How Central Asia Accelerates AI Adoption in Finance
Financial institutions across Central Asia are moving faster on AI. A new study from the National Bank of Kazakhstan (Feb. 13) shows 36% are already using AI and 56% plan to implement within 12 months. Maturity is still early: 38% remain in research, 28% run pilots or partial rollouts, and just 2% operate at full scale.
Where adoption stands
The survey covers 232 organizations across Kazakhstan, the Kyrgyz Republic, and Tajikistan. The headline: interest is high, but most programs have not cleared the jump from pilots to production-grade systems embedded in core processes.
"In 2026, the National Bank will focus on transitioning to the practical implementation of a digital asset market and integrating it into Kazakhstan's existing financial system. Plans also include a unified AI ecosystem, a platform of specialized AI agents within the Bank, and regional collaboration to develop practically applicable AI solutions," said National Bank Governor Timur Suleimenov. The priority is clear: manage risk, ensure institutional readiness, and meet regulatory requirements as AI agents take on more complex work in supervision and analytics.
From operations to risk and decisions
Use cases are moving from simple automation to predictive analytics and risk-focused applications. Fraud detection, credit assessment, and customer data analysis now lead deployment pipelines.
In Kazakhstan, 75% of banks already use AI and 88% plan to expand. Most activity clusters in transactional and customer-facing areas-credit scoring and anti-fraud-while strategic planning, compliance oversight, and enterprise risk remain underdeveloped.
Generative AI and agents: practical upside
Generative AI and semi-autonomous agents are gaining traction. Globally, agent-based systems are projected to grow at about 45% annually over the next five years, and could support or participate in half of business decisions by 2027. Restructuring workflows around AI agents can compress task time by 60% to 90%, depending on the application, according to McKinsey.
McKinsey: the economic potential of generative AI
Global momentum, local gaps
Corporate AI investment reached $252.3 billion in 2024, up 26% year over year. An estimated 78% of organizations worldwide now use AI in at least one function (up from 55% in 2023), while inference costs have fallen roughly 280-fold since 2022-making entry via cloud and low-code far easier.
Central Asian markets are catching up but still outside the top 50 in AI readiness. In 2024, Kazakhstan ranked 76th, Uzbekistan 70th, Tajikistan 131st, and the Kyrgyz Republic 134th. The bottlenecks: technological maturity and-above all-human capital.
Capacity, governance, and risk
The talent deficit is specific: finance expertise blended with data analytics and risk management. Beyond compute, institutions need dependable data supply chains and governance that scale-consistent standards, clear lineage, and access controls that pass audit.
In Tajikistan, 65% of financial executives view AI as critical for competitiveness, yet only 33% of institutions use it today. Many initiatives remain pilots, not integrated into core business processes, which limits impact.
In the Kyrgyz Republic, supervisors prioritize AI for payment monitoring, compliance automation, and analytics. Still, unified risk approaches and secure cross-border data exchange are shared regional challenges.
Regulatory and security outlook
Regulation is moving in step with technology. The EU AI Act introduces obligations including labeling of AI-generated content, while only 38% of systems globally apply watermarking today. Over ten countries have created AI Safety Institutes to set deployment standards.
European Commission: AI Act overview
Security pressure is rising. A cited survey found 68% of organizations in the US and UK experienced data leakage tied to employee use of AI tools, while only 23% had comprehensive AI security policies in place. Environmental costs are part of the equation too: data center energy use is up 72% and annual water consumption reached 560 billion liters-though AI is also used to drive efficiency in agriculture, logistics, and energy management.
What finance leaders should do next
- Stand up AI governance now: model risk management, data usage policies, audit trails, content provenance, and human-in-the-loop controls.
- Fix the data foundation: common data model, quality SLAs, lineage, privacy-by-design, and cross-border agreements where needed.
- Prioritize high-ROI, risk-aware use cases: fraud analytics, credit decisioning, AML/KYC, collections, and treasury forecasting with clear KPIs.
- Pilot AI agents where oversight is measurable: alert triage, regulatory report drafting, reconciliations-keep humans in control.
- Close security gaps: DLP for prompts and outputs, role-based access, secrets management, red-teaming, and prompt-injection defenses.
- Plan for cost and capacity: right-size models, use efficient inference, track unit economics, and set energy/water reduction targets with vendors.
- Build hybrid talent: upskill finance teams in analytics and model risk; partner with universities and regional hubs.
- Collaborate regionally: shared sandboxes, aligned taxonomies, and secure data exchange to speed scaling and lower costs.
- Prepare for AI Act-style obligations: risk classification, documentation, testing, and content labeling/watermarking.
- Review progress quarterly: scale winners, sunset stalled pilots, and codify lessons into playbooks.
Signals for investors and partners
Central Asia's financial sector is moving from experiments to structured build-out. Adoption and budgets are rising, supervisory tech and digital assets are on the agenda, but gaps in infrastructure, talent, and governance still cap scale. That combination points to opportunity for capability-building, targeted platforms, and risk services that meet banks where they are.
For practical playbooks and tools, see AI for Finance and the AI Learning Path for CFOs.
Your membership also unlocks:
