Six Dollars a Net, Forever, Versus One Fixed Bet

SEPTEMBER 17, 2026

Malaria is the rare problem where the cost of a solved unit is genuinely known. Not estimated, not modelled off a theory — measured, repeatedly, by people who distribute the things and then go count who is still alive. An insecticide-treated bed net costs about $6 to buy and hang. Run enough of those through the statistics and you get the number GiveWell used for its 2022–24 grants to the Against Malaria Foundation: roughly $5,500 per life saved. Their broader range across top charities has typically landed between $3,000 and $5,500.

That is one of the most remarkable price tags in economics, and it comes with an asterisk nobody says out loud often enough: the nets wear out in about three years. Then you buy them again. For everybody. Forever.

Malaria Prevention Is a Subscription Business

Look at the shape of the spend rather than the size of it. In 2024 there were an estimated 282 million malaria cases and about 610,000 deaths — 94% of cases and 95% of deaths in Africa, and 75% of the deaths among children under five. Global malaria funding that year was $3.9 billion against a $9.3 billion target, which is 42% of what the plan called for.

Notice that the funding gap is structural, not a bad year. Every tool in the working toolkit — nets, indoor spraying, seasonal chemoprevention at about $7 per child per season, diagnosis, treatment — is a per-person, per-period cost. Double the population at risk and you double the bill. Keep the population flat and you still pay it again next cycle. The 2.3 billion cases and 14 million deaths averted since 2000 are real and enormous, and they were bought on a payment plan that never ends.

Meanwhile the unit economics are quietly deteriorating in the way subscription businesses hate most: the product is losing potency. The World Malaria Report's headline worry in 2025 was drug resistance, sitting alongside years of insecticide resistance. You are paying the same $6 for a net that does somewhat less than the $6 net did fifteen years ago, against a parasite that has been in the arms race longer than we have.

A Gene Drive Is a Fixed Cost With a Spreading Benefit

I wrote up the mechanism separately — how a gene drive works, and the two rival schools of thought about it — but the financial structure is the part that belongs here, and it is genuinely unusual.

A gene drive spreads itself. That is the entire engineering point: an edit that copies itself onto the other chromosome so that nearly all offspring inherit it, instead of the half that ordinary inheritance would give you. Which means that after the research is paid for and a release is permitted, the marginal cost of protecting the next village is not $6 a head. It is approximately zero, paid in mosquito reproduction, on a schedule set by mosquito generations rather than by a donor conference.

In accounting terms: nets are cost of goods sold, and a drive is capitalised R&D with no meaningful COGS behind it. Those two things do not belong in the same column of anybody's spreadsheet, which is exactly why comparing them on dollars-per-life-saved produces nonsense in both directions.

And the research bill, so far, is startlingly small relative to the problem. The Open Philanthropy grant behind the University of California Malaria Initiative's island field programme in São Tomé and Príncipe is $10.2 million. Against $3.9 billion of annual global malaria spending, that is about a quarter of one percent — call it a rounding error with a laboratory attached. UC Irvine's earlier environmental-malaria programme was up to $9.6 million over seven years. These are numbers that would not clear the approval threshold for a mid-sized office refit at a large bank.

So Why Isn't Everyone Funding It?

Because the expected value has a term in it that nobody can price, and I want to be honest about that rather than sell you the optimistic version.

A $6 net has a near-certain, small, temporary effect. A gene drive has an uncertain, enormous, permanent one — and "uncertain" here is doing a lot of work across at least three independent failure modes:

It might not hold. The copy step fails sometimes: the cell repairs the cut by rejoining the ends, which mutates the target site into something the guide RNA can no longer find. That produces a drive-resistant allele, and selection is happy to spread it. The published literature on these strains includes work specifically on resistance alleles, which tells you the researchers rate the risk about as highly as the critics do.

