Uzbekistan's AI Bet Is a Test Case for Korean Edtech Abroad
Uzbekistan wants to train 5 million people in AI by 2030. Korean edtech firms are scrambling to figure out whether to sell finished products or build the tools locals use to make their own.
The numbers are eye-opening
Uzbekistan plans to train five million people in AI by 2030. That is not a vague ambition — it is a hard target set by the government, broken into 4.75 million students and young people, 150,000 teachers, and 100,000 civil servants. The curriculum change starts in the 2026–2027 school year, when “informatics and information technology” will formally include AI basics. Coursera has already translated roughly 3,000 courses into Uzbek. The machinery is being assembled.
What gets less attention is how quickly Uzbek teachers have already started using AI, and how much of a gap remains between enthusiasm and infrastructure.
Teachers are ahead of the classroom
According to the OECD’s TALIS 2024 survey, 62 percent of Uzbek secondary-school teachers reported using AI in their work over the previous 12 months. The OECD average sits at 36 percent. That is a dramatic outlier, and it tells you something important: the demand side of this equation is not hypothetical. Educators in Tashkent and the surrounding regions are already experimenting with AI tools, and they are doing so at rates that exceed many wealthy democracies.
But the flip side matters more for foreign companies trying to sell into this market. Among teachers who did not use AI, 54 percent cited insufficient school infrastructure as the reason. The OECD average for that complaint is 37 percent. Uzbekistan’s teachers want to use these tools. The network conditions and hardware simply do not keep up.
This gap — between usage intent and usable infrastructure — is where Korean edtech firms will find their opening, and also their biggest risk.
What Korean companies are being asked to build
The demand is broadening fast. It is no longer just about delivering online courses. Uzbekistan’s education authorities are looking for AI tutoring systems, automated grading, learning management platforms, digital assessment tools, and teacher-side AI solutions for lesson planning and feedback generation. The Korean companies that will succeed are the ones that can bundle these functions into integrated packages rather than shipping single-product solutions.
There is a more specific niche emerging as well. KOICA has been running edtech demonstration projects at the National World Language University in Tashkent, focusing on Korean-language education. The interest there covers internet-based testing, question banks, AI-powered automated scoring and feedback for the TOPIK II writing section, and LMS analytics that break down performance by category and item. If a Korean company can build a robust Korean-language grading and feedback tool today, the path to expanding into Uzbek and English is shorter than it looks — the core technology transfers.
The bandwidth problem nobody talks about
Here is what most Korean exporters will get wrong: they will assume their cloud-first, always-online product can be dropped into the Uzbek market and work. It cannot. A significant share of schools operate on slow or unreliable networks. Local purchasers are explicitly asking for features like offline-capable answer storage, automatic data sync when connectivity returns, and stable audio and video delivery even on low bandwidth. Products designed for Seoul’s high-speed infrastructure will flounder unless they are rebuilt with those constraints in mind.
This is not a minor localization adjustment. It is a fundamental rethinking of architecture. Companies that treat it as such will gain a durable advantage. Those that ship a Korean product with a language pack will lose credibility fast.
The market is already moving
The private sector is helping pull the demand along. The number of registered private IT education centers in Uzbekistan rose from 390 to 428 between 2024 and last year. Graduates from IT training programs jumped from 24,800 to 37,800 over the same period. Higher-education institutions running IT curricula grew from 102 to 145, with student numbers climbing from roughly 78,000 to 128,000. The ecosystem is expanding in real time, not waiting for government mandates to catch up.
Korean institutions on the ground provide a foothold. Inha University Tashkent, Bucheon University Tashkent, and Ajou University Tashkent are already operating. KOICA is funding a $6.9 million project through 2029 to establish a cyber university at the University of World Economy and Diplomacy. These are not symbolic的存在 — they are distribution channels, validation partners, and recruitment pipelines rolled into one.
The realistic entry strategy
Industry observers at KOTRA’s Tashkent trade office note that individual schools cannot realistically develop their own AI education programs from scratch. That creates a clear B2B and B2G path: partner with the Korean-affiliated universities or KOICA projects, run pilots, demonstrate outcomes, then scale through public education contracts. It is a slower play than a direct software sale, but it avoids the trap of trying to sell finished products into an environment that cannot support them.
An alternative model is gaining traction among Korean firms — rather than exporting completed curricula, provide the authoring tools and AI capabilities that let local educators build their own content. This shifts the value proposition from vendor to platform, and it aligns better with the infrastructure realities on the ground. Local teachers become the creators; Korean companies provide the engine.
What this means beyond Uzbekistan
Uzbekistan is the largest country in Central Asia and a gateway to a region of roughly 70 million people. If Korean edtech firms can prove their products work under constrained-network conditions here, the playbook is directly transferable to Kazakhstan, Kyrgyzstan, and other markets where digital infrastructure lags behind ambition. The 5-million-person target is large enough to matter financially but small enough to serve as a controlled test case. That combination does not come up often.
The lesson for Korean companies is not that Central Asia is a gold rush. It is that the region’s gap between digital aspiration and physical reality is precisely where the opportunity lives — provided you build for the gap, not against it.