NetHope’s AI Report Finds Open Source Can Cost NGOs More Than Licenses

A report built to showcase artificial intelligence wins inside humanitarian groups just admitted that open source software can cost NGOs more than the commercial licenses it was supposed to replace. The finding sits inside NetHope’s newest synthesis of AI deployments at the International Rescue Committee, Mercy Corps, the International Committee of the Red Cross (ICRC) and eight other organizations. It challenges a rule the sector has treated as settled for fifteen years.

Four corporate partners sit on the inside cover of the center that produced the report, and every one of them sells the kind of proprietary software the finding just made a stronger case for.

The Sentence Buried in NetHope’s AI Report

NetHope, a consortium that has spent two decades helping non-governmental organizations adopt technology, published Harnessing AI for Humanitarian Impact earlier this year. The briefing synthesizes 11 case studies of AI deployments from 2024 and 2025, featuring the International Rescue Committee, Danish Refugee Council, Humanitarian OpenStreetMap Team, Pacific Disaster Center, the World Health Organization, Norwegian Refugee Council, Mercy Corps, Catholic Relief Services, ICRC, GiveDirectly and the DEEP Consortium.

Most of it reads like a highlight reel. Displacement forecasting. Flood-triggered cash transfers piloted in Nigeria and Bangladesh. Multilingual chatbots built for displaced populations. Then the cross-cutting analysis reaches licensing, and the tone changes.

Open-source does not eliminate cost, it redistributes it. Licensing fees are replaced by coordination demands: maintaining codebases, documenting tools, onboarding contributors, and governing shared infrastructure. For resource-constrained NGOs, these coordination costs can equal or exceed the price of a commercial license.

NetHope’s own briefing says that, not a critic on the outside. The concession undercuts the total cost of ownership (TCO) math the sector has used for years to justify open source mandates over commercial alternatives.

Follow the Funding

The disclosure the report doesn’t spell out sits on the inside cover of the NetHope Center for the Digital Nonprofit, the unit that produced the analysis. Four partners are named there.

None of that makes the coordination-cost finding wrong. The case studies are documented in detail, and the open-source critique is one paragraph inside a much longer analysis that also holds up fAIr and POMELO as models worth copying for publishing their code on GitHub. Every named partner still has a commercial reason to want NGOs spending less time on open source. That’s a real conflict, and the finding still holds up against evidence collected outside NetHope entirely.

Is Open Source Free for Cash-Strapped NGOs?

Not by itself. Open source software carries no license fee, but someone still has to patch its security holes, document it, train new contributors and keep the servers running. For an NGO without a dedicated engineering budget, that unpaid coordination work is where the real cost shows up.

Open source keeps growing anyway. The Open Source Initiative’s newest survey found that fewer than 2% of organizations cut their open-source use in the past year, based on more than 700 responses spanning a dozen industries. That’s not evidence the coordination-cost warning is wrong. It’s evidence organizations are absorbing the cost rather than walking away from the software.

NetHope frames open source as a cost-reallocation strategy rather than a cost-reduction one. The money that would have gone to a vendor invoice moves to a staff calendar instead, where it’s harder to see and easier to underfund.

When the Funding Cycle Ends, the Tool Follows

NetHope’s coordination-cost line matches what independent organizations, none of them funded by Microsoft, have already learned by watching their own tools nearly die.

OpenLMIS had three options once its original grant funding ran out.

• Charge implementations for ongoing maintenance
• Expand to private clinics as paying customers
• Hand core software maintenance to a private partner

It picked the third option. The supply chain tool survived because a private company needed it inside its own commercial pipeline, not because the code was free to copy.

Development Gateway made a similar argument last year, warning that treating open source as automatically sustainable is “a misunderstanding of both the value of open source and technology lifecycles”.

Where the Coordination Tax Lands

Mikael Hailu of the Norwegian Red Cross put the scale problem plainly in August 2025. Most Global South open-source tools, he said, are “mission-specific, tightly bound to domains like public health, education, or social services”, with far narrower user bases than the Apache or Linux projects the sector loves to cite.

Foundation-backed open source works for Kubernetes because Google, IBM and Red Hat assign full-time engineers to it. Nobody assigns a full-time engineer to an mHealth chatbot running in Malawi.

The money behind the biggest projects makes the gap visible.

• $300 million: what the Linux Foundation collected last year, mostly from corporate sponsors
• $5.8 million: grants its Alpha-Omega security program handed to just 14 open-source projects in 2025
• Two volunteers: the number who kept Ingress-Nginx running before its retirement
• Over 80%: the share of organizations that call open source valuable to their future, even as the maintenance strain grows

Those numbers describe Kubernetes-tier infrastructure. A five-person NGO IT team supporting a refugee chatbot has none of that backing, and no invoice ever arrives to prove the gap exists.

What NGO Procurement Teams Do with This Now

None of this argues for abandoning open source. It argues for budgeting its coordination layer the way a license fee already gets budgeted, as a line item with a named funder attached.

A sustainability plan that only promises the code will be open source is not a plan. It has to name who pays for maintenance, hosting, security patching and contributor time for the next five years, in dollars.

One narrow option exists this week. Zendesk’s $50,000 impact award closing Wednesday is open to nonprofits right now, and it’s exactly the kind of one-time grant that could fund the maintenance contract most donor budgets skip.

Principle 6 of the sector’s own guidelines still calls for open licensing and open-source software wherever it fits, part of nine living guidelines for digital development that donors write into grant conditions. Whether a tool still exists in 2030 will depend on who is named to pay for keeping it alive, not on which license sits on top of it.

Frequently Asked Questions

Does NetHope’s Report Mean NGOs Should Stop Using Open Source?

No. The same briefing that raised the coordination-cost warning also holds up fAIr and POMELO as projects worth copying for publishing their code on GitHub. fAIr’s value comes specifically from community-generated training data that corrects Global North bias baked into most mapping AI models, the kind of public good the sector should keep building in the open.

What Are the Principles for Digital Development?

They are a set of guidelines, including Principle 6’s endorsement of open licensing and open-source software, that donors and implementers use to judge digital development projects. The community-driven effort traces back to the late 2000s, when donors and implementers realized their digital programs were fragmented and rarely survived past their original funding.

How Many People Kept Ingress-Nginx Running Before It Was Retired?

Just two, according to the project’s own retirement notice, for a tool that an estimated half of cloud-native environments rely on directly. The Cloud Native Computing Foundation (CNCF) offered only best-effort maintenance starting in November 2025 and cut support entirely by March 2026, ending releases, bug fixes and security patches.

What Is Total Cost of Ownership in Software Procurement?

Total cost of ownership, or TCO, adds every cost of running software beyond its sticker price, including maintenance, staff time, training and security patching. One widely cited estimate puts commercial software at 20 to 30 bugs for every 1,000 lines of code, a baseline researchers use to model how much unpaid bug-fixing time an open-source alternative can demand without a vendor’s support contract behind it.

Can Underfunded Servers Undo an Open-Source Success Story?

Yes. Development Gateway has pointed to a system once celebrated internationally that went dark because the responsible government never budgeted for ongoing server costs, not because the code failed. The tool kept working right up until the hosting bill went unpaid, a gap a commercial contract would have flagged from day one.