9. Economics: what it costs and what it must beat

In one paragraph. Vietnam's operating inputs are cheap: power at about USD 0.075 per kWh for a plant running around the clock, a minimum wage of about USD 204 a month, factory rent of about USD 5 per m² a month. But in fermentation those inputs are a small part of cost. On our indicative Vietnamese cost stack, capital charges and maintenance are 54 to 67% of the cost of fungal feed protein, which comes to USD 4,050 to 14,700 per tonne of protein. Fishmeal protein costs USD 2,625 (2025 average) to USD 3,846 (August 2026) per tonne, and soybean meal protein about USD 880. So a Vietnamese microbial protein sold only on protein content is marginal against fishmeal and far from soy. It becomes plausible as a functional shrimp or fish-feed ingredient used at 1 to 5% of the diet, where buyers pay for survival and feed efficiency. For plant-based food the arithmetic runs the other way: textured soy protein costs roughly 3,700 to 10,700 VND per 100 g of protein to make, against 34,000 to 70,000 VND at retail for eggs, chicken and pork. Quality, demand and distribution decide that market, not factory cost.

---

#9.1 Input costs in Vietnam

Three rule changes add risk. A decree now lets the power utility adjust the average price by 2% to under 5% after a ministry review; a two-part tariff with a capacity charge is being shadow-billed to large users through 2026 and could apply from 2027 ^21,22 VN-directMedium; and direct renewable power purchase is allowed but its savings (about 10% in one worked example) are unproven ^23,24 VN-directLow.

Feed is VAT-exempt, which is a hidden cost. Under the VAT Law in force since July 2025, animal feed is "not subject to VAT" ²⁵ VN-directHigh. A plant selling a feed protein therefore cannot reclaim the VAT it pays on equipment, power and inputs. We estimate this adds about 5 to 10% to its taxed input costs; a food-ingredient plant can reclaim input VAT VN-directLow.

#9.2 Fermentation: capital is most of the cost

Published benchmarks (full list in tea_benchmarks.csv; read in full, in part or as abstracts, as recorded there) ^{26,27,28,29,30} generalMedium:

Capex varies seven-fold. Per cubic metre of reactor, the off-gas SCP study implies about USD 51,000 and the mycoprotein study about USD 350,000; per annual tonne, USD 16,000 to 23,000 for SCP and mycoprotein and about USD 96,000 for cultivated meat (our derivation) ^26,27,30 generalHigh. Design, hygiene grade and estimating method explain the spread. A Vietnamese plan needs its own equipment-level estimate, not a borrowed ratio.

Capital cost per annual tonne of capacity

USD per tonne of annual capacity; published benchmarks and our assumptions

Cooling in the tropics. Aerobic fermentation releases about 12 to 18 MJ of heat per kg of dry biomass. In southern Vietnam, we assume cooling-tower water arrives at about 31 to 33 °C, too warm to hold a broth at 30 °C, so most cooling must come from chillers. Our indicative estimate of the penalty is USD 70 to 180 per tonne of dry biomass, about 2 to 4% of the cost of fungal feed protein. Strains that grow at 40 to 45 °C avoid most of it ^31,32,33 VN-directLow. Earlier drafts said cooling "may offset" Vietnam's power advantage: that is partly true for the power bill, but because power is a small share of cost, the overall effect is modest.

#Cost stack A: fungal biomass protein for aquafeed, 10,000 t a year

Source: AltProtein Vietnam calculation from sourced inputs ^{1,5,7,16,26,27,30} VN-directLow.

Capital is most of the cost of fermentation protein

Indicative cost of fungal feed protein made in Vietnam, USD per tonne of product, low and high cases

Reading. Capital charges and maintenance are 54% of cost at the low end and 67% at the high end. Labour and power together are under 10%. Two lines deserve attention:

• Carbon. At April 2026 cassava starch prices (USD 520 to 540 per t), the carbon line sits at or above the top of the assumed range once conversion to glucose is added. A near-free side stream such as cassava pulp would cut it, but needs pre-treatment and has almost no protein (Asset map). Domestic sugar at a three-year low is a clean alternative in surplus years.
• Capex. The product nears fishmeal parity only if capex falls below about USD 6,000 per annual tonne, which is well under any verified benchmark. Tolling on existing tanks (Industrial base) is the main way to get there.

#9.3 The prices to beat in feed

Fishmeal (Peru 65%, the World Bank series) averaged USD 1,706 per tonne in 2025 and reached USD 2,500 in August 2026, the highest in the 2020 to 2026 series. Soybean meal was about USD 403 ^34,35 generalHigh. On a protein basis, that is USD 2,625 to 3,846 per tonne for fishmeal and about USD 880 for soybean meal (our calculation).

Source: ^34,36,37,38 and cost stack A VN-directLow. Arithmetic for the functional case: maximum ingredient price = soybean meal price + (performance gain × feed price) ÷ inclusion rate.

The price to beat in feed

USD per tonne of protein

The real competitor is imported microbial protein, not fishmeal list prices. A bacterial biomass from Chinese glutamic-acid fermentation, with at least 70% crude protein, is already imported and sold into Vietnamese pig, poultry, fish and shrimp feed ^39,40 VN-directMedium. Calysseo's 20,000 t a year FeedKind plant operates in Chongqing ⁴¹ generalMedium. By-product protein has near-zero marginal feedstock cost and can be priced anywhere between soybean-meal and fishmeal parity. Our working assumption is that a Vietnamese feed protein should plan to realise 70 to 90% of fishmeal-protein parity VN-directLow.

