I just learned about Zipline, the world’s largest autonomous drone delivery system, from YouTube tech reviewer Marques Brownlee’s recent video, so I was surprised to see Zipline pop up in a GiveWell grant writeup of all places. I admittedly had the intuition that if you’re optimising for cost-effectiveness as hard as GW do, and that your prior is as skeptical as theirs is, then the “coolness factor” would’ve been stripped clean off whatever interventions pass the bar, and Brownlee’s demo both blew my mind with its coolness (he placed an order on mobile for a power bank and it arrived by air in thirty seconds flat, yeesh) and also seemed the complete opposite of cost-effective (caveating that I know nothing about drone delivery economics). Quoting their “in a nutshell” section:
In December 2024, GiveWell recommended a $54,620 grant to Zipline for a six-month scoping project. Zipline will use this time to review ways that they could use drones to increase vaccination uptake, especially in hard-to-reach areas with low vaccination coverage and high rates of vaccine-preventable diseases. …
We recommended this grant because:
Drones are an intuitive way to bring vaccines closer to communities with low coverage rates, especially when demand for vaccination exists, but conditions like difficult terrain, poor infrastructure, weak cold chain, or insecurity make it difficult for families to access immunizations.
This grant aligns with our strategy of making several scoping grants to high-potential organizations to source promising ideas for solving bottlenecks in the routine immunization system, and then testing these concepts.
Okay, but what about cost-effectiveness? Their “main reservations” section says
Evidence on cost-effectiveness of drones for vaccine delivery is limited, and we have not modeled the cost-effectiveness of the types of programs that Zipline plans to consider, nor the value of information for this scoping grant.
An internal review conducted in 2023 focused on drones for health was generally skeptical about there being many opportunities in this area that would meet GiveWell’s bar, although this scoping grant will focus on the most promising scenario (remote areas with high rates of vaccine-preventable diseases and low vaccination coverage rates).
Is there any evidence of cost-effectiveness at all then? According to Zipline, yes — e.g. quoting the abstract from their own 2025 modelling study:
Objectives: In mid-2020, the Ghana Health Service introduced Zipline’s aerial logistics (centralized storage and delivery by drones) in the Western North Region to enhance health supply chain resilience. This intervention led to improved vaccination coverage in high-utilization districts. This study assessed the cost-effectiveness of aerial logistics as an intervention to improve immunization coverage.
Methods: An attack rate model, adjusted for vaccination coverage and vaccine efficacy, was used to estimate disease incidence among vaccinated and unvaccinated populations, focusing on 17 022 infants. Incremental cost-effectiveness ratios of US dollar per averted disability-adjusted life-year (DALY) were evaluated from societal and government perspectives, using real-world operations data. …
Results: In 2021, aerial logistics averted 688 disease cases. Incremental cost-effectiveness ratios were $41 and $58 per averted DALY from the societal and government perspectives, respectively. The intervention was cost-saving when at least 20% of vaccines delivered by aerial logistics replaced those that would have been delivered by ground transportation, with potential government savings of up to $250 per averted DALY. Probabilistic sensitivity analysis confirmed the robustness of these findings.
That’s super cost-effective. For context, the standard willingness-to-pay to avert a DALY is 1x per capita GDP or $2,100 in Ghana, so 35-50x higher. Also:
… we calculated that aerial logistics facilitated the completion of an additional 14 979 full immunization courses… We estimated that 4 children’s lives (95% CI 2–7) were saved in these districts during 2021. … the intervention averted a total of $20 324 in treatment costs and $2819 for caregivers between lost wages and transport.
At a cost of $0.66 per incremental FIC (fully immunized child),this approach outperforms other delivery methods analyzed in the review, including the most cost-effective category of interventions identified, namely “Delivery Approach” interventions, such as monthly immunization by mobile teams in villages and the enhancement of satellite clinic immunization practices.
(GW notes that they’d given Zipline’s study a look and “were unable to quickly assess how key parameters like program costs and the impact of the program on vaccination uptake and disease were being estimated”. Neither can I. Still pretty exciting)
Zipline have been around for about 10 years I think—boy do they have the cool factor. One big issue is that they can only carry as really tiny amount of stuff. Also the places where they can potentially save money have to be super hard to access, because a dirt cheap motorcycle which can go 50km for a dollar of fuel can carry 50x as much weight.
