Script: The Cost Model Explained

Text of the audio overview — read, annotate, or suggest edits via Hypothes.is

Note

This is the text of the cost model audio overview linked from this site. It was generated by AI (OpenAI GPT-4 + TTS-1-HD, May 2026), based on the content of the model documentation, dashboard, and related materials. It has been reviewed by The Unjournal team and is generally accurate; however, it may state some things more definitively than we would ourselves, and it reflects the page content at a point in time.

To suggest corrections or flag inaccuracies: annotate this page via the Hypothes.is sidebar (click < on the right edge). To return to the main pages: Interactive Cost Model · Technical Documentation.

We built this model to answer a deceptively simple question: what would it actually cost to produce cultured chicken meat at commercial scale?

The reason this question is hard is not that the math is complicated — it isn’t. It’s that almost every number that goes into the calculation is genuinely uncertain. We don’t yet know what growth factors will cost in 2036. We don’t know which bioreactor process will dominate. We don’t know how dense cells will grow at scale, or how cheaply media can be made. These aren’t things we can look up — they depend on breakthroughs that may or may not happen.

So rather than pretending we know the answer, the model runs thirty thousand simulated scenarios. Each simulation draws a different combination of parameter values from realistic distributions, and the result is a full distribution of possible costs — not a single number, but a picture of the range of outcomes and what’s driving the uncertainty. That approach is called Monte Carlo simulation, and it’s the right tool for this kind of problem.

The output is expressed as manufacturing cost per kilogram of cultured chicken cell biomass at harvest — wet weight, at the factory gate, before any texturization or blending with other ingredients. That accounting boundary matters, because a lot of published cost figures refer to different things.

What goes into that cost?

We break it into four pieces.

The first is variable operating cost — everything that scales directly with each kilogram you produce. This is almost always the dominant component, and within it, two things dominate: cell culture media and growth factors.

Cell culture media is the liquid nutrient broth that feeds growing cells. The cost per liter varies enormously depending on the grade and formulation. But what matters for cost isn’t just the price per liter — it’s how many liters you consume per kilogram of output. And that depends on how dense the cells grow. Importantly, basal media cost and cell density aren’t fully independent: richer (more expensive) media formulations are often needed to sustain higher cell densities, particularly in nutrient-depleted fed-batch systems. The total media cost per kilogram reflects both effects.

There’s also the question of media composition. Expensive purified amino acids are a major cost driver. Hydrolysates — protein-breakdown products from yeast or plant sources — might substitute for them at a fraction of the cost.

Growth factors are the second component of variable cost, and they are the source of the model’s widest uncertainty band. Growth factors are small proteins — molecules like FGF-2 and IGF-1 — that signal cells to keep dividing. At research scale, they are extraordinarily expensive. At the volumes that commercial cultivated meat would require, no one currently produces them this way. The question is which lower-cost production route will succeed: precision fermentation, plant molecular farming, autocrine cell lines (which produce their own growth factors), or small-molecule substitutes. Any one working at commercial scale by 2036 would be transformative.

Capital costs

The second major component is annualized capital cost — the cost of building and financing the bioreactors and facility, spread across all kilograms produced.

Large-scale bioreactors are currently expensive, primarily because they’re built to pharmaceutical standards that cultivated meat doesn’t require. A food-grade bioreactor could potentially be built for a fraction of the pharmaceutical price. The capital cost per kilogram is also sensitive to scale — large plants spread fixed construction costs over more kilograms — and to financing cost (WACC), which depends on how risky investors perceive the sector.

Fixed costs and downstream

Fixed operating costs — labor, maintenance, overhead — don’t change batch to batch but shrink per kilogram as plant scale increases. Downstream processing adds cost for structured products; for unstructured products it may be minimal.

The maturity factor

Cutting across all four components is what we call the industry maturity variable. In a world where cultivated meat succeeds, many things tend to be true simultaneously: growth factor breakthroughs occur, financing costs fall, companies can afford custom food-grade equipment, and hydrolysate substitution is established. The model uses a single maturity draw to introduce realistic correlation, preventing incoherent scenarios. That’s a modelling convenience — different technologies could advance at different rates — and it’s one of the known structural limitations of the current model.

How to read the model

When you open the interactive dashboard, you see a cost distribution — a histogram of simulated outcomes. The median gives you the central estimate. The tail on the right tells you how bad things could get under pessimistic assumptions. The tornado chart shows which parameters move the median cost most when varied across their plausible range.

The most important thing about the model is what it tells you about uncertainty, not what it tells you about the central estimate. The shape of the distribution — whether tightly concentrated or spread across two orders of magnitude — is the signal.

What we’re asking experts

Rather than anchoring on any single set of parameter assumptions, we want to know what people who actually work in this space think about these distributions. The beliefs form asks workshop participants — TEA researchers, bioprocess engineers, industry operators, and animal welfare analysts — to give their estimates for key parameters: media cost, growth factor cost, achievable densities, dominant process mode. We do this before and after a structured day of evidence review, so we can see how expert views shift when confronted with the full technical picture.

The model is fully open — all equations, parameter ranges, and source citations are in the technical documentation. We especially welcome scrutiny of the growth factor assumptions, the bioreactor capital cost ranges, and the correlation structure. Annotate this page or the main pages via Hypothes.is to flag anything that seems off.

