There is a pattern forming in how startups get built outside the US and China, and it has a name.
Investor William Bao Bean describes it as “AI with emerging markets characteristics”: a phase in which the constraints founders actually face, in infrastructure, capital, and distribution, produce startup models that look fundamentally different from those emerging in better-resourced markets. Rather than copying playbooks written elsewhere, founders in these markets design around the limitations in front of them, and AI is increasingly the tool that turns those limitations into workable business models.
The idea surfaced in a conversation on A Global Tech Podcast and framed the thesis behind a recent round of investments from Orbit Ventures, which included afina alongside Bump, CYPIEN AI, and Tezbor. Each of those companies applies AI to a constraint specific to the market it operates in. For afina, that constraint is one every telecom operator in an emerging market recognizes.
In many of these markets, the identity signals that power digital advertising elsewhere are thin. Third-party cookies are disappearing, mobile advertising IDs are fragmented, and reach is scattered across platforms and devices in ways that make precise targeting difficult. The infrastructure that advertisers in mature markets take for granted is either absent or unreliable.
Telecom operators sit in a different position. They already hold verified first-party data on their subscribers, and they already have the audience at scale. What has historically been missing is the mechanism to turn that data into advertising performance without exposing personal information or exporting sensitive records outside the network. This is the gap afina’s data monetization platform is built to close. By applying machine learning to network-level behavioral data, afina converts an operator’s existing first-party data into audience segments that advertisers can target, with privacy compliance built into the architecture.
That is what makes afina a working example of the pattern Bao Bean describes. The company is not importing an advertising model designed for markets with abundant third-party identity signal. It is building around the specific reality of the markets it serves, where the operator’s own data is the most reliable signal available, and using AI to activate it. The constraint that looks like a disadvantage, limited third-party identity infrastructure, becomes the reason the operator’s first-party data matters more, not less.
For telecom operators weighing how to monetize subscriber data in these conditions, the strategic takeaway is straightforward. The advertising value of first-party network data rises precisely as third-party alternatives erode, and the operators who build the capability to activate that data now are the ones positioned to define how it earns over the next several years.