My lukewarm take is that hey have done well, but as with most things haven’t quite lived up to their initial hype.
Interesting, I got the opposite impression from their about page (“4,000+ hospitals and health centers served, 51% fewer deaths from postpartum hemorrhaging in hospitals Zipline serves, 96% of providers report increased access to vaccinations in their area” which I assume means they’re already targeting those hard-to-access areas), but of course they’d want to paint themselves in a good light and I’d be inclined to trust your in the field experience far more (plus general skepticism just being a sensible starting point).
Actually your point about a cheap bike being able to carry a lot more stuff makes obvious sense, and so me wonder how Zipline’s modelling study in Ghana can claim that their cost per incremental fully immunised child was cheaper than “monthly immunization by mobile teams” which I assume includes dirt bikes.
Don’t be inclined to trust my in-the-field experience, Zipline has plenty of that too!
I just had a read of their study but couldn’t see how they calculated costing (the most important thing).
One thing to note is that vaccine supply chains currently often unnecessarily use trucks and cars rather than motorcycles because, well, GAVI has funded them so they may well be fairly comparing to status quo rather than other more efficient methods. For the life of me I don’t know why so many NGOs use cars for si many things that public transport and motorcycles could do sometimes orders of magnitude cheaper. Comparing to status quo is a fair enough thing to do (probably what I would do) but might not be investigating the most cost effective way of doing things.
Also I doubt they are including R and D and the real drone costs in the costs in of that study, but I’ll try and dig and get more detail.
It annoys me that most modeling studies focus so hard on their math method, rather than explaining now about how they estimate their cost input data—which is really what defines the model itself.
The modelling study has a “costs” section (quoted below), but for what it’s worth GiveWell said they “were unable to quickly assess how key parameters like program costs… were being estimated” so I don’t think this quote will satisfy you:
Given the Ghana Health Service (GHS)‘s dominant role, the government perspective in this analysis included healthcare treatment costs and incremental last mile delivery (LMD) costs. The societal perspective also accounted for externalities such as caregivers’ wage loss and transport costs.
To calculate the total cost for aerial LMD of vaccines, we analyzed Zipline’s monthly operational costs and the depreciation of capital expenditures for the GH4 distribution center in the Western North Region. These were adjusted to 2023 US dollar values, and the corresponding portion attributed to vaccine delivery was determined, resulting in a cost per dose of $0.27.
To estimate the incremental cost of the intervention, we took into account that the impact of aerial logistics on vaccination rates can be explained through either a pure expansion of access (ie, health facilities receiving vaccine doses that they otherwise would not have) or more efficient access (ie, health facilities receiving the same number of vaccine doses they would have otherwise received but in a more timely manner, leading to fewer missed opportunities of vaccination). Anecdotal evidence suggests that the impact is likely a combination of both factors. The distinction is significant when computing costs in an ICER: in the former, aerial logistics LMD cost is an additional expense to the existing supply chain cost for the government, whereas, in the latter, aerial logistics LMD replaces the traditional supply chain cost for transporting those vaccines. …
Due to the absence of detailed data on traditional LMD, we were unable to differentiate between incremental and replaced doses within the number of doses delivered with aerial logistics during the intervention period. To mitigate the impact of this uncertainty on our estimations, for our primary ICER calculation, we proceeded with the conservative assumption that all doses delivered by aerial logistics during this period were incremental. This approach may inflate our incremental cost estimates but ensures the solidity of our findings amid the well-known ambiguous quality and high variance of the traditional LMD data that were used for illustrative purposes in the sensitivity analysis.
But no input numbers, just methods and a dash of conservatism.
I share your annoyance re: modelling studies. Garbage in garbage out as they say (not accusing Zipline of putting garbage data into their model of course!)
