Fermentation is already cheap, for some things
The same technique, a microbe or cell in a tank, makes products that cost from under a dollar to tens of thousands of dollars a kilogram. What separates them is not the biology; it is what the process has to pay for.
What a kilogram costs, from antibody to grain
Cost of goods, not price: reviews put today's antibody manufacturing at tens to hundreds of dollars per gram, plotted at $50 a gram. Mammalian cells, sterile fed-batch, 5–8 g/L, and multi-step chromatography. Selling prices are far higher, near $2,000 a gram.
Six layers between a cell and a kilogram
A production organism is only the first layer. Most of the cost of a bio-manufactured product is set by the equipment that keeps it alive, fed, uncontaminated, and separated from everything else.
Strain
A cell line or microbe engineered to turn feedstock into product: CHO cells for antibodies, bacteria for amino acids, yeast for ethanol and food proteins.
- Measure
- Titer · yield · specific productivity
- Failure boundary
- A strain that performs in a 2 L flask but loses productivity or stability over the generations a production tank requires.
Where the frontier moves
DNA synthesis, genome editing, and automated screening have made new strains fast to build. This is no longer usually the slowest layer.
Feedstock and media
Sugar for microbes; for animal cells, a chemically defined broth of glucose, amino acids, vitamins, salts, and sometimes growth factors.
- Measure
- $/kg product · carbon yield
- Failure boundary
- Pharmaceutical-grade media priced for grams, not tonnes. Animal-cell media has never been made at food scale.
Where the frontier moves
Plant hydrolysates and food-grade amino acids instead of pharmaceutical inputs; recombinant growth factors made cheaply by microbes.
Sterile vessel
A pressure-rated steel tank or single-use bag, steam-sterilised, sealed, with every valve, probe, and gas line kept free of contaminants for days or weeks.
- Measure
- $/litre installed · batch failure rate
- Failure boundary
- One contaminating microbe can outgrow the product organism and lose the whole batch.
Where the frontier moves
Robust hosts that tolerate open or semi-sterile operation: acid-, salt-, or heat-loving organisms that other microbes cannot survive with.
Oxygen and heat
Aerobic cells need oxygen dissolved in water, where it is sparingly soluble, and they release heat in proportion to the oxygen they consume.
- Measure
- Oxygen transfer rate · W/L removed
- Failure boundary
- At large volume, mixing time and gas transfer lag behind demand; cells in part of the tank starve or overheat, and bubbles damage animal cells.
Where the frontier moves
Better spargers and reactor geometry, anaerobic or low-oxygen product pathways, gas fermentation, and continuous processes that run smaller tanks harder.
Harvest and purification
Separating product from cells, broth, host proteins, DNA, and viruses: centrifugation, filtration, chromatography, crystallisation, or distillation.
- Measure
- % recovered · $/kg purified
- Failure boundary
- For injectable proteins, each step loses product and costs resin, buffer, and time; this is now where most antibody cost sits.
Where the frontier moves
Continuous chromatography, cheaper affinity ligands, and products (food, feed, materials) that do not need injectable purity.
Quality and release
Proving each lot is what it says it is, at the purity, potency, and safety its use requires.
- Measure
- Days to release · lots rejected
- Failure boundary
- Release testing and documentation take fixed time whatever the tank's output.
Where the frontier moves
In-line analytics and real-time release, and regulators that accept them.
Three bills, and only one of them shrinks with titer
Cost per kilogram is feedstock divided by yield, plus tank time divided by titer, plus purification, all divided by what purification recovers. Better cells shrink the middle term. The first is bounded by chemistry and the last by purity.
Carbon sets the feedstock floor. Glucose is 40% carbon by mass; protein is roughly half carbon. Even if every carbon atom ended up in product, a gram of protein would need about 1.3 grams of glucose (Derived). Cells also burn sugar for energy, so real yields are lower. At sugar near $0.40 a kilogram the feedstock floor for fermented protein is on the order of a dollar a kilogram: cheap for a drug, significant for food.
Oxygen and heat set the tank-time floor for aerobic processes. Oxygen dissolves in water at only a few milligrams per litre, so it must be transferred continuously; industrial fermenters are designed for transfer rates of the order of a few hundred millimoles per litre per hour. Aerobic metabolism releases roughly 460 kJ of heat per mole of oxygen consumed, so a tank running at 300 mmol/L/h must remove close to 40 watts per litre (Derived). A 200 m³ fermenter at that rate rejects several megawatts of heat. Productivity per litre is capped by how fast the tank can breathe and cool, which is why ethanol, an anaerobic process, is the cheapest fermentation product at scale.
Purity sets the purification floor. An injectable antibody must be separated from host-cell proteins, DNA, and potential viruses to parts-per-million levels. Feed-grade lysine can be dried with much of its broth. Most of the five orders of magnitude in Figure 2 is the difference between those two standards.
What sets the cost of a kilogram grown in a tank?
Every fermented product pays three bills: tank time, which falls as titer rises and batches shorten; feedstock, which is fixed by chemistry; and purification. Move titer and see which bill starts to dominate.
That is $50 a gram.
- Tank time (vessel, sterility, oxygen, cooling)
- $23,333 · 47%
- Feedstock (sugar or media ÷ yield)
- $133 · 0%
- Purification
- $26,603 · 53%
- Feedstock floor alone
- $133
Presets are illustrative and calibrated only to the order of magnitude of the prices in Figure 2. They are not audited plant data. The antibody preset lands near $50 a gram and the ethanol preset near $0.75 a kilogram before co-product credits such as distillers' grains.
