Central Asia AI Talent Market Scan 2026

Demand side vs. supply side for AI and data talent in Kazakhstan, Kyrgyzstan, Uzbekistan, Tajikistan, and Turkmenistan. All figures traceable to the primary sources listed in the Sources section (URL + access date). Derived estimates are explicitly labeled as such and never presented as observed data.

Executive summary

Central Asia's five republics are pursuing some of the most ambitious digital-economy targets of any emerging region — Kazakhstan aims for $5 billion in IT services exports by 2030, Uzbekistan targets 300,000 IT jobs, and both countries have announced programs to train millions of people in AI-related skills. The demand signal is real and measurable: Kazakhstan alone has a live stock of several hundred open data-engineering postings, ministerial-level estimates put near-term IT labor demand at 12,000 workers, and the country's flagship tech hub plans to train 10,000 AI specialists by 2030 — a target that implicitly concedes the current stock of AI specialists is far smaller.

The supply side is thinner and harder to measure. UNESCO data show that ICT programs account for between 3.6% (Uzbekistan) and 8.1% (Kazakhstan) of annual tertiary graduates — shares that look respectable until set against the ambitions above and against the fact that "ICT graduate" is a far broader category than "AI/data-science graduate." Verified AI-specific degree output is not published by any of the five countries' statistical agencies. National programs (Kazakhstan's AI-Sana, Uzbekistan's "One Million Uzbek Coders") are producing impressive certificate-level volumes, but these are short courses, not degree-level AI/ML training.

Bottom line: this scan finds a directional but well-evidenced talent gap. It is a gap of depth, not just headcount: employers report acute shortages at Middle and Senior levels, while the education pipeline is optimized for volume at the junior and certificate level. English-taught BSc programs in AI and Data Science in the region would be entering a market where employer demand is documented, state policy is actively subsidizing supply, and the specific product — degree-level, English-medium AI/ML competence — is nearly absent from local provision.

Methodology

Demand side

Primary attempt — hh.ru public API (HeadHunter)

The scan attempted to query the public vacancies endpoint https://api.hh.ru/vacancies (documented at https://api.hh.ru/openapi) with text=<keyword>, area=<country>, per_page=0 for keywords machine learning, data scientist, data engineer, AI engineer, машинное обучение, дата-сайентист, ML инженер across country areas Kazakhstan (40), Uzbekistan (97), Kyrgyzstan (48), Tajikistan (86), Turkmenistan (93) — area IDs verified via the API's /areas endpoint. All vacancy queries were rejected with HTTP 403 from the provider's DDoS-protection layer on 17 October 2026; reference endpoints (/areas, /professional_roles) from the same host remained reachable, so the block is specific to the vacancy search path from this network. LinkedIn job counts were not used (login required). Direct hh.kz page fetches returned vacancy listings but the text-fetch pipeline did not expose the result-count field.

Working demand sources (all accessed 16–17 October 2026):

1. GorodRabot.kz live vacancy count — the aggregator kazahstan.gorodrabot.kz (which republishes hh.kz and other boards) renders its live result count in page text. Query data engineer, country-level Kazakhstan page: 728 open vacancies, with a computed average posted salary of 953,000 ₸/month (min 150,000 / max 1,800,000). Page: https://kazahstan.gorodrabot.kz/data_engineer, accessed 17 October 2026.
2. Official government statements — Kazakhstan Ministry of Labor and Social Protection vacancy/resume data via Enbek.kz, reported by The Times of Central Asia (2026-05-12); Kazakhstan Minister of Education statement on sectoral skilled-worker demand (2026-09-08, via China News Service).
3. Industry targets and program announcements — Astana Hub / Ministry of AI and Digital Development roundtable (reported by The Times of Central Asia, 2026-04-20); IT Park Uzbekistan figures via The Times of Central Asia; Uzbekistan "Outsource to Uzbekistan" official talent portal (outsource.gov.uz); Kyrgyz High Technology Park (HTP) figures via HTP directorate interview (The Times of Central Asia) and the HTP GITEX 2025 press kit.
4. Labor-market commentary — HeadHunter Uzbekistan study cited in a peer-reviewed article (Latin American Journal of Education, February 2026); World Bank "Promise of AI" findings reported September 2026; industry analysis of the Uzbek IT market (vc.ru, September 2026).

