Bioconversion using insects

Denis

Short answer: I don’t know of any study that weighs insect welfare against the environmental benefits of waste bioconversion in a single framework. But I’m not sure one is needed here. The two promises, producing protein and treating waste, pull against each other. Protein that is safe and sellable needs clean, consistent inputs, which aren’t really waste. Real waste brings contaminants and variability, and in the EU much of it can’t legally be fed to farmed insects at all. Operators end up trading one goal against the other, and rarely deliver both. And whichever goal you pick, there is usually an established alternative that does at least as well environmentally and costs less: anaerobic digestion or composting to treat waste, and for feed, soybean meal or simply giving clean side-streams to livestock directly. So the welfare concern isn’t up against a clear sustainability gain. It comes on top of a case that is already weak.

Some context on where I’m coming from: I’m the scientific director of ONEI (Observatoire National de l’Élevage d’Insectes), a French non-profit working on the impacts of insect farming. My PhD, in economics, was on alternative proteins, and I’ve published several academic papers on insect farming. Disclosure: I co-authored one of the reviews cited below (Biteau et al., 2025).

I’m not an expert on sentience, so I’ll keep this part short. As far as I know, though, several serious lines of work suggest insects are credible candidates for sentience:

Jonathan Birch classes insects as “sentience candidates” in The Edge of Sentience , meaning the evidence is strong enough that ignoring the possibility in decisions that affect them would be irresponsible.
Gibbons et al. (2022), a review co-authored by Birch and Lars Chittka, assessed six insect orders against eight criteria. Adult flies met six. Fly larvae met three or four, which the authors still count as substantial evidence for pain. Black soldier flies are flies, but the assessment was made at the level of the order, and the authors point out that evidence on juveniles has large gaps.
The New York Declaration on Animal Consciousness (2024), now with more than 600 signatories, states that there is at least a “realistic possibility” of conscious experience in insects.

I’ll focus instead on the environmental and economic arguments, since those are the ones made for this route, and I have a lot of doubts about them.

The benchmark should be the other ways of treating the same waste: anaerobic digestion (AD), composting and, for clean streams, direct feeding to livestock. Not soy or fishmeal. Diverting waste from landfill and recovering nutrients are real benefits, but composting and AD deliver them too, so they don’t set insects apart. On that comparison no advantage has been established. Across the twelve assessments we reviewed, treating a tonne of biowaste with insects came out anywhere between −432 and +877 kg CO2e, and the favourable figures usually leave out collection, pre-treatment, transport or construction (Biteau et al., 2025). Some studies do favour insects, but usually against weak comparators (landfill, unaerated composting) or thanks to a large credit for the protein. Unlike AD the process recovers no energy, and in temperate climates the larvae have to be kept warm. Modelling in Nature Sustainability (Javourez et al., 2025) found that feed-grade streams do best fed directly to animals, and that turning residues into novel ingredients such as insect meal is competitive only under certain conditions (decarbonised energy, and replacing meat rather than animal feed). So “lower emissions than composting or AD” doesn’t hold as a general claim.
In Europe, most real waste is off limits. Farmed insects are farmed animals under EU law, so feed rules govern what they eat: catering waste, household biowaste, manure and former foodstuffs containing meat or fish are excluded. As I read the texts (I’m not a lawyer), calling the activity “waste management” doesn’t change that, and no insect process has been authorised as a waste treatment. The US is restrictive too, with feed-grade inputs only for black soldier fly larvae used in feed. Operators therefore end up on clean side-streams like brewers’ grains or unsold fruit and vegetables, much of which could go straight to livestock, which is where point 1 says the benefit is larger. Genuine waste is hard to handle anyway: Entosystem in Quebec runs incoming grocery waste through nine sorting machines before the larvae see it.
The protein doesn’t rescue the climate case. In the life-cycle assessment Ricardo did for Defra (2025), black soldier fly meal reared on food waste still had a higher climate impact per kg of protein than fishmeal, and several times that of soybean meal. The report concludes that insect protein “may not currently be the solution to decarbonising the animal feed industry”, while warning that results are very sensitive to methodological choices (with low-carbon electricity and a different allocation method, insect meal can come out ahead). This protein also goes into feed for poultry, fish or pets. It replaces soy or fishmeal, not meat, so it doesn’t reduce the number of farmed animals. It adds insects to the count.
Frass, the other output, is an immature fertiliser. Its composition isn’t standardised and large field trials are still scarce (same Defra/Ricardo work). It’s also hard to sell at a price that covers costs.
The business model depends on being paid to take the waste. FlyBox’s CEO says the economics “only really make sense when you’re aligned with waste and getting paid to take it” (AgNavigator, 2026). One model finds that for insect products to compete in Dutch broiler diets, the substrate would need a negative value of about −109 € per tonne of dry matter in the best case (Leipertz et al., 2024). Goterra, the best-known operator built on this model, went into administration in June 2026 and liquidation in July (Startup Daily). AgriProtein, Enterra and InSeco, all relying on waste streams, failed before it.
Escapes and contaminants. Mass-reared species are selected for traits that predict invasiveness, and escapes are frequently reported (Bang & Courchamp, 2021; Manfrini et al., 2026). Haplotypes typical of farmed black soldier flies have already been found in wild flies at four sites in France (Guilliet et al., 2022). On contaminants, cadmium bioaccumulates in the larvae, most other metals and pesticides end up in the frass, and pathogens can survive in both. Every fix, whether cleaner inputs, sterilising the substrate or heat-treating the frass, either removes the waste the model is sold on or degrades the products it depends on.

In your position, I would lead with the comparison on standard sustainability metrics, which the client will recognise, and then raise welfare as a further issue their analysis leaves out.

