The same firm publishing the warning is selling the AI and offering a method for evaluating whether Africa should buy.
Akua Gyekye is making a public argument: the continent's projected $1tn AI dividend by 2035 will slip away unless governments, businesses, and citizens trust the systems being built. The African Development Bank has named trust, alongside data, computing capacity, skills, and capital, as one of five conditions for capturing that gain.
The firm making that case is also one of the largest sellers of the AI systems Africa would adopt. Gyekye's framing, circulated this month through a PUNCH op-ed and echoed by BusinessDay NG, treats trust as something to be designed into AI from the start, not bolted on after deployment. That choice has consequences for which "trust" standards the continent ends up measuring itself against.
The AfDB estimates that inclusive development and deployment of AI could add as much as $1tn to African GDP by 2035, equivalent to nearly one-third of the continent's current economic output. About 58% of that gain, or roughly $580bn, is projected to come from five sectors: agriculture, wholesale and retail, manufacturing, financial services, and healthcare. These figures are projections to 2035, not realized value.
The trust gap is already visible in adoption. Microsoft's own Global AI Diffusion Report for Q1 2026 found that generative AI usage among working-age populations stood at 27.5% in the global north versus 15.4% in the global south. The 12-point gap is one of the few concrete, vendor-acknowledged measurements of the trust claim. That gap may reflect trust, infrastructure, or a mix of the two. Gyekye's column is trying to define which.
What does "trust" mean in this context? Gyekye's argument runs through Microsoft's broader Africa data-governance work. In a separate Microsoft feature on data governance, the company positions itself as a partner on localized data centers, sovereign cloud offerings, and training programs. Trust, in that frame, becomes less about external regulation and more about whether Microsoft's own deployment choices, data residency, and audit practices meet the standards it is calling for. A trust framework written with the vendor in the room is different from one written against it.
The AfDB's five-condition checklist is the concrete lever in the story. Data, computing capacity, skills, capital, and trust are each something governments, donors, and companies can fund, build, or refuse. Treating trust as a buildable input rather than a soft requirement is what gives Gyekye's column its policy weight. It is also what makes the vendor framing worth reading carefully: Microsoft is selling the AI and offering a method for evaluating whether to buy.
Trade-press coverage has, to date, reported the warning at face value. The PUNCH and BusinessDay NG both framed the message as a Microsoft call for African policymakers to act on trust. Who actually sets the trust standard has been left for African regulators, multilateral lenders, and the continent's own AI strategies to take up. The African Union's continental AI strategy and the Africa AI Governance Index are where independent definitions of trust are being written.
The trust question is no longer whether the continent will adopt AI. It is whether the systems adopted are the ones Africa would have chosen.