Supply side

1. UNESCO Institute for Statistics (UIS) Data API (https://api.uis.unesco.org/api/public/data/indicators, accessed 16 October 2026) — indicator FOSGP.5T8.F600 (percentage of tertiary graduates from Information and Communication Technologies programmes, both sexes) and FOSGP.5T8.F500600700 (STEM share), for KAZ/KGZ/UZB/TJK/TKM, years 2015–2025. Latest available year per country used.
2. National statistical releases — Kazakhstan Bureau of National Statistics higher-education bulletin (2023/2024 graduates); Uzbekistan National Statistics Committee higher-education press release (2024 graduates, published 2026-05-28); Kyrgyz National Statistics Committee education releases (2026-10-02).
3. Official program data — Kazakhstan AI-Sana program and Tomorrow School figures (reported October 2026); Uzbekistan official IT talent portal figures; Kazakhstan state grant allocations (2026–2027 academic year).

Gap sizing rule

A sized comparison is made only for Kazakhstan, where both an official demand figure (12,000 IT workers needed) and a derivable ICT-graduate estimate exist. The ICT-graduate estimate multiplies the UIS ICT share by the national total of higher-education graduates; it is labeled derived estimate throughout. Everything else is directional and labeled accordingly.

Data tables

Table 1 — Demand-side indicators (observed)

Table 2 — Supply-side indicators (observed)

Table 3 — The gap: Kazakhstan (sized — derived estimate, labeled)

Key findings

1. Demand is documented at the state level, not just on job boards.

Kazakhstan's education minister quantified near-term IT labor demand at 12,000 skilled workers (September 2026), the labor ministry reports 500+ open software-developer vacancies on the state Enbek.kz portal (May 2026), and a live aggregator count showed 728 open data-engineer postings in October 2026. These are three independent demand observations pointing the same direction.

2. The region's digital ambitions imply demand growth far beyond current hiring.

Kazakhstan's IT services exports more than doubled in late 2025 to $1.142B annualized, with a $5B-by-2030 target; Uzbekistan's IT exports approached $1B in 2025 with a 300,000-job target. Export-led IT growth is labor-intensive in engineering talent, and both governments are explicitly framing workforce development as the binding constraint (Kazakhstan's "AI Country Report" itself cites shortages of experienced specialists).

3. The supply pipeline is built for certificate-level volume, not degree-level AI depth.

Kazakhstan's AI-Sana issued 706,000+ AI course certificates and Uzbekistan's "One Million Uzbek Coders" counts 1.7M graduates — genuine achievements in digital literacy, but these are short courses. Meanwhile, no Central Asian statistical agency publishes AI/ML-specific degree output, and the closest international proxy (UNESCO ICT graduate shares: 8.1% KZ, 4.6% KG, 3.6% UZ) bundles everything from telecom technicians to software engineers. Degree-level AI/data-science supply is effectively unmeasured — which, for planning purposes, means it cannot be shown to be adequate.

4. The sharpest reported shortage is at Middle/Senior level — a quality gap, not a headcount gap.

Industry analysis of the Uzbek market describes junior supply as abundant while "the market experiences a severe talent hunger for Middle+ and Senior positions" (vc.ru, September 2026). Kazakhstan's salary data corroborates the premium: data-engineer postings average 953,000 ₸/month. This pattern — plentiful juniors, missing seniors — is exactly what a young degree pipeline produces, and it is the gap an English-taught BSc in AI/Data Science is positioned to address over a 3–4 year horizon.

5. English-medium delivery is a named bottleneck in the region's internationalization strategy.

Analysis of Kazakhstan's higher-education strategy notes that "accessible, high-quality English instruction must expand significantly" for the country to compete internationally across disciplines (Study in Kazakhstan / PIE reporting, 2026). With 64% of Uzbek workers already using AI on the job and AI tooling overwhelmingly English-first, English-medium AI programs have a structural employability advantage.

6. Tajikistan and Turkmenistan cannot be sized — and that is itself a finding.

UNESCO's latest ICT-share data for Tajikistan is from 2017; Turkmenistan's 2024 figure (0.02%) is not credible as a real share and signals a reporting break. No verifiable vacancy or graduate statistics were found for either country. Any regional talent strategy must treat these two markets as data-dark: the gap there is unmeasurable, not necessarily small.