Re: NGOs using trucks and cars unnecessarily, I’m just speculating here but I wonder if it’s got a bit to do with the NGOs wanting to attract “top talent” (salary difference being the main attractor but also “you get to ride in a car instead of on a bike” being implicitly part of the “comp package”, sort of like how top talent in higher-income countries are lured to prestigious industries by not just pay but “comped stays in nice hotels” or whatever). This paper I read awhile back made me think of that: The unintended consequences of NGO-provided aid on government services in Uganda. It argues that NGOs sometimes “poach” scarce local skilled government workers via higher pay, resulting in various adverse effects, although I guess it’s a bit different in this case because the adverse effects happen as a result of the pay structure (NGO workers who would’ve otherwise distributed health products instead sell household products like soap and fortified oil because they get paid on a per-piece basis).
Can confirm; Zipline is ridiculously cool. I saw their P1 Drones in action in Ghana and met some of their staff at EA conferences. Imo, Zipline is one of the most important organizations around for last-mile delivery infrastructure. They’re a key partner for GAVI, and they save lives unbelievably cost-effectively by transporting commodities like snakebite antivenom within minutes to people who need it.
Their staff and operations are among the most high-performing of any organization I’ve ever seen. Here are some pics from a visit I took to their bay area office in October 2024. I’d highly recommend this Mark Rober video, and checking out Zipline’s website. If any software engineers are interested in high-impact work, I would encourage you to apply to Zipline!
Thanks for the links! And for the pics, makes me feel like I’m glimpsing the future, but it’s already here, just unevenly distributed. Everything you say jives with both what GiveWell said about Zipline in their grant writeup
Though this is our first engagement with Zipline, there are some early signals that we might be well-aligned as partners. Zipline’s proposal specifically calls out a few elements that are important to GiveWell: a) emphasis on cost-effectiveness , b) plans to establish an M&E framework early on for any potential pilots, and c) interest in evaluation and learning
Yep Snakebite is one of the few slamdunk usecases for me here. Until we design a cheap, heat stable antivenom I think drones that can get there in under an hour might be the best option in quite a wide range of places.
Nice! I’ve been enjoying your quick takes / analyses, and find your writing style clear/easy to follow. Thanks Mo! (I think this could have been a great top level post FWIW, but to each their own :) )
That’s really kind of you Angelina :) I think top-level posting makes me feel like I need to put in a lot of work to pass some imagined quality bar, while quick takes feel more “free and easy”? Also I hesitate to call any of my takes “analyses”, they’re more like “here’s a surprising thing I just learned, what do y’all think?”
I just learned about Zipline, the world’s largest autonomous drone delivery system, from YouTube tech reviewer Marques Brownlee’s recent video, so I was surprised to see Zipline pop up in a GiveWell grant writeup of all places. I admittedly had the intuition that if you’re optimising for cost-effectiveness as hard as GW do, and that your prior is as skeptical as theirs is, then the “coolness factor” would’ve been stripped clean off whatever interventions pass the bar, and Brownlee’s demo both blew my mind with its coolness (he placed an order on mobile for a power bank and it arrived by air in thirty seconds flat, yeesh) and also seemed the complete opposite of cost-effective (caveating that I know nothing about drone delivery economics). Quoting their “in a nutshell” section:
Okay, but what about cost-effectiveness? Their “main reservations” section says
Is there any evidence of cost-effectiveness at all then? According to Zipline, yes — e.g. quoting the abstract from their own 2025 modelling study:
That’s super cost-effective. For context, the standard willingness-to-pay to avert a DALY is 1x per capita GDP or $2,100 in Ghana, so 35-50x higher. Also:
(GW notes that they’d given Zipline’s study a look and “were unable to quickly assess how key parameters like program costs and the impact of the program on vaccination uptake and disease were being estimated”. Neither can I. Still pretty exciting)
Zipline have been around for about 10 years I think—boy do they have the cool factor. One big issue is that they can only carry as really tiny amount of stuff. Also the places where they can potentially save money have to be super hard to access, because a dirt cheap motorcycle which can go 50km for a dollar of fuel can carry 50x as much weight.
My lukewarm take is that hey have done well, but as with most things haven’t quite lived up to their initial hype.
Interesting, I got the opposite impression from their about page (“4,000+ hospitals and health centers served, 51% fewer deaths from postpartum hemorrhaging in hospitals Zipline serves, 96% of providers report increased access to vaccinations in their area” which I assume means they’re already targeting those hard-to-access areas), but of course they’d want to paint themselves in a good light and I’d be inclined to trust your in the field experience far more (plus general skepticism just being a sensible starting point).
