Kazakhtelecom plans to launch its first 50 MW data center at Data Center Valley in Ekibastuz in the second quarter of 2027. The capacity of the project's first phase, totaling 215 MW, has already been fully reserved, the company said. This was reported by Qazaqyia.kz citing Kursiv Media.

NVIDIA also announced the expansion of Kazakhstan's computing infrastructure together with Kazakhtelecom and Firebird AI in its official blog. The company presented plans for deploying graphics processing units, supporting local startups, and developing AI applications.

The announcements coincided with the first visit of NVIDIA founder and CEO Jensen Huang to Kazakhstan on October 23, 2026. In Astana, he met with President Kassym-Jomart Tokayev and also took part in an open conversation with Deputy Prime Minister and Minister of Artificial Intelligence and Digital Development Zhaslan Madiyev at the Alem.AI international center.

Kazakhtelecom Chairman of the Board Bagdat Mussin presented the Data Center Valley project to Huang, describing the site's infrastructure capabilities and energy advantages. The data center cluster is being built with the prospect of increasing energy capacity to 1 GW.

"Our task is to turn Kazakhstan from a consumer of computing resources into one of the global suppliers of capacity for artificial intelligence. We want the technologies of the future not just to come to Kazakhstan, but to be created and developed here," Mussin said.

According to NVIDIA, Kazakhtelecom has already deployed 400 NVIDIA H200 and 480 NVIDIA B300 graphics processors in Astana and Almaty. In 2027, the company plans to place more than 1,000 GPUs at Data Center Valley and another more than 1,000 NVIDIA Vera Rubin accelerators in the Ekibastuz area.

Firebird, which previously created its first "AI factory" in Armenia, has received a US export license to supply 45,000 NVIDIA Grace Blackwell GPUs to Kazakhstan. The first phase of the project is being built jointly with Kazakhtelecom at the Data Center Valley site. Its commercial launch is scheduled for the third quarter of 2027.

Taking into account the planned second phase, the total potential capacity of Firebird's AI infrastructure in Kazakhstan could reach about 500 MW. A signing ceremony for the company's investment agreement with the Government of Kazakhstan was held at the Akorda. Firebird co-founder and CEO Razmig Hovaghimian took part in it.

Firebird also intends to create Firebird Labs Kazakhstan to support local AI startups, research, and training of specialists. The laboratory will provide funding, access to computing resources, and opportunities for research partnerships.

Higgsfield AI, founded by natives of Kazakhstan, is also expanding its own infrastructure. The company is increasing capacity based on NVIDIA Vera Rubin through cloud providers as its models, products, and demand for media content generation applications grow. During the event in Astana, NVIDIA's investment in the startup was also announced. Huang explained the decision primarily by the quality of the team and the scale of the founders' ambitions.

In a conversation with Madiyev, the head of NVIDIA suggested that Kazakhstan use its energy resources as the basis for a new export direction.

"Instead of just exporting energy, in the future you could convert energy into intelligence and export intelligence," Huang said.

Madiyev, in turn, emphasized that Kazakhstan needs artificial intelligence that understands the Kazakh language and takes into account the country's cultural characteristics. Huang noted the importance of national data, calling it a natural resource reflecting the culture, history, and values of the state.

The interlocutors also discussed Physical AI — artificial intelligence for robots and other systems interacting with the physical world. Madiyev identified industry, mines, agriculture, transport, and logistics as potential areas of development. Huang supported the idea of turning Kazakhstan into a center for training such models.

Speaking about security, the head of NVIDIA described OpenShell — a solution for restricting AI agents' access to files and tools and monitoring their actions. He advised young specialists to study not only how to use AI but also the principles of how it works, its strengths, and limitations.