The gap

Kazakhstan (sized, labeled)

A derived estimate puts annual ICT graduate output at roughly 12,700 (157,106 total graduates × 8.06% ICT share). Against this, the state projects demand for 12,000 additional skilled IT workers in the coming years, there are 728 live data-engineer postings from a single keyword query, and the national plan calls for 10,000 AI specialists to be trained by 2030. On headline numbers, supply and demand look roughly balanced — but the balance is misleading in three ways, all labeled here as directional assessment: (a) the 12,700 is all ICT fields, of which AI/ML is a small fraction; (b) the 12,000-worker need is explicitly additional demand on top of replacement hiring; (c) the binding constraint reported by employers and the government's own AI report is experienced (Middle+) specialists, which a junior-heavy pipeline does not produce. Directional conclusion: Kazakhstan faces a structural AI/ML depth gap, not a general ICT headcount gap.

Uzbekistan (directional)

25,000+ annual IT university graduates (official portal) against a 300,000 IT-job target and ~$1B in IT exports growing fast. The HeadHunter study's finding — 64% of workers already using AI, with demand for ML engineers and data scientists "continuing to grow" — combined with reported Middle+/Senior shortages, indicates demand is outrunning the degree pipeline's ability to produce experienced talent.

Kyrgyzstan (directional)

A 500-company export IT sector ($130M in 2024) built on a small domestic graduate base; ICT share of graduates is 4.56%, below Kazakhstan's. The sector's "live in Kyrgyzstan, work with the world" model makes English-medium AI talent directly monetizable through remote and export work.

Tajikistan / Turkmenistan

No defensible sizing possible (see Limitations).

Implications for English-taught BSc AI / Data Science programs in Central Asia

1. The product-market fit is documented, not assumed. Employer demand (live postings, ministerial demand figures, export-growth targets) and state supply-side subsidies (grants weighted 60% to engineering/technical fields in KZ; IT Park ecosystems in UZ and KG) mean a new English-taught BSc in AI/Data Science would launch into a market where both buyers and policymakers are already pulling in the same direction.
2. Compete on depth, not volume. The region does not lack junior digital talent — it lacks Middle/Senior AI/ML practitioners. A BSc program should be designed and messaged around producing graduates who can grow into those roles: strong mathematics/statistics foundations, production ML engineering (MLOps, deployment), and real project portfolios — the exact profile employers say is missing.
3. English-medium is a differentiator with labor-market value, not just a marketing label. AI tooling, documentation, and research are English-first; 64% of Uzbek workers already use AI on the job. English-taught graduates can serve both domestic employers and the region's export-oriented IT sector (notably Kyrgyzstan's remote-work model), widening the addressable job market beyond national borders.
4. Co-design with the existing ecosystems. IT Park Uzbekistan (2,990+ companies), Astana Hub, and Kyrgyzstan's HTP (500+ companies) are natural partners for internships, capstone projects, and hiring pipelines. Their published growth targets (300,000 IT jobs in UZ; 10,000 AI specialists in KZ) give a program concrete placement narratives.
5. Plan for a 3–4 year credibility window. The gap this scan identifies is structural and slow-moving: state training programs target 2030, and degree pipelines take years to produce graduates. A BSc launched now would graduate its first cohort just as the current junior-heavy pipeline's limits become most acute — provided the program resists the temptation to compete on certificate-level speed and instead builds degree-level depth.

These findings are relevant context for any new English-taught AI/Data Science degree provision in the region.

Limitations

• hh.ru vacancy API blocked: direct keyword×country vacancy counts from hh.kz/hh.uz could not be obtained — the provider's DDoS-protection layer returned HTTP 403 for all /vacancies API queries from this network on 17 October 2026 (area IDs 40/97/48/86/93 verified; reference endpoints reachable). One live aggregator count (gorodrabot.kz, KZ, "data engineer") substitutes, but it is a single query on one aggregator and subject to its deduplication and coverage choices.
• No LinkedIn data: job counts requiring login were excluded by design.
• Supply side is patchy by construction: UNESCO publishes ICT graduate shares, not absolute numbers; absolute ICT graduate figures are derived estimates (labeled). No Central Asian statistical agency publishes AI/ML-specific degree output. Uzbekistan's official portal figures (25k/120k IT graduates) use broader definitions than UNESCO's and are not directly comparable.
• Turkmenistan data anomaly: the UIS 2024 ICT share of 0.02% is not credible as a true value and likely reflects a reporting break; it is excluded from all reasoning.
• Tajikistan staleness: latest UIS ICT-share data is from 2017; no current demand-side figures were verifiable.
• Certificate ≠ degree: program figures (AI-Sana, One Million Uzbek Coders) count course completions and must not be read as degree-level supply.
• Currency of press figures: ministerial and industry figures are cited with their publication dates; labor markets move quickly and these are point-in-time observations.

Sources

All accessed 16 or 17 October 2026 unless noted.