Kazakhstan AI Infrastructure Plan Explained: How Firebird, NVIDIA and Data Center Valley Fit Together for Everyday Tech Readers
A practical way to read this story is as a staged build rather than a single finished machine. Firebird and NVIDIA are outlining plans for Kazakhstan AI infrastructure that involve tens of thousands of GPUs, new data centre capacity and the Vera Rubin platform. The first large public number is a US export licence for 45,000 NVIDIA Grace Blackwell GPUs, with commercial operations scheduled for the third quarter of 2027.
Grace Blackwell is central to the first large GPU order, while the wider programme also includes existing systems, construction work and later additions. The story became more visible during the Kazakhstan visit of NVIDIA founder and CEO Jensen Huang on 23 October 2026.
The Astana Visit and the Political Context
During the visit, Jensen Huang met President Kassym-Jomart Tokayev in Astana. The discussions included Deputy Prime Minister Zhaslan Madiyev, the country’s minister of artificial intelligence and digital development.
For readers, the meeting signal is that the project sits at the intersection of national technology policy, power infrastructure and commercial AI computing.
Announced Steps So Far
Firebird has started constructing the first phase of its installation at Data Center Valley in Ekibastuz, with Kazakhtelecom as the local partner.
The NVIDIA announcement points to a second phase that may lift total AI infrastructure capacity in Kazakhstan to approximately 500 MW.
A longer-term expansion target for Data Center Valley is 1 GW.
Earlier agreements valued at $10 billion were announced by Kazakhstan’s government with Firebird and NVIDIA, including plans for a computing cluster containing up to 100,000 GPUs.
Construction updates published in August confirmed work on an initial 50 MW facility and the acquisition of a 215 MW electrical substation. The first facility is expected to enter operation in June 2027.
Existing GPU Systems and Future Routes
Kazakhtelecom’s current footprint includes 400 NVIDIA H200 GPUs and 480 NVIDIA B300 GPUs in Astana and Almaty. Those H200 and B300 systems are already in place. The operator has a 2027 plan for more than 1,000 additional GPUs at Data Center Valley and deployments exceeding 1,000 Vera Rubin GPUs in Ekibastuz.
The Vera Rubin architecture combines next-generation GPUs with the Vera CPU and high-bandwidth memory in rack-scale computing systems. As eeNews Europe previously reported, NVIDIA began supplying customer samples of Vera Rubin hardware earlier this year, ahead of broader deployments.
At-a-Glance Phase View
| Stage | What the plan includes | Reader takeaway |
|---|---|---|
| First Firebird phase | US export licence for 45,000 NVIDIA Grace Blackwell GPUs; construction at Data Center Valley in Ekibastuz with Kazakhtelecom; commercial operations in the third quarter of 2027 | Earliest operational checkpoint for the project |
| Second phase potential | Total AI infrastructure capacity in Kazakhstan may reach approximately 500 MW | Capacity growth is described as staged |
| Broader Data Center Valley target | Longer-term expansion target of 1 GW | The gigawatt figure is a future target, not current installed capacity |
| Already deployed operator systems | 400 NVIDIA H200 GPUs and 480 NVIDIA B300 GPUs across Astana and Almaty | Some GPU infrastructure exists outside the new Firebird phase |
Beyond Hardware: Startups, Video AI and Applications
The plan extends beyond building data centre halls. Firebird plans to establish Firebird Labs Kazakhstan, which would give local startups and researchers access to computing resources, funding and research partnerships.
NVIDIA also announced an investment in Higgsfield, a Kazakhstan-founded company developing generative AI tools for video and image production. The investment amount was not disclosed. Higgsfield intends to expand its computing capacity using Vera Rubin infrastructure supplied through cloud service providers.
During the visit, Jensen Huang discussed the possibility of using Kazakhstan’s energy resources to support AI computing services for international customers. Zhaslan Madiyev also outlined government interest in developing Kazakh-language models and applying AI to manufacturing, mining, agriculture and logistics.
Energy availability remains central to the economics. Kazakhstan’s government has identified competitive electricity costs and access to existing power infrastructure as advantages for the Ekibastuz development, although the proposed expansion will require substantial additional generating and distribution capacity.
Common Misconceptions to Avoid
- A licence quantity is not the same as installed capacity.
- An agreement value is not the same as a single construction bill.
- A gigawatt target is not the same as the amount of power now being used.
- GPU systems already deployed do not mean every announced GPU plan is complete.
FAQ for General Tech Readers
Is the AI infrastructure already fully operational?
Not in the full planned sense. The programme includes already deployed GPU systems, facilities currently under construction, future hardware deployments and a much larger proposed expansion.
Why is Ekibastuz mentioned so often?
Because it is the named location for Data Center Valley and the Firebird construction phase. Energy and grid access are central to that site, with a 215 MW electrical substation acquisition and an initial 50 MW facility reported in August.
What is the difference between the first facility and commercial launch?
The first facility is expected to enter operation in June 2027, while Firebird commercial operations are scheduled for the third quarter of 2027.
Does the plan only concern hardware?
No. It also connects to local startups and researchers through Firebird Labs Kazakhstan, to generative AI tools through Higgsfield, and to possible national AI application areas such as manufacturing, mining, agriculture and logistics.
How Readers Can Follow the Story Without Getting Confused
- Separate dates: the visit on 23 October 2026, construction updates in August, first facility operation in June 2027, and Firebird commercial operations in the third quarter of 2027.
- Separate units: GPU counts such as 45,000 NVIDIA Grace Blackwell GPUs, power figures such as 50 MW, 215 MW, approximately 500 MW and 1 GW, and money figures such as $10 billion.
- Separate actors: NVIDIA provides platforms and investment context, Firebird is a US-based AI infrastructure company behind planned installations, and Kazakhtelecom is the local telecom partner and operator.
- Look for conversion points: a licence, construction, power acquisition and hardware sample availability are stages; actual computing capacity comes when systems become operational.
Final Takeaway
The practical summary is that Kazakhstan AI infrastructure is moving through several layers at once: an approved GPU pathway, physical construction, existing operator systems, next-generation hardware and broader AI application ambitions. For everyday tech readers, the most reliable way to judge progress is to watch whether the first Firebird phase can turn those staged commitments into available computing capacity from the third quarter of 2027.