--- title: "Script: The Cost Model Explained" subtitle: "Text of the audio overview — read, annotate, or suggest edits via Hypothes.is" format: html: include-after-body: text: | <script src="https://hypothes.is/embed.js" async></script> --- ::: {.callout-note} This is the text of the **cost model audio overview** linked from this site. It was generated by AI (OpenAI GPT-4 + TTS-1-HD, May 2026), based on the content of the [model documentation](docs.qmd), [dashboard](index.qmd), and related materials. It has been reviewed by The Unjournal team and is generally accurate; however, it may state some things more definitively than we would ourselves, and it reflects the page content at a point in time. **To suggest corrections or flag inaccuracies:** annotate this page via the Hypothes.is sidebar (click `<` on the right edge). To return to the main pages: [Interactive Cost Model](index.qmd) · [Technical Documentation](docs.qmd). ::: --- We built this model to answer a deceptively simple question: what would it actually cost to produce cultured chicken meat at commercial scale? The reason this question is hard is not that the math is complicated — it isn't. It's that almost every number that goes into the calculation is genuinely uncertain. We don't yet know what growth factors will cost in 2036. We don't know which bioreactor process will dominate. We don't know how dense cells will grow at scale, or how cheaply media can be made. These aren't things we can look up — they depend on breakthroughs that may or may not happen. So rather than pretending we know the answer, the model runs thirty thousand simulated scenarios. Each simulation draws a different combination of parameter values from realistic distributions, and the result is a full distribution of possible costs — not a single number, but a picture of the range of outcomes and what's driving the uncertainty. That approach is called Monte Carlo simulation, and it's the right tool for this kind of problem. The output is expressed as manufacturing cost per kilogram of cultured chicken cell biomass at harvest — wet weight, at the factory gate, before any texturization or blending with other ingredients. That accounting boundary matters, because a lot of published cost figures refer to different things. --- **What goes into that cost?** We break it into four pieces. The first is variable operating cost — everything that scales directly with each kilogram you produce. This is almost always the dominant component, and within it, two things dominate: cell culture media and growth factors. Cell culture media is the liquid nutrient broth that feeds growing cells. The cost per liter varies enormously depending on the grade and formulation. But what matters for cost isn't just the price per liter — it's how many liters you consume per kilogram of output. And that depends on how dense the cells grow. Importantly, basal media cost and cell density aren't fully independent: richer (more expensive) media formulations are often needed to sustain higher cell densities, particularly in nutrient-depleted fed-batch systems. The total media cost per kilogram reflects both effects. There's also the question of media composition. Expensive purified amino acids are a major cost driver. Hydrolysates — protein-breakdown products from yeast or plant sources — might substitute for them at a fraction of the cost. Growth factors are the second component of variable cost, and they are the source of the model's widest uncertainty band. Growth factors are small proteins — molecules like FGF-2 and IGF-1 — that signal cells to keep dividing. At research scale, they are extraordinarily expensive. At the volumes that commercial cultivated meat would require, no one currently produces them this way. The question is which lower-cost production route will succeed: precision fermentation, plant molecular farming, autocrine cell lines (which produce their own growth factors), or small-molecule substitutes. Any one working at commercial scale by 2036 would be transformative. --- **Capital costs** The second major component is annualized capital cost — the cost of building and financing the bioreactors and facility, spread across all kilograms produced. Large-scale bioreactors are currently expensive, primarily because they're built to pharmaceutical standards that cultivated meat doesn't require. A food-grade bioreactor could potentially be built for a fraction of the pharmaceutical price. The capital cost per kilogram is also sensitive to scale — large plants spread fixed construction costs over more kilograms — and to financing cost (WACC), which depends on how risky investors perceive the sector. --- **Fixed costs and downstream** Fixed operating costs — labor, maintenance, overhead — don't change batch to batch but shrink per kilogram as plant scale increases. Downstream processing adds cost for structured products; for unstructured products it may be minimal. --- **The maturity factor** Cutting across all four components is what we call the industry maturity variable. In a world where cultivated meat succeeds, many things tend to be true simultaneously: growth factor breakthroughs occur, financing costs fall, companies can afford custom food-grade equipment, and hydrolysate substitution is established. The model uses a single maturity draw to introduce realistic correlation, preventing incoherent scenarios. That's a modelling convenience — different technologies could advance at different rates — and it's one of the known structural limitations of the current model. --- **How to read the model** When you open the interactive dashboard, you see a cost distribution — a histogram of simulated outcomes. The median gives you the central estimate. The tail on the right tells you how bad things could get under pessimistic assumptions. The tornado chart shows which parameters move the median cost most when varied across their plausible range. The most important thing about the model is what it tells you about uncertainty, not what it tells you about the central estimate. The shape of the distribution — whether tightly concentrated or spread across two orders of magnitude — is the signal. --- **What we're asking experts** Rather than anchoring on any single set of parameter assumptions, we want to know what people who actually work in this space think about these distributions. The [beliefs form](https://uj-cm-workshop.netlify.app/beliefs.html) asks workshop participants — TEA researchers, bioprocess engineers, industry operators, and animal welfare analysts — to give their estimates for key parameters: media cost, growth factor cost, achievable densities, dominant process mode. We do this before and after a structured day of evidence review, so we can see how expert views shift when confronted with the full technical picture. The model is fully open — all equations, parameter ranges, and source citations are in the [technical documentation](docs.qmd). We especially welcome scrutiny of the growth factor assumptions, the bioreactor capital cost ranges, and the correlation structure. Annotate this page or the main pages via Hypothes.is to flag anything that seems off.