Chapter 3: Scaling

With the ratification of the National AI plan, the closing of zypl.ai’s seed round, and TajRupt’s student enrollment in the hundreds, Azimi returned home after his graduate studies with expanded roles across Tajikistan’s private, civil, and public sectors. While the AI strategy set ambitious benchmarks for the entire country, much of the responsibility of achieving those targets fell on Azimi and his team. They identified three distinct approaches to scale their existing capabilities to achieve the AI Strategy’s goals. Would these three be enough? How might they best contribute to the country’s progress? How could Azimi be sure their three initiatives worked synergistically rather than at cross-purposes?

Area AI

In his master’s thesis, Azimi called for the creation of curated economic zones with special-purpose laws to promote business creation and growth through greater collaboration between the private and research sectors. Azimi envisioned Dushanbe, the capital of Tajikistan, becoming “Area AI” of not just the country, but the entire region.

Azimi’s vision for a curated economic zone built upon his experience at Stanford University in Silicon Valley, where business leaders, researchers, and top technology talent lived and worked nearly side-by-side. “Area AI” would attempt to recreate this dynamic by bringing AI infrastructure closer to the nation’s top universities while launching initiatives and incentives to attract both skilled talent and AI companies. In order to realize the economic impacts at the order of magnitude of Tajikistan’s new AI economic targets, “Area AI” would not only have to spur the creation of several new high-growth AI startups, but also attract a major multinational company to establish an office in the city.

Digital Decolonization 

Digital decolonization was a theory popular among investors in emerging markets. It posited that global technology titans from wealthier economies tend to under-invest in the distribution channels, infrastructure, and partnerships needed to capture small to mid-cap opportunities in frontier markets historically dominated by external parties. This creates  opportunities for indigenous tech entrepreneurs to develop solutions tailored to local needs, sometimes producing billion-dollar valuations in the process, and enabling the benefits of the digital technologies and wealth creation to remain in the community rather than being transferred to foreign corporations.

The merit of this hypothesis became evident in the early 2000s. zypl.ai’s lead investor, David Halpert, had witnessed it firsthand with two of his portfolio companies: Gojek, Indonesia’s ride-hailing giant, and Andela, Nigeria’s online talent marketplace. Both had exploited gaps left by multinationals, built localized solutions, and scaled into “unicorns”, valued at over one billion US dollars. After creating previously unseen business success in their countries, many early employees from these startups, flush with cash after lucrative exits, went on to start new ventures, fueling the growth of their respective startup ecosystems and mirroring the dynamic often seen in Silicon Valley, like with PayPal’s “mafia”.

Azimi hoped to create a similar phenomenon through Zypl.ai. For example, two precocious TajRupt students turned Zypl.ai interns, Nazirjon and Samandar, built an algorithm that verified banking information using passport documents. Once deployed, it generated $5,000 in monthly recurring revenue for Zypl.ai. Even without an exit, Azimi envisaged a scenario where TajRupt students could test and deploy fintech machine-learning solutions through Zypl.ai, and receive funding from its shareholders, including from Azimi himself, to start their own ventures. This would foster local entrepreneurship and also allow Azimi to help set the terms for venture capital in Tajikistan, where no such financing structures currently existed. 

To Halpert, digital decolonization presented not only a savvy investment opportunity, but the political inevitability of developing nations refusing to have a global economy that circled entirely around six zip codes in California and Washington State, concentrating wealth and power in such few hands. Halpert saw a world where local people demanded a greater say on what appeared on their phone, from food delivery options and news feeds to educational content and holiday entertainment. The great tech companies - be they in US, China, or Japan - were starting to realize that they couldn’t simply expect Asians, Africans, and Latin Americans – Google’s famous Next Billion – to just sign up as customers for the entire suite of digital products that they sold back home in Atherton, California, without some significant local input. That local input, Halpert believed, would become rapidly more sophisticated as emerging markets quickly developed greater digital familiarity and understanding of employee ownership, incentivizing entrepreneurs like Azimi to build locally, serve their home markets, and retain control. 

Halpert, seeing parallels between zypl.ai and his past unicorn successes, urged Azimi to think big and avoid an early exit. “If zypl.ai achieved the right scale,” Halpert argued, “a billion-dollar exit could drive a fivefold increase in GDP over a decade”—a timeline extending beyond the AI Strategy’s targets but with transformative long-term implications.

Upskilling the Next Generation 

In Tajikistan’s "Concept of Digital Economy until 2040" workforce agenda and in the President’s 2021 annual address, the national government emphasized the need to develop curriculum offerings focused on critical thinking, foreign language mastery, and postsecondary study abroad, goals directly aligned with TajRupt’s program design. The national vision was based on the premise that enabling Tajik youth to pursue undergraduate studies at top global institutions would both enhance the country’s human capital and help students achieve their aspirations. One such example was Khosiyat, one of TajRupt’s earliest students, who became the first native Tajik undergraduate at Stanford University.

TajRupt’s early successes, along with Azimi’s rising national profile, opened the door for partnerships with schools and educational leaders across the country eager to integrate elements of its curriculum into public school programs. Seeing this opportunity, the AI Council’s Education team began adapting the AI Academy’s curriculum for nationwide deployment, aiming to scale its impact in alignment with the AI Strategy. By 2026, their vision was for every high school student in Tajikistan to have access to an “AI for Beginners” elective, providing foundational machine learning skills. University students who remained in Tajikistan could then advance to “AI Full” courses at the postsecondary level.

Under this model, TajRupt’s curriculum, blending technical machine learning with entrepreneurship and leadership, would be adopted by all schools throughout Tajikistan, giving every Tajik student the opportunity to expand their horizons, follow their dreams, and learn high-demand skills. Although not all TajRupt students would pursue AI career paths, the nation would be enriched through remittances from a higher skilled Tajik diaspora and economic output produced by those who stayed local, like the TajRupt alumni working at zypl.ai.

In many ways, this initiative to scale TajRupt’s educational impact built on Azimi’s original mission: to cultivate a new generation of Tajik-educated changemakers, both at home and abroad.