Kazakhstan to Establish International Computing Hub in Pavlodar Region

Kazakhstan plans to develop an international computing hub centered in the Pavlodar region, based on the emerging “data center valley” in Ekibastuz, one of the country's key energy production zones. The announcement was made by Deputy Prime Minister and Minister of Artificial Intelligence and Digital Development Zhaslan Madiev during an expanded government meeting. The government had earlier unveiled plans to establish a cluster of high-performance data centers in Ekibastuz, a city that hosts extensive coal-fired power generation and robust energy infrastructure. Madiev now says the project will evolve beyond a national cluster into a full-fledged international computing hub. According to Madiev, two critical enablers for the project are affordable electricity and cross-border telecommunications infrastructure. A land plot has already been designated for development, and 300 megawatts of power capacity have been reserved at Ekibastuz GRES-1. Negotiations are ongoing with international investors. AI Center to Anchor Digital Expansion The hub’s anchor facility will be a 50 MW AI-focused data processing center currently under construction by national telecom provider Kazakhtelecom. Scheduled for commissioning in mid-2027, the new facility is expected to be ten times more powerful than Kazakhstan’s recently launched state supercomputer facility. Astana is already home to Central Asia’s most powerful supercomputer, unveiled earlier this year, marking a significant step in the country’s digital infrastructure ambitions. Officials envision this computing expansion as the foundation for a new economic model driven by artificial intelligence and big data. Concerns Over Domestic AI Model Usage At the same government session, President Kassym-Jomart Tokayev expressed dissatisfaction with the uptake of the national large language model, KazLLM. He noted that public interest remains limited, with many users favoring international platforms such as ChatGPT. Madiev acknowledged that while KazLLM is used by approximately 600,000 people, about 3% of the population, foreign AI tools enjoy far broader adoption. He emphasized the importance of ongoing training and updates to local AI models, aligning with global best practices. In addition to KazLLM, a second model, Alem LLM, is also under development. Both systems have already been integrated into numerous digital platforms, including search engines and services targeted at students and IT enterprises.