It might not be permitted. This is the one the financial case consistently underweights, and 2025 settled the argument. On 11 August 2025, Target Malaria — the best-funded, furthest-advanced programme in the field, roughly a decade into community engagement in Burkina Faso — released about 16,000 modified male mosquitoes in the village of Souroukoudindan. Eleven days later the government suspended the project by ministerial directive. Researchers killed the remaining insectary stock; a government team sprayed insecticide on the release site. It was still suspended into 2026, with knock-on doubt across other African programmes.

Eleven days. A decade of groundwork and a consent process that was, by any normal standard, exhaustive — and the binding constraint turned out not to be the biology or the money at all. Any expected-value calculation that omits a political-termination probability is not a calculation, it is a brochure.

And it might work and still not be the cheapest thing you could have done with the same ten million, because counterfactual reasoning at this altitude is genuinely hard.

What I Actually Think the Arithmetic Says

Two things, and the second one is the point.

First: they are not competitors, and the framing that pits them against each other is the mistake. Nets save lives this year at a known price. Drives might remove the need to buy nets at all, at an unknown price, on an unknown date, with a real chance of never. A portfolio holds both, for the same reason a sane balance sheet holds both operating expense and research — and if you are choosing where a personal donation goes, the known-price option has the stronger claim precisely because it is known. That is not a knock on the science; it is what the word "uncertain" costs.

Second, and this is what struck me: the asymmetry in the numbers is enormous and nobody is arbitraging it. Humanity spends $3.9 billion a year renting a solution. The entire serious attempt to buy one outright, across the UC programme's field sites, runs on eight figures. Whatever you believe about gene drives — and there are respectable reasons to be nervous — a ratio like 400-to-1 between the annual rent and the total purchase attempt is not a considered allocation of capital. It is what a market looks like when nobody has repriced the asset.

Where I Could Be Wrong

The $10.2 million is one grant to one programme's field work, not the total global spend on malaria gene drives — Target Malaria, Transmission Zero and others have their own budgets I have not added up, so my 400-to-1 is illustrative of the gap in order of magnitude, not a carefully constructed total-R&D-to-total-operations ratio. Read it as the shape of the disparity, not a precise multiple.

The "approximately zero marginal cost" claim is the strongest thing I said and the most vulnerable. It holds for the spread of the edit itself. It does not include surveillance, regulatory compliance, the monitoring programme that would have to run for years, or the rebuild if resistance beats the drive and a second-generation construct is needed. A real cost-per-life-saved figure for a working drive does not exist yet because no drive has been released into a wild population, and I would distrust anyone who quotes one.

I am also aware that the tidy "rent versus buy" framing is doing rhetorical work, and that the people who object to gene drives are not primarily objecting on cost grounds. An irreversible ecological intervention with a global externality profile is not made acceptable by being cheap, and I have not engaged that argument here — it deserves its own treatment rather than a paragraph in a financial piece.

Nothing here is advice about where to give money. It is one reader's arithmetic, shown so you can check it.

Sources

1. GiveWell. How Much Does It Cost to Save a Life? — the $5,500-per-life-saved figure for 2022–24 AMF grants and the $3,000–$5,500 top-charity range. givewell.org
2. GiveWell. Our Top Charities — approximate $6 cost per insecticide-treated net, ~$7 per child per season for chemoprevention. givewell.org
3. World Health Organization. World Malaria Report 2025 — 2024 estimates: 282 million cases, ~610,000 deaths, $3.9bn funding against a $9.3bn target. who.int
4. UN News. Malaria: Drug resistance and underfunding threaten progress towards eliminating killer disease. December 2025. news.un.org
5. UC ANR. UC Davis $10.2 Million Grant Targets Malaria in São Tomé and Príncipe. ucanr.edu
6. Science. After 'humiliating' raid, Burkina Faso halts 'gene drive' project to fight malaria. 2025. science.org
7. Nature Africa. Mosquito gene drive cancellation disrupts Africa's malaria research. 2025. nature.com
8. Carballar-Lejarazú, R. et al. Dual effector population modification gene-drive strains of the African malaria mosquitoes. PNAS, 2023. pnas.org