Last year, we surveyed firms in emerging markets backed by private equity (PE) and venture capital (VC) funds, which we refer to as frontier firms, and found that almost 40 percent were using artificial intelligence (AI). This year, the picture looks considerably different.
The International Finance Corporation’s (IFC) latest survey of 1,666 frontier firms with significant activity in emerging markets, shows that AI has moved into the mainstream: 71 percent now report using AI, whereas half of all adopters say they began using AI during 2025. As the World Development Report 2026 highlights, AI offers developing economies a pathway to boost productivity, but only if it is adopted, adapted to local conditions and used at scale. Frontier firms are typically the first to adapt global technologies to local contexts, and pave the way for other firms to follow.
Adoption, however, remains shallow for many firms. We distinguish between internal adopters, which use AI in their own operations, and external adopters, which build it into products or services for customers. More than half of internal adopters (57 percent) and nearly half of external adopters (45 percent) are still experimenting with AI or deploying it on a limited scale. Far fewer have made AI central to how they operate: just 10 percent of internal adopters have embedded it across their core internal processes, and 19 percent of external adopters describe AI as core to their business model.
What prevents firms from moving further up the AI ladder? Skills constitute the most frequently cited constraint, ahead of other commonly discussed barriers such as costs or technical infrastructure.
When firms were asked about barriers to (further) AI adoption, more than four in ten non-adopters identified skills shortages as their main obstacle, compared with about one quarter of current adopters.
Perceptions of barriers also shift with depth of adoption: firms using AI more extensively are more likely to say their current level of adoption is appropriate. This could reflect selection, as more capable firms both deploy AI more deeply and face fewer constraints, or experience, as firms that successfully move from pilots to broader use develop clearer use cases and invest more heavily in AI as a result. Among shallow adopters, however, the same answer may partly reflect limited awareness of what deeper use of AI could offer their business.
