Fermentation protein at $4–6/kg is a modelled cost. Beef at $6–15/kg is a market price.

GFI's June 2025 meta-analysis puts biomass fermentation at $4–6/kg against beef and pork at $6–15/kg. One is a modelled production cost, the other a market price — and GFI's own protein-normalised figures imply its fungal models span 12.5% to 54% protein.

The headline number from the Good Food Institute’s June 2025 fermentation economics report is $4–6 per kg: the modelled cost of production for fungal biomass protein, drawn from five techno-economic models in three publications. GFI sets it against $6–15 per kg for beef and pork, and concludes that “biomass proteins are closing the price gap with several incumbent proteins.”

Read the two numbers carefully and they are not the same kind of number. The first is a modelled ex-works cost of production — capital and operating cost to make a kilogram of material, with no packaging, no freight, no margin. The second is a market price, which by GFI’s own glossary is cost of goods sold plus markup. GFI says this explicitly in a callout box on page 24 and then makes the comparison anyway, which is defensible practice in early-stage techno-economics. What is harder to defend is that the units on the left of that comparison are not internally consistent either. Work backwards through GFI’s own protein-normalised figures and its five fungal biomass models imply protein contents ranging from 12.5% to 54% — which is the difference between a wet paste and a spray-dried powder, both counted as one kilogram of “biomass.”

Methodology note: what GFI counted

This piece rests on someone else’s dataset, so the boundaries matter.

GFI, working with Hawkwood Biotech as co-author, screened 190 publications on biomanufacturing techno-economics. Of those, 48 publications (78 models) were judged in scope on grounds of scale and food relevance; 35 publications comprising 55 techno-economic models met the quality criteria. Twenty publications (36 models) covered the primary ingredients — biomass protein, precision-fermentation single proteins, and microbial fats and oils. The remaining 15 publications (19 models) covered secondary ingredients: amino acids, binders, enzymes, sophorolipids, sugar alcohols and sweeteners.

Biomass fermentation accounts for 25 of those models: five for filamentous or mycelial fungi, four for yeast, one microalgal, one aerobic bacterial, and fourteen for gas fermentation including multistage systems. Only four of the 25 modelled conventional Gen 1 systems using refined sugar — the feedstock almost every commercial biomass producer actually runs. Despite commercial solid-state fermentation being in the market, the exercise found no solid-state models at all, and fed-batch fermentation, the industry standard, is “largely missing from published models.” So the $4–6/kg range is drawn overwhelmingly from processes that are not the ones currently in commercial operation, and GFI says so.

Practical consequence: if you are benchmarking a sugar-fed fed-batch process, only four of the 25 models in this dataset share your feedstock class and none share your fermentation mode. Treat the range as a literature summary, not a peer set.

Cost or price? The comparison GFI flags and then makes

GFI’s glossary is precise. Cost of production is capital plus operational cost to make the product; cost of goods sold is COP plus packaging, palletising and shipping; market price “typically represents COGS plus additional ingredient markup.” The report states the rule directly: “production cost is not market price.”

That is a fair disclosure. The problem is what survives compression. GFI’s own key-insights bullet — reproduced verbatim on the resource landing page — reads: “Biomass cost of production from published models converges around $4-$6/kg, comparable to market prices of $6.0-$15.0/kg for beef and pork.” MycoStories, covering the report in July 2026, ran the headline “Fermentation Protein Costs Are Closing In on Beef and Pork” and wrote that fungal biomass sits “within a comparable range of wholesale beef and pork prices.” Each step is a defensible paraphrase of the one before it. The cumulative effect is that a modelled ex-works cost has become a statement about price parity.

Practical consequence: before you quote $4–6/kg to an investor or a customer, add the missing layers. Dewatering or drying, packaging, palletising, cold chain and freight sit between COP and COGS, and margin sits between COGS and price. None of them are in this number.

Dry or wet? Every figure with its stated basis

Here is each cost and price figure in this story with the basis actually stated by its source. Where a source does not state a basis, the table says so rather than guessing.

