A window of opportunity: Armenia as a hub for academic AI — interview with Grant Khachatryan

A few days ago, a data processing center equipped with a new high-performance supercomputer was opened at Yerevan State University. This event has become one of the notable steps in the development of the country’s digital and scientific-technological infrastructure. In the near future, Firebird AI plans to launch an even larger computing center, and by the end of the year, the second phase of this project is expected to be completed, offering expanded capabilities and significantly increased power.

Against the backdrop of these developments, a key question arises: what new prospects and practical opportunities are opening up for Armenia with the emergence of such centers and the growth of computing capacity? These and other topics were discussed on the program “Did You Know?” by Grant Khachatryan, head of the machine learning research group at Yerevan State University.

Not to compete, but to create our own
According to Khachatryan, when artificial intelligence is discussed today, the focus is mostly on using existing models—those developed by major players like OpenAI, Google, and others. However, he emphasized that it is also possible to work toward creating one’s own solutions. This does not mean that countries like Armenia should try to compete with them in building universal models—he described that as “not the most reasonable idea.” Rather, the focus should be on developing solutions for specific, specialized tasks.

To develop such solutions, he explained, appropriate computing resources are essential. “As long as they are lacking, possibilities remain limited. It ties your hands—you cannot reach a higher level of quality in AI development,” he said.

He gave the example of a laboratory at Yerevan State University, where limitations were felt even at the research level. In particular, when working on AI systems for processing satellite imagery, the main challenge was consistently the lack of sufficient computing power.

Cloud rental: expensive, unpredictable, and not always available
In theory, computing power can be rented from cloud providers like Amazon or Google, but this is expensive and unpredictable—prices can surge sharply during periods of high demand—and sometimes simply unavailable, with companies responding that there are no free resources regardless of how much one is willing to pay.

Khachatryan emphasized that having one’s own supercomputer provides full control over the resource and the opportunity to learn how to manage it. He noted that this is a separate skill in itself—understanding how such systems operate and what challenges arise in their administration.

At Yerevan State University, this stage is currently underway: a management system for the supercomputer is being developed, along with mechanisms for granting access to different research groups. This also creates an opportunity to invest in human capital—training specialists who can effectively work with such мощности in a university environment. Khachatryan added that having such a resource in an academic setting seems to give researchers in this field “wings,” describing it as the first important effect.

Computing resources as a “magnet”
But the advantages go further. A supercomputer makes Armenia attractive to international researchers and graduate students, including those from abroad. In the past, it was extremely difficult to identify what could draw talented scientists to the country, as high salaries in Big Tech outweighed many factors. However, computing resources have proven to be a powerful incentive.

Khachatryan explained that graduate students and researchers who want to remain in academia—publishing papers and pursuing fundamental science rather than moving to companies like Google—often cannot find sufficient computing power anywhere in the world. Even at top U.S. universities, resources are limited because demand is so high.

Creating an environment in Armenia where a university supercomputer guarantees consistent access to resources opens a “window of opportunity.” Armenia is now actively using this to attract talent and develop its own expertise in artificial intelligence.