What broke in 2025
In Q1 2025 USAID was effectively dismantled and folded into the State Department. The administration's stated policy was to route remaining funds directly to host-country governments, bypassing the INGO consortia that had historically converted those grants into operational primary-care delivery. Within months, the implementation organizations that ran the world's largest CHW programs began mass layoffs and program closures.
What INGOs actually did
The dollar number gets the headlines. The structural problem is the function that disappeared. INGOs weren't just disbursement vehicles. They were the operating layer that converted grant money into trained, supervised, supplied health workers. That layer is what nobody is building back.
Ministries are operating. They're stretched in two directions at once.
Ministries inherited the responsibility. They are spending what's flowing direct. Patients are still being seen, nurses are still showing up, CHW networks are still attached to district health offices. The system isn't paralyzed — it's stretched. Stretched on budget (~20% pure shortfall), and stretched on capability (the trainers, supervisors, and M&E officers that used to come with the money aren't there anymore).
Building INGO-equivalent in-house operational capacity from scratch — hiring trainers, building M&E systems, running supervisory cycles — is a decade-long project that costs more than ministries have in absolute dollars. The donor cycle that used to fund that work is itself the thing being unwound. Software is the only substitution path that fits the budget. That's why this is a venture opportunity rather than a development-aid opportunity.
International aid was always supposed to make itself unnecessary.
The forty-year unspoken contract of foreign aid is that the recipient eventually graduates from it. In practice, off-ramp planning has been structurally absent — politically uncomfortable for donor governments to discuss, harder still for recipient governments to publicly endorse. The 2025 cuts forced the conversation by removing the alternative. The transition is brutal. The destination — country-level independence from donor cycles — is one almost everyone in global health agrees on in principle.
luma sits on the off-ramp side of that argument. We are not trying to rebuild USAID-as-it-was. We are not selling "donor-as-a-service." We are building software that lets ministries do the operational work themselves — durably, at scale, paid for by a commercial wedge that doesn't depend on the donor cycle at all. The pharma RWE revenue stream subsidises the ministry license. The same operational data feeds both. As the ministry pays less and the pharma layer pays more, the share of luma's cost recovered from international aid trends to zero. That's the off-ramp.
Software-native substitute for the INGO operational layer
Most of what INGOs did was operationally heavy but conceptually patternable. Supervision is "review case against protocol; flag deviation; suggest escalation." M&E is "collect indicator; aggregate; surface anomaly." Technical assistance is "translate WHO guideline into district workflow; train; iterate." All three were delivered by humans because the technology to do them wasn't viable. In 2025 it is.
- Multi-year USAID/PEPFAR cooperative agreement signed with an INGO consortium
- 15–30% NICRA indirect cost rate on top of direct programmatic spend
- Expat technical advisors + local in-country staff (50–500 per major program)
- Quarterly clinical mentor visits, quarterly M&E reports, annual training cycles
- Per-CHW operational cost typically $1,500–$7,500/yr all-in
- 3–6 month feedback loops between CHW practice and program adjustment
- Direct contract with the ministry of health — no donor intermediary required
- No NICRA. Standard SaaS pricing with volume tiers per CHW
- 2–4 in-country staff total per country (deployment + clinical localization)
- Real-time decision support, real-time M&E, weekly performance dashboards
- Per-CHW operational cost $250/yr at standard tier
- Sub-second feedback loop between CHW question and protocol-grounded response
Cost-substitution calculator
What it cost to supervise a CHW network through an INGO partnership versus what it costs to run the same network on luma. Move the sliders. The math is intentionally simple and conservative — full sources at the bottom of the calculator.
How the numbers were chosen. INGO operational cost per CHW is sourced from PEPFAR Country Operational Plan partner-level breakouts (where published) and triangulated with the academic literature on CHW program economics (per-CHW cost ranges of $1,500–$7,500/yr cover most published Lesotho/Botswana/Malawi/South Africa program evaluations). NICRA indirect rates of 15–30% are applied to direct cost components and included in the INGO total. luma's $250/CHW/yr is the published standard tier for the workflow line (see pricing). Five-year horizon assumes a 5% CHW network growth rate.
This is a 5–7 year window
The functional gap created by the INGO collapse will be filled by something. In the optimistic scenario, a patchwork of African-government implementation units rebuilds slowly over a decade. In the realistic scenario, software substitutes show up first because they're cheaper to deploy and faster to scale. The vendor that establishes the workflow standard during this window owns the African primary-care data layer for the next twenty years.
Whoever owns the CHW interaction layer through 2030 owns Africa's primary-care data layer through 2050.
Switching costs in workforce-software are extreme. Once a ministry's entire CHW network has trained on a particular interface, integrated their protocols, and built reporting around the data structure, replacement requires a multi-year reconfiguration that ministries don't have the bandwidth to undertake. That's the moat — not the chatbot, not the projections, not any single feature. The moat is being the standard at the moment ministries had to choose one.
luma is moving on that window now. See the operational dashboard · see the four-country expansion plan · read the investor brief.