Insider Brief
- Firebird is advancing the second phase of its AI supercomputing project in Armenia, securing U.S. export licensing to deliver an additional 41,000 Nvidia GB300 GPUs and bringing the total investment to roughly $4 billion.
- The expansion builds on a $500 million first phase and is intended to support large-scale AI training and inference, positioning Armenia as one of the world’s largest AI GPU clusters by installed capacity and a trusted destination for advanced U.S. AI hardware.
- Firebird said the project reflects a broader strategy to deploy U.S.-designed AI infrastructure in aligned markets, combining global commercial demand with local research, enterprise use, and long-term ecosystem development in Armenia.
Firebird is moving ahead with the second phase of its AI supercomputing project in Armenia, expanding a buildout that is quickly turning the country into a significant node in the global AI infrastructure map. The expanded deployment pushes the overall effort to roughly $4 billion and places Armenia among the world’s largest AI GPU clusters by installed capacity, according to the company.
The U.S.-based AI cloud and infrastructure company announced it has secured U.S. export licensing and regulatory approvals to deliver an additional 41,000 Nvidia GB300 GPUs to Armenia, a step that materially increases the scale of the project.
Firebird said the approvals confirm the project’s compliance with U.S. export controls and signal confidence from U.S. regulators in Armenia as a destination for advanced AI hardware.
“I hope that close and transparent cooperation established between Armenia and the United States in the high-tech sector will allow us to further develop and strengthen the mutually beneficial partnership between Armenian and U.S. companies,” Prime Minister of the Republic of Armenia Nikol Pashinyan said in a statement.
Phase Two builds on a previously announced $500 million first phase, which established Armenia’s initial high-performance AI computing footprint. Together, the two phases are intended to support large-scale AI training and inference workloads, as well as research in areas such as life sciences, robotics, space systems and advanced simulation, Firebird said.
The announcement came during a U.S. delegation visit to Yerevan and highlighted the geopolitical dimension of modern AI infrastructure, where access to advanced chips is tightly regulated and increasingly aligned with broader technology and trade policy. Firebird said the project reflects a strategy of exporting U.S.-designed AI infrastructure into markets viewed as trusted partners under U.S. regulatory frameworks.
“This means new markets and new jobs, both for the American workforce and companies, as well as for Armenia,” U.S. Vice President JD Vance said. “These are chips that simply do not exist in most countries in the world, they are now going to be developed, and the data centers using those chips are going to be built in Armenia thanks to the Prime Minister’s leadership.”
The company positions the Armenia deployment as both a commercial and ecosystem play, aimed at anchoring a long-term technology hub built around secure infrastructure, international partnerships and local talent development. Firebird said the cluster is designed to serve global customers while also enabling domestic research and enterprise use inside Armenia.
“Firebird extends U.S. AI technology leadership globally, aligned with our vision of enabling AI for the benefit of all,” co-founder and CEO of Firebird Razmig Hovaghimian said in the announcement. “This new cluster establishes Armenia as a global supercomputing hub, demonstrating how trusted U.S. infrastructure can power emerging economies. We’re grateful to the U.S. and Armenia governments for their partnership in enabling American technology to operate globally at scale.”
Image credit: Firebird