Actually your point about a cheap bike being able to carry a lot more stuff makes obvious sense, and so me wonder how Zipline’s modelling study in Ghana can claim that their cost per incremental fully immunised child was cheaper than “monthly immunization by mobile teams” which I assume includes dirt bikes.
Don’t be inclined to trust my in-the-field experience, Zipline has plenty of that too!
I just had a read of their study but couldn’t see how they calculated costing (the most important thing).
One thing to note is that vaccine supply chains currently often unnecessarily use trucks and cars rather than motorcycles because, well, GAVI has funded them so they may well be fairly comparing to status quo rather than other more efficient methods. For the life of me I don’t know why so many NGOs use cars for si many things that public transport and motorcycles could do sometimes orders of magnitude cheaper. Comparing to status quo is a fair enough thing to do (probably what I would do) but might not be investigating the most cost effective way of doing things.
Also I doubt they are including R and D and the real drone costs in the costs in of that study, but I’ll try and dig and get more detail.
It annoys me that most modeling studies focus so hard on their math method, rather than explaining now about how they estimate their cost input data—which is really what defines the model itself.
The modelling study has a “costs” section (quoted below), but for what it’s worth GiveWell said they “were unable to quickly assess how key parameters like program costs… were being estimated” so I don’t think this quote will satisfy you:
But no input numbers, just methods and a dash of conservatism.
I share your annoyance re: modelling studies. Garbage in garbage out as they say (not accusing Zipline of putting garbage data into their model of course!)
Re: NGOs using trucks and cars unnecessarily, I’m just speculating here but I wonder if it’s got a bit to do with the NGOs wanting to attract “top talent” (salary difference being the main attractor but also “you get to ride in a car instead of on a bike” being implicitly part of the “comp package”, sort of like how top talent in higher-income countries are lured to prestigious industries by not just pay but “comped stays in nice hotels” or whatever). This paper I read awhile back made me think of that: The unintended consequences of NGO-provided aid on government services in Uganda. It argues that NGOs sometimes “poach” scarce local skilled government workers via higher pay, resulting in various adverse effects, although I guess it’s a bit different in this case because the adverse effects happen as a result of the pay structure (NGO workers who would’ve otherwise distributed health products instead sell household products like soap and fortified oil because they get paid on a per-piece basis).
Can confirm; Zipline is ridiculously cool. I saw their P1 Drones in action in Ghana and met some of their staff at EA conferences. Imo, Zipline is one of the most important organizations around for last-mile delivery infrastructure. They’re a key partner for GAVI, and they save lives unbelievably cost-effectively by transporting commodities like snakebite antivenom within minutes to people who need it.
Their staff and operations are among the most high-performing of any organization I’ve ever seen. Here are some pics from a visit I took to their bay area office in October 2024. I’d highly recommend this Mark Rober video, and checking out Zipline’s website. If any software engineers are interested in high-impact work, I would encourage you to apply to Zipline!
Thanks for the links! And for the pics, makes me feel like I’m glimpsing the future, but it’s already here, just unevenly distributed. Everything you say jives with both what GiveWell said about Zipline in their grant writeup
as well as the vibe I get from their about page, stuff like
and 3 out of their 4 most prominent “output statistic” claims being health-oriented
Yeah the pointer to snakebite antivenom delivery feels useful, you reminded me of how big a problem it is.
Yep Snakebite is one of the few slamdunk usecases for me here. Until we design a cheap, heat stable antivenom I think drones that can get there in under an hour might be the best option in quite a wide range of places.
Nice! I’ve been enjoying your quick takes / analyses, and find your writing style clear/easy to follow. Thanks Mo! (I think this could have been a great top level post FWIW, but to each their own :) )
That’s really kind of you Angelina :) I think top-level posting makes me feel like I need to put in a lot of work to pass some imagined quality bar, while quick takes feel more “free and easy”? Also I hesitate to call any of my takes “analyses”, they’re more like “here’s a surprising thing I just learned, what do y’all think?”