In 2019, I boarded a one-way flight from New York to Accra with a simple ambition: help tackle youth unemployment. The scale of the problem was already well-documented. Millions of young people enter the labor market every year, most of them greeted not by a job offer but by informal work, under-employment, or the decision to leave entirely. Like most early-stage entrepreneurs, I did not yet have a complete solution. But I had a conviction: that data skills, applied to companies that were flying blind on instinct, could create a new category of employment that the continent needed. That conviction became the foundation for Blossom Academy.
The early years were a slow grind of persuasion. We were not just training young people; we were convincing organizations that analytical talent could transform how they made decisions. A few took the chance. Those early partners became our most powerful advocates, and as our graduates began delivering results inside their companies, the word spread. Slowly, skepticism gave way to demand. Hundreds of young people moved into data careers. For a few years, we felt like we were ahead of the curve. Then the curve moved.
Companies that had once hired four or five of our graduates at a time began hiring one, sometimes two. We assumed a competitor had emerged. The truth was both simpler and more unsettling: the competitor was AI. The same platforms we had helped these companies adopt were now doing in seconds what teams of analysts had previously spent days on. We had succeeded in making our partners more data-driven. In doing so, we had helped build the very infrastructure that reduced their need for the people we were training.
I want to be honest about what that moment demanded of me. It would have been easy to dismiss it as an edge case, a blip in an otherwise positive trajectory. But I had seen enough to know that naivety about automation carries actual costs; not abstract costs, but the kind measured in the careers of specific people who trusted us to prepare them for the future. The honest reckoning was this: betting on a single technical skill set, however promising it looks today, is no longer a viable strategy. Technology does not wait for your program to catch up.
Before that experience, I would have read the AI conversation the way most people in my position do; with cautious optimism, nodding at the opportunities, quietly assuming the disruption would land somewhere else. It didn’t. And that is why so much of the current conversation about AI in Africa leaves me uneasy.
The dominant framing today is one of competition. Can African countries build their own large language models? Can they close the compute gap? Can they secure a seat at the table of the global AI race? These are important questions, and they deserve serious attention. But for policymakers and development partners thinking about employment and economic transformation, they may be the wrong starting point. They orient us toward a race we are not positioned to win, while distracting us from a contest where the structural advantages are already ours.
Consider agriculture. Africa holds roughly 60 percent of the world’s available arable land, yet the continent imports billions of dollars’ worth of food annually– not because the land is unproductive, but because the systems built around that land remain chronically underdeveloped. During a visit to a village near the Côte d’Ivoire border, I watched foreign buyers move through local farms sourcing raw cashew nuts for export to Asia. Those cashews travel thousands of miles to be processed, packaged, and, in a bitter irony, sold back into African markets at a premium. What leaves with those trucks is not just commodity value. It is an entire ecosystem of jobs in processing, logistics, and distribution that could, and should, exist within the communities where the crops are grown. And building that ecosystem is precisely the conversation about AI that Africa’s policymakers need to be having.
In a program we are currently implementing with a United Nations agency, we are exploring precisely this intersection. We’re training young professionals to apply AI tools across agricultural value chains. What has become clear, faster than I expected, is something that surprised even us: the biggest barrier to investment in these value chains is not capital, it is legibility. Investors and development partners may not be avoiding African agriculture because they lack appetite. They may be avoiding it because the data needed to make a confident decision – where the losses are, what the infrastructure gap actually costs, what a credible return looks like – has never been systematically assembled. AI changes that. It is not, in this context, a displacement force. It is a visibility tool.
Critics would argue: if AI disrupted data roles, why wouldn’t it eventually do the same to agricultural jobs? The answer lies in understanding what kind of disruption AI actually causes and where. In mature, digitized industries, AI replaces tasks that are already well-defined, repetitive, and optimized. But African agriculture does not suffer from overcapacity. It suffers from the opposite: missing infrastructure, fragmented markets, absent cold chains, and financing gaps that keep the sector locked in subsistence rather than scaled production. In this environment, AI does not arrive to replace a functioning system. It arrives to help build one. And when those missing layers; the processing plants, the distribution networks, the aggregation hubs, are finally constructed, they bring jobs at every level of the value chain.
This reframing matters enormously for how we think about skills and workforce development. For years, the instinct has been to double down on technical training: more coders, more data scientists, more AI engineers. Those skills remain valuable. However, the workforce that will actually drive Africa’s economic transformation will look different. It will combine deep domain knowledge in sectors like agriculture, mining, and logistics – knowledge built from proximity, from lived experience, from understanding how a supply chain breaks down in the rainy season – with the capacity to deploy AI tools that make that domain knowledge more powerful. A young professional who understands both the mechanics of a smallholder cooperative and how to use predictive analytics to manage its inventory will be more resilient, and more valuable, than someone trained only in abstraction.
Africa does not need to win the global AI race. It needs to win the application race by using these tools to unlock productivity in the sectors it already owns. The opportunity is not in the model, but in the land, the labor, the value chains that remain only partially built, and the young people ready to finish building them if we give them the right tools and the right frame. The continent’s future of work will not be decided by who builds the most powerful models, but by who uses them to transform what already exists.
