What Uzbek startup founders are building

Only funded startups are announced in Uzbekistan, so the process of building a startup, including its attempts, concentrations, and failures, remains unknown to the public. This article contributes to filling that information gap using the 3,573 applications submitted to UzCombinator’s (UzC) second batch in June–July 2026, of which 1,360 complete applications form the analytical sample. It describes the applicant pool by sector, stage, customer type, and revenue model, and compares it with the accepted cohort (n=31) and with Y Combinator’s Summer 2025 batch. AI and education are the most common sectors among applicants, and AI applied to education is the largest AI combination. AI appears mainly as a function added to existing sectors to solve domestic problems. Selection differs sharply from the pool: no idea-stage startup was accepted, although idea-stage founders made up 26.2% of applicants, while revenue- and growth-stage startups were roughly three times more common among those accepted. The accepted cohort is more AI-based (51.6%) and mostly B2B (87.1%), moving it closer to the Y Combinator profile. Subscription is the most common revenue model across the pool. The study is descriptive and is not a census of Uzbek startups. It covers founders who chose to apply to one accelerator in one window, so founders who never applied or never heard of the program are absent by construction, and therefore, the findings do not establish causes. Keywords: startup ecosystem, Uzbekistan, accelerators, entrepreneurship, artificial intelligence, emerging markets, Central Asia