Figure	What it measures	Dry or wet	Scale assumed	Source	Year / currency
Range $1.30–18.10/kg, median $4.30/kg, “strong alignment” $4–5/kg	Modelled cost of production per kg biomass, all bioprocesses in the biomass set	Not stated; the set mixes both	60 to over 100,000 t/yr	GFI meta-analysis, all biomass TEMs	USD; base year not stated
$3.80–6.00/kg, stated by GFI alongside $7.00–48.00/kg-protein (Table 1 gives $3.90–5.90)	Modelled COP per kg biomass, fungal protein only	Not stated	3,400–40,500 t/yr	GFI, 5 fungal TEMs in 3 publications	USD; base year not stated
$7.50/kg (median)	Modelled COP per kg biomass, yeast	Not stated	60–10,000 t/yr	GFI, 4 yeast TEMs	USD; base year not stated
$2.53/kg (bounds $2.25–2.90)	Modelled COP, spray-dried single-cell protein	Dry, 54% protein by weight	32,000 t/yr, titre 142 g/L	Hawkwood FEL-1 pro forma, published inside the GFI report	USD; base year not stated; CapEx accuracy +40%/−25%
$3.55/kg	Modelled COP, mycoprotein at the fermenter	Wet, approximately 73% moisture	2,000 kg/h production rate, continuous airlift	Risner et al. 2023	USD; model published 2023
$6.00–15.00/kg	Market price, beef and pork, whole, ground and sliced	Wet commodity meat	Not applicable	GFI Figure 10, compiled from USDA ERS, USITC and FRED	USD; date of price observation not stated
$11.83/kg net farm, $12.29/kg wholesale, $23.12/kg retail	Choice beef values per pound of retail weight	Wet	Not applicable	USDA ERS Meat Price Spreads, July 2026	USD, July 2026
$10.78/kg retail, $2.66/kg net farm	Pork values per pound of retail weight	Wet	Not applicable	USDA ERS Meat Price Spreads, July 2026	USD, July 2026

Three things fall out of that table. First, GFI reports its own fungal range three different ways in the same document: $4–6/kg in the executive summary, $3.80–6.00/kg in section 1.2.4, and $3.90–5.90/kg in Table 1. Small differences, but in a number now used as a sector benchmark.

Second, the two most specific biomass figures in the literature have opposite moisture bases. Hawkwood’s pro forma is a dried product at 50–57% protein. Risner’s mycoprotein is a wet biomass at roughly 73% moisture. Both are quoted in dollars per kilogram of biomass. On a dry-matter basis Risner’s $3.55/kg wet is around $13/kg dry — outside the $4–6 range entirely, and above every fungal figure GFI reports.

Third, GFI’s beef and pork band contains both ends of the supply chain. In July 2026, USDA ERS put Choice beef net farm value at $11.83/kg and pork retail value at $10.78/kg. Both sit inside $6–15/kg. They are not the same kind of quantity.

Practical consequence: ask every supplier quoting a cost per kilogram for a single number — moisture content as sold. Without it the figure cannot be compared to anything, including their competitor’s figure.

Protein or biomass? The conversion, worked

GFI does publish protein-normalised figures alongside the biomass ones, which is more than most of this literature manages. Those figures are where the basis problem becomes arithmetic rather than argument. Dividing GFI’s per-kilogram cost by its per-kilogram-of-protein cost recovers the protein content that was assumed.

Material	Cost or price, $/kg as sold	Protein, g/100 g	Where the protein figure comes from	$/kg protein (= col 2 ÷ col 3 × 100)
Hawkwood pro forma dried SCP	2.53	54	Stated in GFI Table 3 for the hypothetical commercial case	4.69
GFI published biomass, median	4.30	22.3	Back-calculated from GFI’s own $19.30/kg-protein median	19.30
GFI fungal biomass, low end	3.80	54.3	Back-calculated from GFI’s $7.00/kg-protein floor	7.00
GFI fungal biomass, high end	6.00	12.5	Back-calculated from GFI’s $48.00/kg-protein ceiling	48.00
Risner mycoprotein, wet	3.55	12	Risner’s stated assumption; Derbyshire and Delange give 11 g/100 g	29.58
US ground beef, retail, July 2026	15.18	18.4	USDA Foods fact sheet, raw 85/15 ground beef	82.50
US Choice beef, net farm value, July 2026	11.83	18.4	USDA Foods fact sheet, raw 85/15 ground beef	64.29
US Choice beef, net farm value, March 2022	6.56	21.8	Beef mince, stewed — the figure Risner used	30.09
US Choice beef, net farm value, March 2022	6.56	18.4	Raw 85/15 ground beef, same price, corrected basis	35.65

The unit conversions behind the USDA rows: ERS reports in cents per pound of retail weight. July 2026 Choice beef retail value was 1,048.7 cents/lb; the wholesale-to-retail spread was 491.1 cents/lb, giving a wholesale value of 557.6 cents/lb; the farm-to-wholesale spread was 21.2 cents/lb, giving a net farm value of 536.4 cents/lb. That last figure checks against ERS’s published farmer’s share of 51.1% (536.4 ÷ 1,048.7 = 51.1%). Converting at 2.20462 pounds per kilogram: 536.4 c/lb → $5.364/lb → $11.83/kg; 557.6 → $12.29/kg; 1,048.7 → $23.12/kg. Retail ground beef at 688.5 c/lb → $6.885/lb → $15.18/kg.