Buyer economics. Feed is 50 to 60% of shrimp production cost ³⁷ VN-directMedium. At 15% inclusion, fishmeal adds about USD 256 per tonne of shrimp feed at the 2025 price and USD 375 at the August 2026 price (our calculation from ³⁴). Shrimp feed prices rose by VND 1,200 to 1,500 per kg in June 2026 ³⁷ VN-directMedium. Mills are motivated to cut fishmeal, but they already use established substitutes (poultry by-product meal, fermented soy, soy protein concentrate) ³⁶ VN-directMedium.

#9.4 Plant-based food: factory cost is not the problem

#Cost stack B: textured soy protein, 3,000 t a year on one extrusion line

Source: AltProtein Vietnam calculation ^1,5,7,42 VN-directLow.

Retail protein prices to beat (per 100 g of protein): at a national grocery chain on 23 September 2026, eggs 34,000 to 40,000 VND, chicken breast 39,500 to 45,500 and lean pork 63,000 to 69,500 ^43,44,45 VN-directMedium; in our retail audit on 16 and 20 September 2026, tofu 39,000 to 77,000, dried soy mock meat about 51,000 and an imported plant-based burger about 542,000 ⁴⁶ VN-directMedium.

Price per 100 g of protein, shelf products against meat and eggs

Median VND per 100 g of protein by product type (retail audit) with reference ranges for eggs, chicken and pork

Reading. Textured protein costs between about one-twentieth and one-third of the retail price of meat and egg protein to make. The gap is taken by distribution, brand and retail margins and by product quality. Raw material is about 70% of cost and is imported, so Vietnam's cost edge over a Thai or Chinese plant is small, perhaps USD 50 to 150 per tonne from labour and rent VN-directLow. The import it would replace is cheap: Chinese textured protein and concentrates entered Vietnam at about USD 1.6 per kg in 2023 (China-reported unit value, before freight and duty), inside our Vietnamese cost range ^48,49 VN-directMedium. A domestic textured-protein plant competes on freshness, logistics, customisation for local dishes and supply security, and on replacing imported textured soy protein at scale (Alt protein today, Plays play T1).

Chicken and eggs set the floor. Any plant-based product aimed at mass-market Vietnamese consumers is priced against chicken and eggs at about 34,000 to 46,000 VND per 100 g of protein ^43,44 VN-directMedium. Imported premium analogues sit an order of magnitude above that.

#9.5 Precision fermentation and cultivated meat

• Precision fermentation. We could not verify a peer-reviewed cost model for food proteins this round. The logic from the regional study still applies: in a 10,000 t a year precision-fermentation plant model, labour was 17% and utilities 4 to 5% of cost, and depreciation was the largest line ⁵⁰ generalHigh. For Vietnam, this favours high-value, low-dose products (enzymes, functional proteins, specialty ingredients) made under contract, not commodity dairy proteins (Technology fit).
• Cultivated meat. At USD 37 to 51 per kg of wet cells ³⁰ generalHigh, cultivated meat costs roughly 7 to 15 times the retail price of lean pork or chicken breast in Vietnam (our calculation from ^43,44). With capex near USD 96,000 per annual tonne, there is no near-term production case for Vietnam; the case is for research and regulatory readiness (Plays play T8).

#9.6 What policy can change in the numbers

The regional study modelled incentives for representative plants and found that a 50% capital grant or non-dilutive funding raised company value the most (a reported NPV gain of USD 73 million), followed by a 10% cut in feedstock cost and a loan guarantee. Tax holidays ranked near the bottom ⁵⁰ generalHigh. In Vietnam's case, that points to:

• Capital: shared or tolled tanks, capital grants and loan guarantees rather than more tax holidays (Policy options).
• Feedstock: stable access to glucose from cassava starch or domestic sugar at fermentation grade.
• Tax neutrality: allowing new feed ingredients to reclaim input VAT.
• Time: every month of feed or food approval is carried as capital cost (Rules).

#9.7 What this means

• Investors: model capital charges, not operating costs. Test any feed-protein pitch against USD 2,625 to 3,846 per tonne of fishmeal protein, against Chinese by-product protein, and against capex per annual tonne (verified benchmarks: USD 16,000 to 23,000). A claim below USD 6,000 needs evidence.
• Startups: sell function, not tonnage. Choose thermotolerant strains or design for hybrid cooling from day one. Start texturing in a ready-built factory at about USD 5 per m² a month.
• Manufacturers: feed mills can buy functional ingredients at a premium only with species-specific trial data. Food makers can produce textured protein in Vietnam at a small fraction of meat's retail price; demand and quality are the constraints.
• Policy makers: the cheapest levers on cost are capital sharing (a public or tolled pilot line), input-VAT neutrality for new feed ingredients, and early signals on the capacity charge.
• Research bodies: two public goods are missing and cheap: an open Vietnamese fermentation cost model with local inputs, and a public series of domestic feed-ingredient prices.

Related: Costs and benchmarks (full input tables, both cost stacks, benchmarks), Feed and aquafeed market (fishmeal, inclusion rates, trials), Technology fit.