Editor’s Note: Since 2023, Microsoft’s Office of Responsible AI has partnered with the Strategic Foresight Hub at the Stimson Center to convene a diverse group of experts from the Global South to evaluate the impacts of AI in emerging markets. Guided by the question of how AI-related risks and benefits might manifest in various social, cultural, economic, and environmental contexts, program participants identify technological and regulatory solutions that can help mitigate risks and maximize opportunities across the globe. Fellows also have the opportunity to publish at Stimson; in the RAI Case Studies, Fellows share insights about responsible AI governance from within their own thematic and geographic areas of expertise.
By Giulia Neaher, Managing Editor for RAI Case Studies
Introduction
In the late 1990s, the prospect of widespread internet connectivity and personal computers evoked images of boundless possibilities: streamlined government services, flourishing tech economies, and a burgeoning generation of digital natives. Fast forward to the present, much of this promise has indeed materialized — but unevenly and with distinct regional variations. Countries like Kazakhstan, Kyrgyzstan, Tajikistan, and Uzbekistan have navigated economic volatility, shifting geopolitical alliances, and infrastructural complexities to arrive at a moment filled with opportunity — especially with the rapid rise of mobile technology.
The reality that has unfolded tells a fascinating story of technological leapfrogging and adaptation, especially when it comes to mobile technology adoption. Recent data reveals a striking phenomenon: Mobile phone penetration in Central Asian countries exceeds 100% of the population, with many individuals maintaining multiple SIM cards for different purposes. In Kazakhstan, for instance, mobile subscriptions reach 130% of the population, while Kyrgyzstan reports similar figures. More remarkably, most of these connections are through 4G-enabled smartphones, with penetration rates exceeding 80% in urban areas in countries neighboring China. This organic evolution demonstrates how digital adoption can follow unexpected paths when shaped by local constraints and conditions.
This example of mobile-first development stands in sharp contrast to traditional computer adoption metrics. According to recent World Bank and International Telecommunication Union (ITU) reports, Central Asian nations significantly underperform in computers per household compared to other middle-income countries. In Kyrgyzstan, for example, fewer than 20% of households own a computer, while smartphone ownership exceeds the total population. This disparity reflects a conscious choice in technological development paths, where both consumers and governments have prioritized mobile connectivity over traditional computing infrastructure.
Existing Mobile-centric Development
This mobile-centric approach is perhaps best exemplified by Kazakhstan’s Kaspi.kz, which transformed from a traditional bank into a super-app ecosystem that now dominates the country’s digital landscape. Kaspi’s journey is remarkable: It has evolved to handle everything from peer-to-peer payments to marketplace transactions, effectively becoming Kazakhstan’s version of a combined PayPal, Amazon, and Commercial bank. The app processes over 45% of all digital payments in Kazakhstan, with monthly active users exceeding 14 million – remarkable in a country of 20 million people.
The region’s digital transformation has taken particularly fascinating turns in response to political and social constraints. In Uzbekistan, under the former Karimov regime, which maintained strict control over traditional media and communications, Telegram messenger evolved far beyond its intended purpose. What began as a simple messaging app transformed into a comprehensive digital ecosystem: 70% of Uzbek citizens turned to Telegram channels not just for peer-to-peer communication, but as their primary source of news, entertainment, and perhaps most significantly, commerce.
The platform’s group and channel features became de facto digital marketplaces, where entrepreneurs could reach customers directly without needing sophisticated technical infrastructure or having to navigate complex regulatory requirements. These Telegram-based businesses range from small handicraft sellers to large-scale retail operations, creating a parallel digital economy that thrived despite, or perhaps because of, the restrictive political environment. The platform’s encryption features and ability to host large group chats made it particularly resilient to government interference, while its minimal technical requirements meant anyone with a basic smartphone could participate in such a digital marketplace.
A similar phenomenon emerged in Kyrgyzstan, though driven by different factors. Despite numerous government initiatives to promote digital literacy, computer skills remained limited among the population. However, this didn’t prevent the emergence of a vibrant digital commerce ecosystem — it just took an unexpected form. Instagram, typically viewed in the West as a social media platform for sharing photos and building personal brands, transformed into Kyrgyzstan’s leading e-commerce platform with most of merchants using it to promote goods and services.
The platform’s visual nature and simple mobile interface proved perfectly suited for a population that was comfortable with smartphones but less familiar with traditional e-commerce websites. Small businesses, particularly in fashion, handicrafts, and food services, found they could build successful enterprises entirely through Instagram. Sellers use posts and stories to showcase products, direct messages for customer service, and maintain relationships with customers through regular engagement. The platform’s built-in features like location tagging and hashtags created an informal but effective marketplace infrastructure.
Such adaptation reflects a broader pattern in Central Asian digital development: Rather than building new platforms from scratch or waiting for traditional e-commerce infrastructure to emerge, entrepreneurs and consumers repurpose existing mobile-first platforms to meet their needs. This organic evolution demonstrates how digital adoption can follow unexpected paths when shaped by local constraints and conditions.
Another case study is the region’s experience with shared computing resources, best exemplified by the Institute of Smart Systems and Artificial Intelligence (ISSAI) at Nazarbayev University, which represents an innovative approach to creating public infrastructure for AI development. The model treats advanced computing capabilities as a public good rather than a private resource, democratizing access to the tools needed for AI research and development. The ISSAI approach could thereby serve as a template for other regions seeking to build AI capabilities without excessive duplication of expensive infrastructure.
