GPU instead of Oil: How Firebird is Turning Armenia into an AI Hub

Artificial intelligence is rapidly changing the global economy, and with it, the map of technological influence. While until recently oil and gas were considered key resources, today there is more and more talk about a different shortage — computing power. Against this backdrop, Armenia unexpectedly finds itself among the countries gaining access to one of the most sought-after resources of the new economy. Speaking at Yerevan Dialogue 2026, Alexander Yesayan, co-founder and CEO of Firebird AI, presented the opportunities and potential of this project.

According to him, it is not just about building a data center. It is about access to a resource that will determine the competitiveness of countries in the coming years — computing power.

Trust as a Point of Entry

According to Yesayan, the Firebird project was announced a year ago, but its real significance has only become clear now. A key role was played by a factor that is rarely spoken of directly — trust from the USA. Firebird is an American company, and this is exactly what allowed it to receive permission to host a large AI cluster in a country belonging to the category of BIS licensed countries. Such permissions are issued to very few. This automatically moves the project beyond the framework of an ordinary investment.

Additional momentum, according to him, was given by the TRIPP initiative, which strengthened attention to the region and created a political and economic basis for such projects. For Firebird, this became a signal: the South Caucasus is ceasing to be a periphery and is beginning to be considered as part of a new infrastructure map.

From 6,000 to 41,000 GPUs

Yesayan said that while at the first stage the talk was about launching 6,000 GPUs, the project is already reaching a completely different level. The second phase involves about 41,000 chips and investments of about $4 billion. Such capacities are oriented not towards the local market, but towards the largest global players — the so-called hyperscalers who develop artificial intelligence products. And here a key feature of the current moment manifests itself: the demand for computing resources exponentially exceeds the supply. "Every GPU that works in the world today is, in essence, already sold," Yesayan noted. The first phase of the project will be launched in the coming weeks, and its capacities are already bought out with a large margin ahead. For the second phase, contracts are also effectively formed.

Why Armenia Ended Up at This Point

The choice of the country, according to Yesayan, was not accidental — it was the result of the coincidence of several factors at once.

The first is energy. At the time the decision was made, Armenia was in a unique situation: with the growth of solar generation, a surplus of electricity was formed in the country. There were periods when capacities had to be limited due to excess. For the AI industry, where data centers consume huge amounts of energy, this is a critical advantage.

The second factor is geography and connectivity. Armenia is within a five-hour flight from a number of markets — from Europe to Central Asia. This allows it to be used as a regional hub for the provision of digital services. Thus, the country simultaneously possesses both an energy base and a logistical advantage — a rare combination for such projects.

Access as a Strategic Advantage

According to his estimates, the main effect of the project for Armenia is not the investments themselves or even the infrastructure. The key is access to computing resources. Part of the capacity, according to Yesayan, is reserved for the Armenian side — for the state, universities, research centers, and startups. This fundamentally changes the capabilities of the entire ecosystem. If earlier the limitation for the development of AI projects was precisely the access to GPUs, now this limitation is partially removed.

A Bet on the Ecosystem, Not Just on Business

Firebird, according to Yesayan, is making a conscious choice — not to limit itself to commercial contracts, but to develop the local environment. The company is ready to provide GPUs to startups, research groups, and universities, effectively investing in them through access to infrastructure. This is already beginning to change the educational and scientific landscape.

Negotiations are underway with leading US universities — including Stanford University, Harvard University, and Massachusetts Institute of Technology — on joint research with Armenian universities. The contrast that Yesayan brings is indicative: in these universities, researchers often compete for access to one or two GPUs, whereas Armenia gets the opportunity to work with much larger capacities.