A wave of artificial intelligence tools is spreading through African health supply chains, automating everything from procurement planning to fraud detection, but new research warns that almost all remain dependent on donor funding with no clear route to financial sustainability.

A study released on Monday by Salient Advisory, a US-based consultancy, and funded by the Bill & Melinda Gates Foundation, identified at least 20 AI-enabled solutions deployed across the continent. They target persistent weaknesses such as inaccurate demand forecasting, fragmented data systems, and time-consuming manual paperwork. Yet of five case studies examined in depth, only one generates revenue directly from its users. The rest are propped up by grants from the Gates Foundation, the Global Fund and other development financiers, raising questions about their future once external support wanes.

“Evidence of cost savings appears thin, representing a critical gap in understanding,” the report states, in an unusually frank assessment of a sector that has drawn growing interest from both technology vendors and global health donors. All impact figures cited are self-reported by the solution providers and have not been independently verified, the authors caution.

The inventory of tools spans from an Ethiopian government-mandated forecasting platform that claims to have saved $38m in planned procurement expenditure, to a Kenyan procurement system that slashed planning time from days to under an hour per facility, to a Nigerian document-processing agent that cut invoice-to-pay cycles from 13 to three days. In Morocco, a private-sector inventory management service said it had reduced pharmacy stock levels by 20 per cent.

But the report’s deep-dive profiles expose the fragile underpinnings of many early deployments. In Ethiopia, Opian Technologies’ ForLab Plus platform — used by more than 5,700 public health facilities — achieved national reach after the government required facilities to submit forecasts and procurement commitments through the system. Its development and rollout, however, were funded by the Gates Foundation, with “a long-term aim to shift to government payments,” the report notes. Whether Addis Ababa will allocate budget lines for the service remains an open question.

In Kenya, the SMArT tool, co-created with the Ministry of Health, now supports procurement planning across around 1,000 facilities in five counties and has won national government endorsement. But the project is financed by the Patrick J. McGovern Foundation and the Gates Foundation. A proof-of-concept in Ghana, designed to unify fragmented HIV commodity data and project an 80 per cent improvement in supply chain visibility, has been paused since 2024 because of a lack of funds, the report says.

Only Distripha, a web-based platform that helps private pharmacies in Morocco and Spain manage inventory, operates on a commercial subscription model. The company says its AI engine has reduced internal customer support needs by “10–20x”, a signal that well-designed tools can scale without ballooning costs. But the contrast with donor-backed initiatives is stark.

The sustainability problem is compounded by a pervasive lack of hard data on cost savings. Of the five solutions profiled, just one — ForLab Plus — presented a primary impact metric linked to cost reduction. Others emphasised operational gains such as time saved, document processing speed, or stock availability. “Solutions that cannot articulate a credible sustainability plan beyond the pilot phase may struggle to attract the multi-year commitments needed to demonstrate and sustain impact,” the report warns.

This picture of promise tethered to grant financing arrives as official development assistance for health is declining, forcing governments to absorb programmes once underwritten by foreign donors. The funding squeeze makes the absence of clear commercial or publicly budgeted pathways for AI tools an acute vulnerability.

The report’s authors urge solution providers to develop “data-backed business cases” with verified cost offsets, and to embed change management as a core requirement, not an afterthought. Global health organisations, they argue, should use their influence with large AI firms to negotiate equitable access to infrastructure for African innovators, including subsidised licensing and compute credits. They should also fund independent impact evaluations to establish what works and under what conditions.

For governments, Salient Advisory advises a sober assessment of whether to buy off-the-shelf tools, build proprietary systems, or partner with local providers. It lays out a framework based on data quality, existing system architecture, policy constraints and available funding. In practice, many ministries may find they have strengths in one dimension and critical gaps in another, making the procurement decision fraught.

“Governments should establish a baseline understanding of current supply chain costs before procuring AI solutions,” the report says, noting that without such a baseline, any claimed savings remain unverifiable. It also calls for updated decision-making processes that ensure AI-generated insights actually inform planning — a human-in-the-loop principle that proved essential for building trust in the Kenyan SMArT tool.

Despite the cautionary tone, the research signals that AI is beginning to move beyond pilots and into operational use. Sixty per cent of identified solutions are deployed exclusively in health supply chains, and 85 per cent are enterprise or mid-scale platforms, suggesting the market has progressed past early experimentation. A third of providers are headquartered in emerging markets across Africa, Asia and Latin America, reflecting tools shaped by local operating realities.

“What is needed now is coordinated action to build on these early gains,” the report concludes. Without it, there is a risk that even the most promising tools join the long list of donor-funded digital health pilots that flickered briefly and then faded.