At AI & Digital Bridge in Kazakhstan’s capital Astana on 1 October, the county’s President, Kassym-Jomart Tokayev, proposed an AI Challenge built around problems identified by industry. Kazakh and international teams would compete to solve them, with effective projects taken into production. The proposal captures the central question facing Kazakhstan’s technology strategy: how to turn digital capabilities into measurable economic gains.
That question matters for an economy seeking to broaden its sources of growth beyond natural resources. Technology offers two routes towards that goal. Software businesses can generate exports, while AI can improve the performance of existing industries. For Kazakhstan, opportunities to reduce equipment downtime, improve freight handling or simplify administration give the digital agenda an immediate economic purpose.
Kazakhstan starts with a useful foundation. It ranked 24th among 193 countries in the UN’s 2024 E-Government Development Index and tenth for online services. Its experience in moving public administration online provides a base for deploying AI in services people already use. The next step is to make those services easier to navigate and more responsive to individual needs.
There are signs of that transition. In August, the government reported that its eGov GPT assistant offered access to 50 public services and had been used by more than 150,000 citizens. It also reported that digitalisation of frequently used services had reduced visits to public service centres by 30%. Such results suggest why AI adoption can build on previous investment: the administrative systems and user demand are already there.
Tokayev’s designation of 2026 as the Year of Digitalisation and Artificial Intelligence has been followed by a broader policy framework. The Digital Qazaqstan strategy, approved in June, runs through 2029. Its August implementation plan sets targets that connect technology spending to outcomes, including cutting the time needed to deliver public services and social support by at least half and connecting at least 70% of fuel and energy assets to digital monitoring by 2029.
These remain targets, but they establish a basis for assessing progress. The same logic should guide the proposed industrial AI Challenge. A maintenance system should demonstrate that it reduces breakdowns; a logistics application should show that it speeds up cargo movement. Competitions become economically useful when successful prototypes can secure customers and be deployed beyond a demonstration.
Computing infrastructure is another part of the strategy. At the forum, Tokayev’s meeting with NVIDIA vice-president Rev Lebaredian highlighted two already launched supercomputing clusters using the company’s technology. They also discussed Data Center Valley in Ekibastuz, a planned expansion of computing infrastructure. The distinction matters: existing capacity can support work today, while larger developments require sustained investment and reliable electricity supplies.
The value of this infrastructure depends on what researchers and businesses build with it. KazLLM, developed at Nazarbayev University’s Institute of Smart Systems and Artificial Intelligence, illustrates the importance of local expertise. Based on the Llama 3.1 architecture, it was designed to strengthen capabilities in the Kazakh language and local context. Adapting established technology can help address needs that receive less attention in global AI development.
Language capability has practical consequences. Public services become more accessible when citizens can interact in the language they use daily. Educational applications also need to explain concepts accurately in that language. A model alone does not deliver those benefits: it still needs integration with reliable information, testing and continued development. Kazakhstan’s opportunity lies in building that expertise alongside its infrastructure.
The recently opened Qazaq AI Research University adds an institutional component. Its programmes include AI and machine learning, physical AI and AI+X, which connects artificial intelligence with other disciplines. Access to national computing infrastructure could help students move from coursework to applied research. Meanwhile, the national action plan aims for at least 80% of school, college and university graduates to possess basic AI skills by 2029.
Commercial activity suggests the ecosystem is gaining substance. The Kazakhstan AI Country Report, prepared by RISE Research with partners including Mastercard and Freedom Bank, identified more than 100 AI startups. It estimated that venture investment in AI projects rose from approximately $14 million in 2023 to $73 million in 2025. Many companies focus on business applications, where customers can assess a product against a specific operational need.
A visible example is Higgsfield. The AI video company is headquartered in San Francisco and has a large engineering hub in Almaty. In August, it announced a $400 million funding round at a $5.4 billion valuation. Its growth shows how engineers working in Kazakhstan can contribute to products sold internationally, linking domestic expertise to global capital and customers.
There is also evidence of international demand. Tokayev told Digital Bridge that Kazakhstan’s IT services exports exceeded $1 billion in 2025. Export growth gives domestic developers opportunities beyond their home market. Companies that improve their products through demanding international customers can, in turn, bring that experience into local projects.
Scaling will require further work. The AI Country Report identifies shortages of high-quality industry data and experienced specialists, uneven regional access to infrastructure and limited late-stage venture funding. These constraints help explain the emphasis on education and industry partnerships. Buyers also need confidence that new products will be maintained and supported after deployment.
Regulation will shape that confidence. Kazakhstan has adopted an AI law and a Digital Code, but Tokayev warned at the forum: “We must not overregulate this sphere to the point where promising teams, technologies and capital begin moving to other jurisdictions.” Achieving that balance means protecting personal data and ensuring accountability while allowing developers to test useful applications. People need ways to challenge consequential errors in automated services.
Kazakhstan’s approach deserves attention because it connects AI ambition with an existing base of digital services and concrete economic needs. Its next advances will be judged through implementation: whether services save citizens time and whether businesses gain productivity. Delivering those benefits would give the country a stronger technology sector and offer other emerging economies useful evidence about how to make AI investment pay.