Calculation and boundaries
Tank time per kg = (cost of one litre of installed, sterile, aerated, cooled tank for one day) × batch days × (1,000 ÷ titer in g/L). Feedstock per kg = feedstock price ÷ product yield. Total = (tank + feedstock + purification) ÷ purification recovery. Tank cost bundles capital recovery, utilities, labour, and quality systems, which is why it spans four orders of magnitude between a pharmaceutical suite and an ethanol plant. Excludes co-product credits, turnaround time between batches, failed batches, and margins.
From building the cell to building the tank
For forty years the hard part was getting cells to make enough. That problem was largely solved. The binding constraints are now sterility, oxygen, heat, purification, and the capital they require.
Strains became cheap
Reading, writing, and editing DNA made building a production organism a matter of weeks. Titer rose a hundredfold, mostly through media and process control.
Upstream stopped dominating
As titer climbed past a few grams per litre, reviews describe the production bottleneck moving into downstream purification, which can account for more than half of antibody manufacturing cost.
Capital became the bill
In fed-batch antibody plants about 59% of cost of goods is capital: a sterile, validated plant costs the same whether the cells are productive or not.
Food needs a different machine
For products worth a few dollars a kilogram, tank time must fall by orders of magnitude, not percentages. That is a reactor-design and sterility problem, not a genetics problem.
What a field gets for free
A cornfield is a bioreactor that is never sterilised, powered by sunlight rather than electricity, aerated by the atmosphere, cooled by evaporation, and harvested without purification. Its productivity per unit of volume is tiny compared with a fermenter; its cost per unit of output is tiny too. Nature pays with land and time instead of steel and energy, and it accepts a mixture, grain, stalk, and root, rather than one molecule at one purity.
Industry has partial copies of each of those tricks. Ethanol skips oxygen transfer and tolerates imperfect sterility because yeast at high sugar and alcohol concentrations outcompetes most contaminants. Citric acid is made by a fungus at a pH few competitors can tolerate. Research groups have run engineered salt-loving bacteria in open, unsterilised fermentations to make bioplastics, and gas fermentation plants turn steel-mill off-gas into ethanol at commercial scale. Egg-based influenza vaccine production uses biology's own sealed vessel. Each of these avoids a layer of the stack rather than optimising it.
Cultivated meat is the hardest test. Humbird's 2021 techno-economic analysis concluded that low growth rate, metabolic inefficiency, CO₂ inhibition, and bubble damage limit practical bioreactor size and cell density, that contamination safeguards make the equipment expensive, and that suitable amino acids and growth factors are not made at food scale. The Good Food Institute disputes several of its cost assumptions, including equipment prices and installation factors. Neither side disputes the structure: animal cells in a sterile, aerated tank carry every bill in the equation, and food prices leave no room for any of them.
The optimistic view, with conditions
The fermented commodity class widens
If robust hosts, food-grade media, and simpler reactors let proteins and fats be made the way lysine and citric acid already are, a new class of products could reach a few dollars a kilogram without needing nature's land.
Cheaper biologics
Higher titer and continuous purification push antibody cost of goods toward low tens of dollars a gram, which matters for global access more than for margins.
Food ingredients by precision fermentation
Microbial production of specific proteins and fats in large, aerobic tanks with food-grade purification: the lysine playbook applied to higher-value molecules.
Open, robust bioprocessing
Hosts that tolerate unsterile operation, gas feedstocks, and continuous processing that together cut the tank-time bill by orders of magnitude.
Where each measure stands
View the scorecard
| Measure | Record through September 2026 | What the optimistic case needs |
|---|---|---|
| Antibody titer, commercial | >8 g/L | Rising slowly; the cost now sits elsewhere |
| Antibody cost of goods | $10s–100s per gram | Low tens of dollars per gram for global-access biologics |
| Aerobic bulk fermentation | About $1–2/kg (lysine, citric acid) | The same cost class for proteins and fats |
| Animal cells as food | Not demonstrated at commodity scale | A few dollars per kilogram, without pharmaceutical media or sterility |
| Non-sterile production | Anaerobic ethanol; acid-tolerant citric acid | Engineered robust hosts making higher-value products in open tanks |
Baselines are reported results or approximate market prices. Targets are editorial benchmarks, not forecasts.
What cheap biomanufacturing actually needs
- Cheaper tank timeReactors that cost less per litre-day, run continuously, and do not need pharmaceutical sterility for food and materials.
- Oxygen-efficient productsPathways and hosts that need less oxygen per kilogram, or none, so heat and gas transfer stop capping productivity.
- Food-grade inputs at scaleMedia and growth factors made at tonne scale and priced like feed, not like reagents.
- Purity matched to usePurification designed for what the product must be, not inherited from injectable drugs.
- UtilisationPlants that run full, because capital, not sugar, is most of the bill.
Sources, method, and boundaries
The titer curve uses ranges from industry reviews and one development-scale measurement; it shows the shape of the curve rather than audited plant averages. The price ladder mixes one cost-of-goods estimate with approximate 2025 market prices, which include margin. The cost equation and calculator are editorial frames calibrated to order of magnitude only. Heat and carbon figures are derived from standard stoichiometry and are marked Derived.
- Titer
- Concentration of product in the culture at harvest, in grams per litre.
- Fed-batch
- A culture fed nutrients during a run of one to three weeks and harvested once.
- Cost of goods
- The manufacturer's cost to produce a unit of product, including capital recovery, before margin.
- Downstream processing
- Everything after the tank: harvest, purification, formulation.
