Firebird launches a major AI data center in Armenia with NVIDIA GPUs and DSX infrastructure

Firebird, an emerging AI cloud operator, has opened what NVIDIA describes as the largest AI factory in the Commonwealth of Independent States region, positioning Armenia as a new hub for large-scale AI computing. The facility is built to supply the computing capacity used to train, fine-tune, and deploy AI models across industries, and it is intended to serve developers, startups, enterprises, universities, and public institutions within the region rather than relying solely on global AI services.

The opening ceremony brought together Armenia’s prime minister, Nikol Pashinyan, along with Kazakhstan’s deputy prime minister, Zhaslan Madiyev, and David Allen, the U.S. chargé d’affaires, a.i. in Armenia. NVIDIA also said the project reflects ongoing investment and cooperation in frontier computing infrastructure across multiple countries in the CIS region.

Firebird’s stated buildout plan calls for deploying more than 70,000 NVIDIA Rubin and Blackwell GPUs and delivering 300 megawatts of AI infrastructure capacity in Armenia by the end of 2027. Rubin and Blackwell are NVIDIA’s AI-focused GPU platforms designed for accelerating deep learning workloads and large-scale model development. At that scale, the company said the project’s operational focus includes the energy efficiency needed to convert power into usable computing at an economically sustainable pace.

NVIDIA said the factory is based on its DSX platform, which NVIDIA describes as a codesigned approach that integrates accelerated computing, networking, and the power and cooling systems needed to run AI workloads as a coordinated system. NVIDIA also claimed that DSX can run up to 40% more GPUs on the same footprint, supporting higher throughput and more “tokens per dollar” while improving utilization of each megawatt of capacity.

Firebird’s Armenia facility is deployed using Dell Technologies’ high-performance AI infrastructure, with NVIDIA saying the system is built on Dell PowerEdge servers. The power portion of the facility was supported by Schneider Electric, which provided components that include medium- and low-voltage switchgear, three-phase uninterruptible power supply (UPS) systems, and rack enclosures, aimed at keeping the electrical buildout aligned with the compute installation schedule.

For cooling, NVIDIA said Vertiv supplied a chilled-water architecture combining chilled-water technology, advanced controls, and Vertiv TrimCooler for heat rejection. Vertiv’s iCOM CWM Chilled Water Manager is described as a centralized controller intended to coordinate cooling resources and improve visibility, efficiency, and responsiveness as AI workloads shift in intensity.

NVIDIA also highlighted that the project was delivered in just over six months, framing it as proof of Firebird’s ability to convert large infrastructure targets into operational capacity quickly. The company said that speed is part of Firebird’s broader expansion strategy, which includes an approximately 2-gigawatt AI infrastructure roadmap spanning Armenia, Kazakhstan, and additional markets.

Beyond the buildout, the announcement also addressed funding and capacity expansion. NVIDIA said it intends to invest in Firebird, following an earlier investment by CoreWeave this year. NVIDIA did not outline investment terms in the blog post, but it tied the additional backing to Firebird’s goal of expanding its global infrastructure and operational footprint, including the development of large compute clusters across frontier markets.

The stated demand indicates include AI-native companies, with NVIDIA citing Perplexity as an early customer. NVIDIA said Perplexity is working with Firebird to access high-performance AI infrastructure for Perplexity’s AI agent platform and answer engine. NVIDIA also cast the factory as part of a wider effort to make the CIS region more attractive to global AI companies that both build and run models locally, while also encouraging local ecosystem growth.