A $2 Billion Computing Project Takes Shape
Freedom Holding Corp. plans to develop a $2 billion artificial intelligence hub in Kazakhstan with 100 megawatts of available power, while opening its first public GPU cloud to businesses and developers in 2027. Chief executive Timur Turlov says demand already stands at twice the computing capacity planned for the hub's first stage.
Contents
- A $2 Billion Computing Project Takes Shape
- What Freedom and the Ministry Will Do
- Why a Public GPU Cloud Matters
- Demand Is Twice the First Stage's Capacity
- Three Developments With Different Milestones
- NVIDIA's Role Across the Announcements
- An Existing Cloud Business Near Almaty
- Turning Local Skills Into Working Services
- Key Points
The company set out its latest plans in an Astana announcement dated October 26, 2026. Its AI infrastructure announcement identifies Freedom as the principal financing and implementation partner, as well as the future operator, under an agreement signed in November 2025 with Kazakhstan's Ministry of Artificial Intelligence and Digital Development.
The proposed Sovereign AI Hub would support AI model training, large volumes of data processing, and services for companies and government agencies. Freedom is also developing an AI Super Cluster, whose first phase is scheduled for the first quarter of 2027, and building a new Freedom Cloud data centre near Almaty.
For potential customers, the practical change would be access to powerful shared computers without having to build and maintain their own server facilities. For Kazakhstan, the project combines domestic computing infrastructure with a national effort to develop AI skills and services.
What Freedom and the Ministry Will Do
Freedom will take responsibility for financing, delivering and operating the hub. The ministry's role, described when the agreement was announced on November 7, 2025, includes supporting conditions for hosting and operating large AI systems and leading related programmes to train workers.
The division of responsibilities connects a commercial operator with government support for infrastructure and skills. The project is intended to serve businesses, public institutions and developers, rather than a single government department or one part of Freedom's business.
Zhaslan Madiyev, Kazakhstan's deputy prime minister and minister of artificial intelligence and digital development, described the NVIDIA collaboration in the November 2025 announcement as an important step in implementing the national AI strategy. He said:
We are building the foundation of a sovereign AI ecosystem that will strengthen our economy
The term sovereign AI refers here to developing domestic computing capacity and the ability to create and operate AI services within Kazakhstan. Freedom's plans bring together the physical machines, cloud access and technical workforce needed to turn that ambition into services people can use.
Why a Public GPU Cloud Matters
Freedom plans to make its first GPU cluster available through a public cloud in 2027. A GPU, or graphics processing unit, is a chip that can perform many calculations at once. Groups of these chips are widely used to train AI models and run demanding AI applications.
A cluster connects many processors so they can work together. Cloud access lets customers use that computing power remotely instead of purchasing all the equipment themselves. Public cloud means a service offered to customers; it does not mean that access will be free.
Businesses and developers could use the planned service to build applications, train models on data or run existing AI tools. Model training is the process through which an AI system learns patterns from examples, often requiring substantial computing power.
Freedom says its investment is intended to help businesses turn local expertise into AI applications and digital products. The cloud model would give customers a way to obtain the underlying computing resources while concentrating their own work on software and services.
Demand Is Twice the First Stage's Capacity
Turlov says demand has reached twice the capacity planned for the hub's first stage. Put simply, for every unit of computing capacity planned for that initial stage, potential customers are seeking two.
Major prospective clients have also expressed willingness to lease future capacity for five years, according to Freedom. Commitments of that length could help the company plan how much equipment to install and how to organise access as the facility develops.
The figures describe customer interest at the planning stage. Freedom has not disclosed the size of the initial stage, a list of signed customer contracts, cloud prices, the detailed financing schedule or an opening date for the entire hub. Its stated 2027 milestones concern the public GPU cloud and the first phase of the Super Cluster.
The planned site's 100 megawatts describe available electrical power. Computing capacity depends on the chips, servers, networking and other equipment installed there. Electricity also supports the systems that keep a data centre operating, including cooling.
Three Developments With Different Milestones
Freedom's programme includes the Sovereign AI Hub, the public GPU cloud and Freedom AI Super Cluster. A new Freedom Cloud data centre near Almaty adds another physical development to the company's computing plans.
