Mycelium Insulation for Ugandan Homes: A Biomanufacturing Case Study
Corrugated iron has replaced thatch across most of Uganda. It is cheap, it is durable, and it does not rot. It is also thermally brutal. By mid-afternoon a bare metal sheet under equatorial sun turns the room beneath it into an oven, and when the rain comes, the noise is loud enough to stop a conversation.
This is a write-up of a service project exploring whether mycelium insulation, grown by the households that would use it from agricultural waste they already have, could fix both problems at once. It is a concept, not a product. Nothing here has been built or measured yet, and I will be specific further down about what would have to be tested before any of it is real.
The problem is not what North American insulation solves
Most of populated Uganda sits between 1,100 and 1,300 metres of elevation. Kampala runs roughly 17 °C at night and 27 °C during the day, year round. Almost nobody heats or cools a home, so R-value in the conventional sense is close to irrelevant.
What residents actually complain about is narrower and more specific. Radiant heat from the roof. Rain noise. Both are roof problems rather than wall problems, and both are addressed by the same intervention: a ceiling layer hung underneath the iron sheeting.
Targeting the ceiling rather than the walls also avoids a trap. A wall assembly that takes on ground moisture fails, and designing around that would have made the whole thing much harder. A ceiling panel carries only its own weight, stays dry, and needs no structural capacity at all.
Why mycelium insulation, and how the panels grow
The organism is Pleurotus ostreatus, the common oyster mushroom. Mycelium is the root-like network of thread-like filaments, called hyphae, that a fungus sends through whatever it is feeding on. Grown through loose agricultural waste, that network binds the pile into a single solid block. Dry the block in the sun and you have a rigid, lightweight panel.
I chose oyster over a purpose-engineered organism for three practical reasons. It is aggressive enough to outcompete contaminants without sterile technique, which matters when there is no autoclave. It grows on almost anything containing lignin and cellulose. And it is already cultivated in Uganda, so some of the knowledge and supply chain exists.
The inputs are free. Banana pseudostem, coffee husks, maize stover, rice husks, sugarcane bagasse. Uganda produces all of these as waste, and banana is the dominant staple crop, so pseudostem is available year round.
The process is solid-state fermentation, which simply means the fungus grows on damp solid material rather than in a liquid tank. Chop the substrate, wet it to about 65 percent moisture, pasteurise it by steeping in hot water or heating it in black plastic under the sun, mix in the starter culture, and pack it into a mould. Three to four weeks later the loose pile has become a continuous solid.
The panel is a byproduct of a food crop
This is the part that makes the economics work, and it was the single most useful idea to come out of the project.
Selling a novel building product into a cash-constrained market is hard. Extending a livelihood activity that already exists is much easier. So the sequence is deliberately food first: the household harvests two or three flushes of edible oyster mushrooms over about four weeks, then dries the spent block, which is normally thrown away, and hangs it as a ceiling panel.
At typical yields, roughly 300 kg of dry substrate produces somewhere near 180 kg of fresh mushrooms before it becomes the panel. The mushrooms pay for the starter culture. The insulation is what is left over.
Ship information, not material
The obvious version of this business makes starter culture in the United States and ships it to Uganda. That version fails on arithmetic.
Grain spawn, meaning the starter culture grown on sterilised grain, is heavy and alive. A full roof needs roughly 43 kg of it. Air-freighting that much perishable biological material per household, through tropical transit and customs delays, with no refrigeration, is not viable.
The fix is to let the biology do the scaling. Each local amplification step multiplies the mass by roughly fifteen times. Two generations on the ground turn about 190 grams of mother culture into the 43 kg a house needs. That 190 grams is a single 10 mL syringe, weighing under 20 grams shipped, which eventually becomes around 240 kg of finished ceiling panel.
What crosses the border is a shoebox. Everything after that is grown locally. For a 200-household village, the entire import is roughly 2.5 kg of culture, amplified through three generations into the 8.6 tonnes the village needs.
There is a strategic point buried in that. The long-term version of this does not have a US supplier at all. It has a local spawn operation, and the imported culture is just a way to bootstrap it.
