SparkLabs and Mirae Asset’s Central Asia AI Fund Signals a Shift in Venture Geography

The $50 million fund targets Series A and later-stage AI startups in a region previously overlooked by major investors, but its success hinges on local infrastructure gaps and talent retention.

SparkLabs and Mirae Asset have launched a venture fund to back AI startups in Central Asia, according to Pulse 2.0.

The move is notable not just for its focus on AI but for its geographic ambition: Central Asia has rarely attracted dedicated venture capital for AI, despite growing local tech hubs in Kazakhstan and Uzbekistan.

The fund’s emphasis on Series A and later-stage companies suggests confidence in the region’s ability to scale, but it also faces challenges unique to emerging ecosystems.

Why Central Asia? And Why Now?

The fund’s launch aligns with two broader trends. First, venture capital is increasingly looking beyond Silicon Valley and even traditional secondary hubs like Berlin or Singapore.

Recent funds like Misumi’s $50 million robotics and industrial AI fund in the US per Dealroom and the Siri-linked VC’s Japan-focused AI and space fund per Nikkei Asia show a similar geographic diversification.

Second, AI startups no longer require proximity to major tech hubs—cloud infrastructure and remote collaboration tools have lowered barriers.

But Central Asia isn’t just another underfunded region. Its appeal lies in lower operating costs compared to Eastern Europe, a growing pool of engineers, and governments actively courting tech investment. The risk, however, is that the fund’s focus on later-stage startups may clash with the reality of Central Asia’s ecosystem: many local AI ventures are still in seed stage, and scaling them requires solving infrastructure and talent gaps that Series A funding alone can’t address.

The Infrastructure Problem

AI startups in Central Asia face unique hurdles. While cloud providers like AWS and Azure have expanded locally, latency and data sovereignty issues persist. Training large models still often requires partnerships with foreign data centers, adding cost and complexity.

The SparkLabs-Mirae fund doesn’t explicitly address this—unlike OpenAI’s venture fund, which invests in companies that can directly leverage its infrastructure per the WSJ.

The fund’s success may hinge on whether it can help portfolio companies navigate these challenges. Without parallel investments in local compute infrastructure, even well-funded Central Asian AI startups could struggle to compete globally.

Talent Retention vs. Global Competition

Central Asia produces strong technical graduates, but retaining top AI talent is difficult. Many engineers leave for higher salaries in Europe or North America, and those who stay often work remotely for foreign firms. The fund could help by creating high-value local jobs, but it will need to demonstrate that its portfolio companies can offer career growth comparable to global alternatives.

This isn’t just a Central Asian problem—Japan’s AI fund similarly aims to keep talent domestically per Nikkei Asia.

The difference is that Japan’s ecosystem already has mature startups and corporate R&D labs. Central Asia’s AI scene is younger, making talent retention even more critical—and harder.

A Test Case for Regional AI Hubs

If the SparkLabs-Mirae fund succeeds, it could prove that AI innovation isn’t confined to a handful of global hubs. But “success” here means more than financial returns—it means fostering startups that can scale without relocating, attracting follow-on investment, and building a sustainable talent pipeline.

The fund’s focus on Series A and later stages suggests optimism about the region’s readiness. I’m less convinced. Central Asia’s AI ecosystem would benefit more from seed-stage funding paired with infrastructure support. Without that, even promising startups may hit ceilings the fund can’t lift.

For AI builders, the takeaway isn’t just about a new funding source—it’s about whether emerging markets can truly compete in AI without first solving foundational gaps. The answer will shape where the next generation of AI agents are built. For more on global AI agent development, explore the AI agent directory.