Nvidia is planting flags in two countries most people couldn’t locate on a map without squinting, and that’s precisely the point. Armenia and Kazakhstan are each building massive AI computing facilities powered by Nvidia hardware, part of a broader strategy to spread the company’s infrastructure footprint well beyond the usual suspects in Silicon Valley, Northern Virginia, and East Asia.
The combined investment across both projects could reach $10.5 billion, positioning two post-Soviet nations as unlikely but potentially significant players in the global AI compute race.
Armenia’s $500 million supercomputer bet
The Armenian project was announced on June 11, 2025, during Nvidia’s GTC Paris and VivaTech event. It’s a public-private partnership valued at $500 million, aimed at building the Caucasus region’s first major AI supercomputer.
The facility will run on Nvidia’s Blackwell GPUs and is scheduled to go live in 2026, with plans to scale beyond 100 MW of power capacity.
Armenian Prime Minister Nikol Pashinyan and Nvidia CEO Jensen Huang both voiced support for the initiative. The project leverages Nvidia’s DSX platform, which provides what the company calls “AI factory blueprints,” essentially turnkey frameworks for countries looking to stand up sovereign AI compute capacity without building everything from scratch.
Kazakhstan goes bigger with $10 billion Data Center Valley
Kazakhstan unveiled plans for a “Data Center Valley” in the Pavlodar region on June 15, with total investment potential reaching $10 billion. The initial phase alone carries a $5 billion budget. The facility is targeting 125 MW of capacity and a 2027 launch, with deployment plans for up to 100,000 advanced Nvidia GPUs, including the GB300 and future Vera Rubin architectures.
Deputy Prime Minister Zhaslan Madiyev projected the facility could generate over $3 billion in annual export revenues once operational.
Nvidia’s sovereign AI infrastructure strategy
The DSX platform, referenced in both the Armenian and Kazakh announcements, packages Nvidia’s hardware with software frameworks, networking, and deployment guidance, lowering the barrier for countries that want to build AI infrastructure without relying entirely on US or Chinese cloud providers.