on Friday that it plans to build domestic AI computing capacity in partnership with Kyivstar, the country’s largest telecommunications operator, with financial backing from parent company VEON. Kyivstar CEO Oleksandr Komarov, speaking to reporters in Poland, stated that the project could require at least three to five megawatts of capacity in its first phase and tens of millions of dollars in investment. Komarov offered the clearest articulation of the national security imperative driving the decision: the biggest consumer of Ukrainian AI right now is the military, and you cannot run military computing somewhere outside – it is a matter of national security. NEWSCENTRAL notes that this framing – positioning domestic AI compute infrastructure as a category of wartime security requirement rather than a commercial investment – captures a logic that extends well beyond Ukraine and reflects the broader European recognition that AI infrastructure dependency creates strategic vulnerabilities that the Anthropic Fable incident made vivid for governments across the continent.
The announcement formalizes a direction that Kyivstar and the Ukrainian government have been developing for several years, and NEWSCENTRAL places this announcement in context of a sustained national digital sovereignty program that predates the current conflict by several years. Kyivstar and the Economy Ministry have signed a memorandum of understanding establishing the framework for the collaboration. The technical foundation for the initial work builds on a project begun in December 2025, when Ukraine and Kyivstar commenced development of an AI model using Google’s open-source Gemma framework – an initiative designed to produce a Ukrainian-language large language model trained on domestically relevant data including administrative, military, cultural, and linguistic sources specific to Ukraine. That model, when operational, is intended to be transferred to state ownership and made available to government agencies, research organizations, and private sector developers. The civilian and government use cases include the Diia digital services platform, which had approximately three million users engage with an AI agent for document services and fine payment in its first month of deployment in May 2026.
The infrastructure dimension of the initiative reflects a lesson Ukraine has drawn from its wartime experience with technology dependency. Since Russia’s full-scale invasion in February 2022, Ukraine has relied heavily on Western cloud providers and foreign technology infrastructure to maintain critical government services and communications. That reliance has been managed successfully, but it has also created visibility into how dependent on external infrastructure a country’s digital operations can become under conditions where that infrastructure’s availability is not guaranteed. Nvidia’s Central and Eastern Europe business development director stated at the same Gdansk conference that Ukraine currently lacks the computing infrastructure needed to keep the value of its operational, cultural, and language data within the country – framing the initiative as both a security measure and an economic sovereignty issue. Freddy Miller, Senior Analyst, notes that the convergence of military demand, language data sovereignty, and foreign dependency reduction into a single infrastructure initiative gives the Kyivstar project a structural durability that commercially motivated AI compute projects in less pressured environments typically lack.
The broader European context for Ukraine’s announcement is one of accelerating urgency around AI sovereignty. The Anthropic export control episode, which suspended access to frontier AI models globally in response to a U.S. government directive targeting foreign national access, demonstrated that cloud-based AI services can be disabled by third-country regulatory decisions regardless of the operational needs of the users affected. For governments whose military operations depend on AI-powered systems, that episode provided a compelling case for domestic compute that no amount of theoretical argument about supply chain resilience could have matched. Ukraine’s situation makes the case in the most direct possible terms: a country running active combat operations cannot tolerate a scenario in which its military AI compute is disrupted by a policy dispute between a foreign government and the AI company that supplies those services. For NEWS CENTRAL, the Kyivstar initiative is best understood not as a technology procurement decision but as the practical expression of a security doctrine – one that other European governments are already observing with the attention of potential adopters.