Kazakhstan is moving to deepen its partnership with NVIDIA as the country builds out large-scale artificial intelligence infrastructure and positions Ekibastuz as a potential regional hub for data centers, DKNews.kz reports.
President Kassym-Jomart Tokayev met NVIDIA Vice President Rev Lebaredian on October 1 to discuss the next stage of cooperation. NVIDIA technologies are already used in two major supercomputing clusters in Kazakhstan, while the company is now expected to take part in the next phase of the country’s Data Center Valley project in Ekibastuz.
Gov
The project is expected to start with around 300 MW of capacity, with the potential to scale up to 1 GW.
For Kazakhstan, the partnership goes well beyond buying advanced chips. In June 2026, the government announced agreements with Firebird and NVIDIA worth a combined $10 billion in artificial intelligence and digital infrastructure.
The broader ambition is to create the computing backbone needed for large AI models, high-performance computing and next-generation data centers.
Tokayev and NVIDIA discuss the next phase
During the meeting, Tokayev highlighted NVIDIA’s role in shaping the global artificial intelligence ecosystem and stressed the need to expand the practical use of AI across priority sectors of Kazakhstan’s economy.
The partnership already has a tangible foundation. Two supercomputing clusters powered by NVIDIA technologies are operating in Kazakhstan.
The talks also covered talent development, the broader technology ecosystem and the expansion of computing infrastructure.
Lebaredian outlined NVIDIA’s plans to participate in the next stage of the Data Center Valley project in Ekibastuz.
The project is designed for roughly 300 MW of capacity in its initial stage, with the possibility of expanding gradually to 1 GW. Kazakhstan is expected to provide the site, energy resources and basic infrastructure.
Who is Rev Lebaredian?
Rev Lebaredian has worked at NVIDIA since 2002 and has more than 30 years of experience in computer graphics.
Before joining NVIDIA, he worked at Warner Brothers Digital and Disney Dream Quest Images and later founded Steamboat Software.
Today, Lebaredian leads work related to physical AI simulation and NVIDIA Omniverse.
The platform allows developers to create digital versions of real-world environments and use them to train and test robots, autonomous systems and AI models before deploying them in physical settings.
That makes the technology particularly relevant for industrial automation, robotics and large-scale digital infrastructure.
Why NVIDIA matters so much to the AI economy
NVIDIA was founded in California in 1993 and for many years was best known as a maker of graphics processors for computer gaming.
That changed dramatically as graphics processing units, or GPUs, proved highly effective for training neural networks.
The rapid growth of generative AI transformed NVIDIA into one of the most important suppliers of computing infrastructure for the global technology industry.
In fiscal 2026, NVIDIA reported revenue of $215.9 billion, up 65% from the previous year, while net income reached about $120.1 billion.
More recent Fortune figures covering the 12 months to April 30, 2026 put revenue at $253.5 billion and profit at $159.6 billion. NVIDIA employs around 42,000 people.
As of August 18, 2026, Fortune valued the company at roughly $5.32 trillion.
NVIDIA also stands out for profitability. A Visual Capitalist comparison based on Fortune Global 500 companies found that NVIDIA generated $55.60 in net profit for every $100 of revenue.
For comparison, Microsoft posted $36.10, Alphabet $32.80, Meta $30.10 and Toyota $7.60.
The gap reflects major differences in business models. Digital products and computing platforms can scale with comparatively lower marginal costs, while manufacturers and retailers face ongoing spending on raw materials, production, goods and logistics.
Kazakhstan already operates two NVIDIA-powered supercomputers
One of Kazakhstan’s flagship computing systems is the Alem.Cloud national supercomputer.
The system is built on 512 NVIDIA H200 graphics processors. Its computing resources are available to government agencies, universities, research institutions and private companies.
Alem.Cloud ranked 86th in the November 2025 edition of the TOP500 list. In the June 2026 ranking, it stood at 104th with Linpack performance of 20.48 PFlop/s.
That change does not indicate a decline in the machine’s performance. TOP500 positions shift as newer and more powerful systems enter the ranking.
Kazakhstan’s second major cluster is AI-Farabium, launched by Kazakhtelecom.
It is also based on NVIDIA H200 processors. According to TOP500 data, the system has 57,600 computing cores and delivers 17.93 PFlop/s in Linpack performance. It ranked 122nd in June 2026.
Its computing capacity is also used to support the development of Kazakhstan’s KazLLM and AlemLLM language models.
For local universities, researchers, government institutions and businesses, access to this level of computing power is becoming increasingly important as modern AI models require far more processing capacity than conventional IT infrastructure can provide.
Ekibastuz could become the next major AI hub
The next stage of Kazakhstan’s AI infrastructure push is far larger than the existing supercomputing clusters.
On June 15, 2026, Kazakhstan announced a package of agreements worth $10 billion in artificial intelligence and digital infrastructure. The initiative involves U.S.-based Firebird and NVIDIA.
One of its central projects is the Data Center Valley in Ekibastuz.
Under the proposed cooperation model, Firebird is expected to help build infrastructure for artificial intelligence and high-performance computing, while NVIDIA serves as a technology partner. Kazakhstan will provide the site, energy access and core infrastructure.
The $10 billion figure relates to the agreements already announced. Earlier estimates put the potential investment needed for the full Data Center Valley development at around $30 billion.
The first data center under the project is expected to be completed in 2027.
Initial capacity is planned at around 300 MW, with the infrastructure potentially expanding to 1 GW over time.
This is why Tokayev’s meeting with NVIDIA matters beyond a single corporate partnership. Kazakhstan already has two major supercomputing clusters and is now seeking to move from standalone computing systems toward a much larger AI infrastructure platform with regional ambitions.
DKNews.kz previously examined why Kazakhstan is building a Data Center Valley in Ekibastuz, as well as how businesses, universities and researchers can access the Alem.Cloud supercomputer.