Where automation belongs, and where it does not
Only one step in this process has a real throughput problem, and it is strain selection. Screening dozens to hundreds of fungal isolates against several Ugandan substrates, at Ugandan ambient temperatures rather than lab bench conditions, is exactly the kind of repetitive liquid-handling work a robot should do.
The rest of it should stay manual by design. No bioreactors, no controlled environment rooms, no power dependency. Colonisation happens in shallow ground pits, which hold a stable cool temperature for free, and in household-scale bags and moulds. Drying is solar.
A centralised automated facility would introduce capital requirements and a distribution problem that the distributed version simply does not have. Automation belongs at the front, in the lab, where it buys something. On the ground, the scale-up mechanism is the organism’s own exponential growth running in parallel across every household at once.
How the process works, end to end
The project splits into two halves that look nothing alike, and that asymmetry is the whole design.
Half one: optimize the strain in the cloud
The first half happens in a cloud lab, meaning a remotely operated facility where you write the experiment as code and robots execute it, returning structured data rather than a lab notebook. This is how the course I developed the project in runs its experimental work, and it fits this problem well.
The question the cloud lab answers is which fungal strain to send. Not any oyster mushroom strain will do. The one that ships needs to be:
- Fast to colonise, because every extra day of growth is another day contamination can take hold
- Thermotolerant, since it will grow at 25 to 30 °C with no climate control rather than at a comfortable lab bench temperature
- Broad in its substrate range, because a household in a coffee-growing district has coffee husk and a household in a banana-growing district has pseudostem, and the strain has to handle whatever is at hand
- Robust in transit, surviving ten to fourteen days without refrigeration and arriving viable
Screening isolates against those variables is repetitive, parallel, plate-based work. It is exactly what automated liquid handling and time-lapse plate imaging are for. Dozens of candidates, several substrates, several temperatures, measured as millimetres of growth per day rather than judged by eye.
One honest boundary. A cloud lab is built for liquid handling and plate assays, so it can screen growth rate, temperature tolerance, and substrate preference on defined media. It cannot run a termite feeding trial, and it cannot tell you how a strain behaves on a real pile of chopped banana stem in a Ugandan pit. Those tests have to happen on the ground. The cloud lab narrows a hundred candidates to a handful; field trials pick the winner.
Half two: amplify the winner in Uganda
What ships at the end of the first half is tiny. A single 10 mL syringe of liquid culture, about 190 grams, under 20 grams of shipping weight.
Everything after that happens locally, with no equipment. The culture is used to inoculate sterilised grain, producing the first generation of spawn. That generation inoculates a larger batch, producing the second. Each step multiplies the mass roughly fifteen times, so two generations turn the syringe into the 43 kg a household needs, and three generations serve an entire village.
The spawn then goes into the agricultural waste, the fungus grows through it, the household harvests mushrooms, and the spent block becomes the panel.
So the finished product weighs something like 1,200 times what crossed the border. The intelligence is imported. The mass is grown where it is used.
Why this split matters
This is the shape the project kept returning to: US biodesign capability paired with African biomass.
The United States has cloud labs, automated screening, and strain development infrastructure that most of the world cannot access. Uganda has enormous volumes of agricultural waste, year-round growing temperatures, an existing mushroom cultivation tradition, and abundant labour. Neither side can do this alone. A US lab cannot manufacture 400 cubic metres of panel, and a Ugandan household cannot screen two hundred fungal isolates.
Splitting the work along that line means the expensive, capital-intensive part stays small and stays where the infrastructure already exists, while the heavy, high-volume part happens where the raw material is free and the need is real. The thing being exported is not a product, and not aid. It is a design.
Why this sits under Service Projects
This came out of coursework for How to Grow (Almost) Anything, the MIT Media Lab synthetic biology programme, and it stayed with me because I know people in Uganda who live under iron roofs.
It is also a reasonable illustration of how I work. The interesting move in this project was not the biology, which is well established. It was noticing that the shipping arithmetic killed the obvious business model, and that letting the organism do the amplification on the ground solved it. That kind of constraint-first thinking is most of what useful technical strategy consists of.
If you are working on something in biotech or drug development where the hard part is figuring out which version of the idea survives contact with reality, that is the work I do at Scriptome.AI.