The Super Cluster is intended to serve government, business and science. Its first phase has the most specific launch window in the announcement: the first quarter of 2027, meaning January through March. The public GPU cloud has a broader target of sometime in 2027.
The main dates show how the programme has developed:
- November 7, 2025: Freedom announced its agreement with Kazakhstan's Ministry of Artificial Intelligence and Digital Development in a release issued from Las Vegas.
- October 23, 2026: Turlov met NVIDIA founder and chief executive Jensen Huang in Astana to present Freedom's technology businesses and AI infrastructure plans.
- October 26, 2026: Freedom issued its latest update on the $2 billion hub and planned public GPU cloud.
- First quarter of 2027: Freedom targets the first phase of its AI Super Cluster.
- 2027: Freedom plans to open its first GPU cluster to customers through a public cloud.
These milestones describe a programme being delivered in stages. Businesses preparing to use the services will need to plan around the particular computing resource they require and its scheduled availability.
NVIDIA's Role Across the Announcements
The November 2025 announcement described the hub as operating on sovereign exascale NVIDIA AI infrastructure. Exascale refers to an extremely high level of computing performance, with the precise performance depending on the type of calculation being measured.
On October 23, 2026, Turlov met Huang in Astana during what Freedom described as the NVIDIA chief's first visit to Kazakhstan. Turlov presented the group's technology operations and plans for AI infrastructure.
The October update did not name a chip supplier for the hub, while the November 2025 announcement explicitly identified NVIDIA infrastructure. The earlier description provides the stated technology direction; the later statement focuses on customer access, demand and delivery milestones.
Freedom's plans place Kazakhstan within a wider push by governments to develop domestic AI capacity. The original project account also points to a sovereign AI stack offered by NVIDIA and Palantir, Malaysia's consideration of Huawei chips, and European research warning that data centre capacity could limit national ambitions. Kazakhstan's proposal puts a specific power allocation and a commercial operator at the centre of its approach.
An Existing Cloud Business Near Almaty
Freedom Cloud already serves more than 3,400 business and public sector clients, according to the company. Construction of its new data centre near Almaty would add infrastructure to an operation that already has customers.
Those existing relationships give Freedom a channel through which to offer additional computing services. The announced GPU cloud would add resources designed for AI workloads to the group's wider cloud activity.
Freedom Holding, which trades on Nasdaq under the symbol FRHC, operates across banking, brokerage, insurance and telecommunications. Its technology businesses also include ticketing, travel, e-commerce and food delivery.
That range explains why Freedom is presenting the investment as part of a broader technology business. Financial services, commerce and consumer platforms all involve software and data, while the proposed cloud would make computing resources available to customers outside the group.
Turning Local Skills Into Working Services
Turlov says Kazakhstan has developed a strong pool of talent already creating technology in e-commerce and finance. He argues that combining those skills with domestic AI infrastructure is becoming one of the country's main advantages in digital development.
Freedom describes its role as supplying the computing capacity and cloud services that allow businesses and developers to turn that expertise into applications. The ministry's planned training programmes address another part of the same task: preparing people to work with large AI systems.
The plans connect three practical needs. Developers need access to machines powerful enough for their work. Businesses and public institutions need services they can use. The operator needs customers willing to pay for capacity over a period long enough to support a substantial infrastructure investment.
The next scheduled delivery point is the Super Cluster's first phase in the first quarter of 2027, followed by the planned public GPU cloud launch during that year. Those services would give customers their first opportunity to use the new computing resources described in Freedom's programme.
Key Points
- Freedom plans a $2 billion Sovereign AI Hub in Kazakhstan at a site with 100 megawatts of available power.
- The company will finance, deliver and operate the hub under an agreement signed with the digital development ministry in November 2025.
- Turlov says demand is twice the planned first stage's capacity, with prospective clients interested in leases lasting five years.
- The first phase of Freedom AI Super Cluster is targeted for the first quarter of 2027, while the public GPU cloud is planned for 2027.
- A new Freedom Cloud data centre is under construction near Almaty. The cloud business already serves more than 3,400 clients.
- The November 2025 announcement identified NVIDIA infrastructure; the October 2026 update did not name a chip supplier.