The rows worth staring at are three and four. GFI’s five fungal models produce a biomass cost range of $3.80–6.00/kg and a protein cost range of $7.00–48.00/kg-protein. Those two ranges are only mutually consistent if the underlying models assume protein contents from 12.5% to 54.3%. A 12.5%-protein material is wet fungal biomass. A 54%-protein material is a dried powder. The “$4–6/kg fungal biomass” range is therefore not a range of costs for one product. It is a range of costs for materials that differ by a factor of four in what a formulator would actually be buying.

Practical consequence: normalise every quote to dollars per kilogram of protein before you compare anything, and state the protein assay method. On a protein basis, the honest read of this dataset is that fungal biomass runs from $7 to $48 per kg of protein — a spread wide enough that the midpoint carries no information.

What the beef comparator actually is

GFI states its incumbent price sources — USDA ERS, USITC and FRED — and notes that “beef and pork include whole, ground, and sliced products,” gathered from “wholesale and retail price markets.” It does not state which series, which month, or which stage of the chain each end of the $6–15/kg band represents. It also warns that trade pricing “may not fully reflect market values due to product aggregation, limited specification detail, and atypical or negotiated transactions.”

Going to the source: USDA ERS Meat Price Spreads reports monthly farm, wholesale and retail values for the meat of a standard Choice steer and a standard hog, in cents per pound of retail weight. Retail prices come from BLS Consumer Price Index surveys; farm and wholesale prices from USDA AMS under mandatory price reporting, converted using ERS factors of 2.40 pounds of standard steer and 1.14 pounds of wholesale beef per pound of retail beef. Net farm value is gross farm value minus the byproduct allowance for hides, offal, fats and bones.

That gives, for July 2026: Choice beef at $11.83/kg net farm, $12.29/kg wholesale, $23.12/kg retail; pork at $2.66/kg net farm, $10.78/kg retail. Beef retail value is now above the top of GFI’s $6–15/kg band and has been moving fast — the CPI for beef and veal rose 16.4% year on year in December 2025 and was still up 9.4% in July 2026. That matters more than the basis argument: a parity claim benchmarked to an undated meat price band will drift out of validity on the comparator’s side alone, without anything happening in the fermentation plant.

Practical consequence: date-stamp the comparator. If you are writing a business case against beef, cite a specific ERS series, stage and month, and re-run it quarterly.

The private number, and who produced it

The $2.50/kg figure that has travelled furthest from this report is not from the published literature at all. It comes from a pro forma built by Hawkwood Biotech: aerobic submerged stirred-tank fed-batch fermentation of budding yeast or microalgae on corn glucose (DE95), Midwest USA location, spray-dried product at 50–57% protein. The hypothetical commercial case assumes 32,000 tonnes a year, 0.47 g/g yield, 142 g/L titre and 96% downstream recovery, and returns a unit COP of $2.53/kg with CapEx of $237 million and OpEx of $80.8 million a year. Glucose is priced at $0.40/kg and electricity at $0.08/kWh. It is a front-end loading stage 1 estimate, with CapEx accuracy declared at +40%/−25%.

The disclosure that belongs next to that number: Hawkwood is a co-author. GFI’s resource page carries a “Co-authored by Hawkwood Biotech” credit and the author list names Adrienne McKee, Partner at Hawkwood, alongside three GFI staff. The report’s central argument — that published models understate industry performance — is evidenced by comparing public models to Hawkwood’s private models, supplied by Hawkwood, and unverifiable by anyone else. That is not a criticism of the number. It is a statement of what kind of evidence it is.

One arithmetic detail does not reconcile. GFI reports the private yeast model at “a median cost of $2.5/kg and $4.2/kg-protein,” which implies a protein content of 59.5%. The stated protein range for the modelled product is 50–57%, and the hypothetical case value is 54%. At $2.53/kg and 54% protein the answer is $4.69/kg-protein, not $4.20. The two figures presumably come from different models within the private set of three, but the report does not say which, and the pairing cannot be reconstructed from what is published.

Practical consequence: when a private benchmark is the load-bearing evidence in a public report, ask who built it and whether they have a commercial interest in the conclusion. Here the answer is on the cover.

Where the sources disagree

Three disagreements are worth naming.

GFI contradicts itself on the beef and pork band. Section 1.2.4 states that “ground and whole chicken/turkey and beef/pork ranged from $2.0 to $3.4/kg and $6.0-$15.0/kg, respectively.” Two paragraphs later the same section states that mycoprotein “has the potential to compete on a cost per kg-biomass with beef and pork whole and ground products ($2.0–$3.4/kg).” The parenthetical has taken the poultry number. We have used $6.0–15.0/kg as the beef and pork figure, because that is the version repeated in the executive summary and on GFI’s public resource page.