Even more successful has been Kyrgyzstan’s ELQR initiative, a project launched by the National Bank that demonstrates how central banks can drive digital innovation. ELQR created a unified QR code system for the entire country, allowing any payment app to process transactions through a standardized system. This approach contrasts sharply with other regions where competing QR systems continue to create market fragmentation. The initiative has achieved remarkable success, with over 100,000 merchants adopting the system within its first year and online transactions increasing over 120 times in the two years since adoption, showcasing how governmental coordination can accelerate digital adoption.
The region’s technological leapfrogging, however, extends beyond mobile payments. Kyrgyzstan’s Tunduk system represents a revolutionary approach to digital governance. This interoperability platform, accessible through a mobile app, has transformed how citizens interact with government services. Through Tunduk, Kyrgyz citizens can access everything from birth certificates to property records instantly on their smartphones. To the present day, over 65% of the adult population has downloaded the mobile app, and almost a quarter of its users use the app daily to validate their identities. The platform processes over 200 million data exchanges monthly, demonstrating how digital identity systems can enable comprehensive e-governance. What makes Tunduk particularly remarkable is its architecture — instead of building a massive, centralized database, it is based on Estonia’s open interoperability framework, X-Road, that allows different government agencies’ systems to communicate seamlessly.
Furthermore, the region has also seen innovative adaptations in traditional sectors. In Uzbekistan, for instance, bazaar merchants who once relied entirely on cash transactions now routinely use mobile payment systems, with some markets reporting over 80% of transactions occurring digitally. This transformation hasn’t required complex technology — simple QR codes and basic smartphones have been sufficient — but it has fundamentally changed how commerce operates in these traditional spaces.
The region’s experience offers some valuable lessons about technological development. First, infrastructure matters immensely — but not necessarily in the way we once thought. Rather than focusing on laying cables for home internet connections (last mile connectivity), successful initiatives prioritized mobile network coverage. Second, technology adoption follows local patterns and needs, not predetermined Western templates. The low computer penetration rates, rather than indicating technological backwardness, reflect a conscious Skip/Shift of the personal computer era in favor of mobile-first solutions.
As we now stand at the dawn of the era of artificial intelligence, these lessons from Central Asia’s digital transformation might become particularly relevant. The region’s experience suggests that emerging technologies do not necessarily follow the deployment patterns we expect. Instead, the success of platforms like Tunduk demonstrates how technology can be both sophisticated and inclusive, serving populations regardless of their technical literacy.
Opportunities
Looking ahead, companies and policymakers should re-examine existing expectations for AI’s development trajectory, focusing on several key areas:
First, infrastructure flexibility is crucial. Rather than building rigid systems designed for current AI applications, adaptable infrastructure is needed to evolve with technological advances. The success of mobile networks and interoperability platforms in Central Asia demonstrates how flexible infrastructure can enable unexpected innovations.
Second, inclusion must be at the forefront of AI development. Central Asia’s experience shows that technological sophistication and accessibility aren’t mutually exclusive. Future AI systems should be designed with similar principles — sophisticated enough to handle complex tasks but intuitive enough for users of all technical backgrounds.
Third, we must embrace technological leapfrogging while ensuring no one is left behind. The success story of platforms like Kaspi and ELQR illustrates how digital solutions can achieve mass adoption when they address real needs and are designed with user experience in mind. AI development should follow similar principles, focusing on solving tangible problems while maintaining accessibility.
Fourth, mobile-first AI development will be crucial for regions like Central Asia. Given the overwhelming prevalence of smartphones over computers, AI solutions need to be optimized for mobile devices, too. This might mean developing lighter, more efficient AI models (Edge AI or TinyAI solutions) that can run on and be built into smartphones instead of requiring powerful cloud servers.
Finally, interoperability and standardization are crucial. The success of Kyrgyzstan’s Tunduk and ELQR systems demonstrates how standardized platforms can accelerate digital adoption. As AI systems become more prevalent, we’ll need similar frameworks to ensure different AI solutions can work together seamlessly while maintaining security and privacy. An exemplary framework being the Model Context Protocol (MCP) open-sourced by Anthropic, standardizing how AI systems are connected to external tools and data sources. MCP clients can either expose their data through MCP servers or build AI applications to connect to these data sets, eliminating the need for building custom integrations on a model-by-model basis.
These examples of platform adaptation in Uzbekistan and Kyrgyzstan offer important lessons for future technological development. They demonstrate that successful digital transformation doesn’t necessarily require purpose-built solutions or comprehensive infrastructure. Instead, users will creatively adapt available tools to meet their needs, often in ways that platform developers never anticipated. The pattern of organic adaptation and repurposing could prove particularly relevant as AI technologies become more widely available.
Takeaways
The role of government in AI development also deserves special attention. Central Asian nations have shown that state involvement, when properly executed, can accelerate digital transformation. As we move into the Age of AI, governments need to balance the promotion of AI development with appropriate regulation and governance. The region’s experience with digital identity systems like Tunduk might offer valuable lessons about managing sensitive data while maintaining public trust.
Central Asia’s technological journey over the past three decades thereby serves as a reminder that innovation doesn’t always follow predicted paths. The region’s unique pattern of high mobile penetration coupled with low computer adoption challenges conventional assumptions about technological development. Perhaps the most important takeaway is that technological advancement can be both sophisticated and accessible, serving entire populations rather than just tech-savvy early adopters. The next wave of AI innovation must similarly balance capability with inclusivity, ensuring that AI becomes a tool for broader societal development rather than a force for digital division.
In the end, Central Asia’s experience suggests that successful technological development isn’t about following established paths but finding solutions that match local needs and capabilities. As we look toward an AI-enabled future, this lesson becomes increasingly relevant. The challenge will be to develop AI systems that, like the best mobile solutions in Central Asia, are both powerful and accessible, sophisticated and inclusive.