Questions I would put to the proposers:

What exactly is the substrate, and is it legal to feed it to insects where your client operates?
What is the comparator (landfill, unaerated composting, AD?), and is the functional unit a tonne of waste treated or a kg of protein?
Are collection, pre-treatment and heating included?
How much of the benefit comes from crediting the protein and frass?
What gate fee do they assume, and what happens to the waste if the operator fails?

For something you can forward, Ricardo has a short summary of its Defra work, and section IX of Biteau et al. (2025) is the part on waste. We also wrote a note on insects as a waste-treatment technology whose first page is a one-page summary. Since you can’t share the details of the project publicly, feel free to send me a private message, whether you’d like the note or want to talk through the options your client is comparing.

Tom Bry-Chevalier🔸

I’m not an expert on sentience, so I’ll keep this part short. As far as I know, though, several serious lines of work suggest insects are credible candidates for sentience:

Jonathan Birch classes insects as “sentience candidates” in The Edge of Sentience , meaning the evidence is strong enough that ignoring the possibility in decisions that affect them would be irresponsible.
Gibbons et al. (2022), a review co-authored by Birch and Lars Chittka, assessed six insect orders against eight criteria. Adult flies met six. Fly larvae met three or four, which the authors still count as substantial evidence for pain. Black soldier flies are flies, but the assessment was made at the level of the order, and the authors point out that evidence on juveniles has large gaps.
The New York Declaration on Animal Consciousness (2024), now with more than 600 signatories, states that there is at least a “realistic possibility” of conscious experience in insects.

I’ll focus instead on the environmental and economic arguments, since those are the ones made for this route, and I have a lot of doubts about them.

The benchmark should be the other ways of treating the same waste: anaerobic digestion (AD), composting and, for clean streams, direct feeding to livestock. Not soy or fishmeal. Diverting waste from landfill and recovering nutrients are real benefits, but composting and AD deliver them too, so they don’t set insects apart. On that comparison no advantage has been established. Across the twelve assessments we reviewed, treating a tonne of biowaste with insects came out anywhere between −432 and +877 kg CO2e, and the favourable figures usually leave out collection, pre-treatment, transport or construction (Biteau et al., 2025). Some studies do favour insects, but usually against weak comparators (landfill, unaerated composting) or thanks to a large credit for the protein. Unlike AD the process recovers no energy, and in temperate climates the larvae have to be kept warm. Modelling in Nature Sustainability (Javourez et al., 2025) found that feed-grade streams do best fed directly to animals, and that turning residues into novel ingredients such as insect meal is competitive only under certain conditions (decarbonised energy, and replacing meat rather than animal feed). So “lower emissions than composting or AD” doesn’t hold as a general claim.
In Europe, most real waste is off limits. Farmed insects are farmed animals under EU law, so feed rules govern what they eat: catering waste, household biowaste, manure and former foodstuffs containing meat or fish are excluded. As I read the texts (I’m not a lawyer), calling the activity “waste management” doesn’t change that, and no insect process has been authorised as a waste treatment. The US is restrictive too, with feed-grade inputs only for black soldier fly larvae used in feed. Operators therefore end up on clean side-streams like brewers’ grains or unsold fruit and vegetables, much of which could go straight to livestock, which is where point 1 says the benefit is larger. Genuine waste is hard to handle anyway: Entosystem in Quebec runs incoming grocery waste through nine sorting machines before the larvae see it.
The protein doesn’t rescue the climate case. In the life-cycle assessment Ricardo did for Defra (2025), black soldier fly meal reared on food waste still had a higher climate impact per kg of protein than fishmeal, and several times that of soybean meal. The report concludes that insect protein “may not currently be the solution to decarbonising the animal feed industry”, while warning that results are very sensitive to methodological choices (with low-carbon electricity and a different allocation method, insect meal can come out ahead). This protein also goes into feed for poultry, fish or pets. It replaces soy or fishmeal, not meat, so it doesn’t reduce the number of farmed animals. It adds insects to the count.
Frass, the other output, is an immature fertiliser. Its composition isn’t standardised and large field trials are still scarce (same Defra/Ricardo work). It’s also hard to sell at a price that covers costs.
The business model depends on being paid to take the waste. FlyBox’s CEO says the economics “only really make sense when you’re aligned with waste and getting paid to take it” (AgNavigator, 2026). One model finds that for insect products to compete in Dutch broiler diets, the substrate would need a negative value of about −109 € per tonne of dry matter in the best case (Leipertz et al., 2024). Goterra, the best-known operator built on this model, went into administration in June 2026 and liquidation in July (Startup Daily). AgriProtein, Enterra and InSeco, all relying on waste streams, failed before it.
Escapes and contaminants. Mass-reared species are selected for traits that predict invasiveness, and escapes are frequently reported (Bang & Courchamp, 2021; Manfrini et al., 2026). Haplotypes typical of farmed black soldier flies have already been found in wild flies at four sites in France (Guilliet et al., 2022). On contaminants, cadmium bioaccumulates in the larvae, most other metals and pesticides end up in the frass, and pathogens can survive in both. Every fix, whether cleaner inputs, sterilising the substrate or heat-treating the frass, either removes the waste the model is sold on or degrades the products it depends on.

In your position, I would lead with the comparison on standard sustainability metrics, which the client will recognise, and then raise welfare as a further issue their analysis leaves out.

Questions I would put to the proposers:

What exactly is the substrate, and is it legal to feed it to insects where your client operates?
What is the comparator (landfill, unaerated composting, AD?), and is the functional unit a tonne of waste treated or a kg of protein?
Are collection, pre-treatment and heating included?
How much of the benefit comes from crediting the protein and frass?
What gate fee do they assume, and what happens to the waste if the operator fails?