Risner’s parity claim mixes cooked and raw beef. Risner et al. modelled wet mycoprotein at $3.55/kg and 12% protein, giving $29.56/kg-protein, against USDA net farm value for Choice beef of $6.56/kg in March 2022 converted at roughly 21.9% protein to $29.95/kg-protein — hence “achieving price parity with beef protein.” But 21.8 g/100 g is the protein content of beef mince stewed, from Derbyshire and Delange’s table, while ERS net farm value is per pound of retail weight, which is raw. Raw 85/15 ground beef runs 7 g protein per 38 g serving — 18.4 g/100 g — on the USDA Foods fact sheet sourced from FoodData Central. At the same March 2022 price that gives $35.65/kg-protein for beef against $29.58 for mycoprotein. Correcting the basis strengthens the mycoprotein case by about 17%. The point is that nobody checked.

MycoStories tracks the report closely but hardens two edges. Its per-kilogram figures, the $2.50/kg and 32,000 t/yr Hawkwood case, the titre sensitivity (30% up gives 12% cost down, 30% down gives 22% cost up), the 14.3% cost impact from a 40% feedstock price rise and the 11% impact from a 10% loss of downstream recovery all match the report exactly. Two departures: it gives the fungal range as $3.90–6.00/kg, splicing the low end of Table 1 to the high end of section 1.2.4; and it describes the report as “produced with analytical support from Hawkwood Biotech,” where GFI’s own page says co-authored.

The case for the comparison

Comparing modelled cost to market price is standard practice in early-stage techno-economic analysis, for the simple reason that new products have no market price. There is nothing else to benchmark against. GFI does not hide this: it defines the three terms in a glossary, restates the distinction in a boxed callout, and warns twice that “costs of production are not reflective of market prices.”

The directional finding also looks robust. Whatever the basis, biomass fermentation is far cheaper to model than precision fermentation, and the gap to animal protein is narrower than the gap for microbial oils against commodity vegetable oils. The report is equally clear that biomass protein struggles against plant protein concentrates on a protein basis — median $19.30/kg-protein against roughly $6.00/kg-protein for soy and pea concentrates — which is not the finding you would write if you were spinning.

GFI’s central methodological point also stands independently of the price comparison: published models assume volumes averaging around 10,000 t/yr where commercial practice runs at 32,000 t/yr and above, and understate titre. If that is right, the true cost sits below the published range and the comparison to meat is conservative rather than flattering.

The objection is narrower than “the report is wrong.” It is that a number carrying this much weight needs its basis attached every time it is quoted, and it currently is not.

What we could not establish

The base year and currency normalisation. The report is dated June 2025 and quotes USD throughout, but no base year, deflator or currency-conversion method appears anywhere in the body text we could retrieve. Appendix 1, “Materials and methods,” is on pages 68–69; our PDF text extraction truncated at page 58, so we could not read it.
Whether the $4–6/kg range is ex-works, COGS-inclusive or capex-inclusive per model. GFI defines COP as capital plus operational cost and states that CapEx enters as depreciation spread over annual volume, which implies capex-inclusive ex-works. Whether all 25 underlying models applied that boundary consistently is not reported.
Which USDA, USITC or FRED series produced each end of the $6–15/kg beef and pork band, and for which month. GFI publishes an Airtable of incumbent product prices; we were unable to read its contents.
The Mycorena Promyc price of roughly €6–9/kg. No primary source. It circulates through market-research aggregators; the IndexBox page returned no pricing content when fetched. Mycorena’s own technical material states Promyc contains up to 60% protein and 12% fibre, but we found no company-published price — and Mycorena entered bankruptcy in 2024, with Promyc production moving to Naplasol in Bree, Belgium. We have left the figure out.
The March 2022 USDA net farm value of $6.56/kg used by Risner. ERS’s rolling files cover the last 24 months and its historical file was too large to retrieve. We report it as Risner does.
Moisture content for the beef composition figure. The USDA Foods fact sheet gives protein, fat and calories but not water.

What to watch

Whether GFI restates the $4–6/kg range with a moisture basis. The resource page was last updated in December 2025 and the report invites new model submissions through an open form. A dry-matter or protein-normalised headline would settle this.
Whether any published TEM appears for a Gen 1 sugar-fed, fed-batch biomass process at commercial scale. GFI names this as the central gap. One such model would test the claim that published costs are inflated by sub-commercial assumptions, without requiring anyone to trust a private pro forma.
Whether ERS Choice beef retail value stays above $15/kg. It was $23.12/kg in July 2026, already outside GFI’s comparator band. If beef holds, any parity claim benchmarked to the 2025 band understates the commercial opportunity while overstating the technical achievement.
Whether a biomass producer publishes a price with a stated moisture and protein basis. No one in this dataset has. The first company to quote dollars per kilogram of protein at a stated assay will make everyone else’s number look evasive.