Ukraine plans domestic AI computing capacity with Kyivstar

Ukraine plans to build domestic computing capacity for artificial intelligence with Kyivstar, the company said on Friday, as the country tries to harden critical infrastructure during the war.

Kyivstar said it had signed a memorandum of understanding with the Economy Ministry at the Ukraine Recovery Conference ‌in Gdansk, while ⁠parent ⁠VEON would provide financial backing for a first phase that Kyivstar CEO Oleksandr Komarov said ​could need at least 3-5 megawatts of capacity and tens of millions of dollars.

"The ​biggest consumer of Ukrainian AI right now is the military," Komarov told Reuters. "You cannot run military computing somewhere outside. It is a matter of ​national security."


The plan reflects a wider European push ⁠to reduce ‌reliance on foreign technology infrastructure, a concern that has ​grown more ​urgent in Ukraine after Russia's invasion forced the country ⁠to depend heavily on Western providers.

That shift has also ​changed where Ukrainian data is stored.

Microsoft Europe, Middle East and Africa Vice President Jeff Bullwinkel said at the conference that Ukrainian data was moved to data centres across Europe after the invasion to shield it from Russian strikes, underscoring how the war reshaped the country's digital systems.

Komarov said Ukraine's current demand for artificial intelligence computing was still ‌limited but strategically important, adding that Kyivstar could help deliver services to local businesses that may be too small to ​attract global ​cloud providers directly.

At the ⁠same event, Nvidia Central and Eastern Europe business development director Patrycja Sokalska-Pomacho said Ukraine lacked the computing infrastructure needed to keep the value of its operational, cultural and language data at home.

Reuters reported in December that Ukraine and Kyivstar were developing an artificial intelligence model using Alphabet-owned Google's open-source Gemma, part of a broader effort to support military and civilian operations as demand for secure local processing grows.