Kazakhstan has declared 2026 the "Year of Digitalisation and Artificial Intelligence" and is backing the slogan with institutional scaffolding: a dedicated Ministry of Artificial Intelligence and Digital Development created in 2025, an AI governance law (making it the second jurisdiction after the EU to pass one), a national action plan through 2029, and a sovereign wealth fund that seated an AI programme with voting rights on its board. The pace is real, even if the substance remains uneven. The talent pipeline centres on Astana Hub, Central Asia's largest innovation cluster, which runs free programming and AI courses for schoolchildren and a selective adult retraining programme called Tomorrow School. Of 25,000 adult applicants, fewer than 2 percent were admitted. Eighteen months in, 500 people have been trained, with pools opening in 12 regions. The hub has also incubated Kazakhstan's first unicorn, Higgsfield, a video-generation company, and aims for five unicorns by 2030. On the research side, Nazarbayev University's Institute of Smart Systems and Artificial Intelligence built Central Asia's first academic supercomputer and launched MagniSoz, a locally hosted multilingual model, in December 2024. An updated version called Oylan handles images in English, Russian, and Kazakh. The rationale is part data sovereignty — keeping citizen data on domestic servers — and part cultural: training models on Kazakh history, traditions, and reasoning patterns rather than importing Western defaults. The infrastructure underneath this ambition is fragile. Kazakhstan still imports electricity at peak times and runs on aging plants and grids. Large data centres demand stable power and cooling, yet the ministry has offered no public environmental assessment. It claims immersion cooling will replace water-based systems, but as Tlegen Kuandykov of the CAPS Unlock think tank notes, the transparency gap itself is a policy failure. Human rights concerns compound the infrastructure risk. The 2025 AI law includes protective language, but enforcement mechanisms and public oversight structures remain unclear. Biometric identification and facial recognition are rolling out rapidly — a familiar pattern in post-Soviet states where surveillance infrastructure often outpaces the legal frameworks meant to constrain it. Yuri Misnik, CTO of InDrive, frames the opportunity geographically: proximity to China gives Kazakhstan access to eastern hardware and manufacturing ecosystems while western innovation models provide the software and governance templates. The positioning is clever, but it also means Kazakhstan absorbs the tensions between those two AI governance philosophies without having resolved them domestically. The programme is early-stage, genuinely generative in intent, and structurally incomplete. A barista-to-developer pipeline is a real social elevator; an AI board member on a sovereign wealth fund is branding. The difference between the two will determine whether Kazakhstan becomes a regional AI centre or a showcase for digital Potemkin villages.
