Summary
Executive summary
Vietnam's case for making protein at home is a security case, real but narrow. The first openings are domestic textured plant protein, a cassava-to-fermentation-sugar hub and shared pilot fermentation. Firms, not shoppers, would buy first, and the main public lever is the new Food Safety Law.
On this page
New to alternative protein? Read the prologue first: Prologue. Short on time? Read In two minutes.
Protein for people first; feed only as context. This study asks what Vietnam could make at home for people to eat, at what cost and under which rules, and who would buy it, now and to 2050. Demand is a design input, not a marketing plan. Feed appears only as the hidden import behind the meat, eggs, milk and farmed fish that people eat.
Why Vietnam. Vietnamese people eat more meat each year: 136.4 g per person per day in 2020, against a recommended 50 to 80 g. Animal foods supply 42% of protein. About 90% of the meat is produced in Vietnam, but on imported feed. In 2025 about 99% of the soy protein in feed and 74% of the maize were imported, for a feed-ingredient bill of about USD 10 billion.
So the case for making more protein at home rests on the security of supply, then on industrial value and trade (Why Vietnam). Environment and animal welfare motivate some actors. The security case is real but narrow.
Food security: what the evidence supports, and what it does not
We tested the main food-security claims made for alternative protein in Vietnam (Why Vietnam, Protein diet, Drivers to 2050).
- Two thirds of animal protein rests on imports. In 2023, 19% to 23% of the protein supply was imported food. Counting feed, 61% to 71% of animal protein rested on imports (our calculation).
- Imported meat is growing, but not subsidised. Meat imports reached 978,000 t in 2025. For most of it, lower farm costs abroad explain the price, not subsidies. New protein would displace at most about 1.6% of imported meat protein by 2035 (our calculation).
- Disease is the shock consumers feel. Pork prices rose 57% in 2020 after African swine fever. People switched to chicken and imported pork, not tofu.
- We found no evidence that alternative protein lowers disease risk. No study we found measures lower pandemic or antimicrobial-resistance risk from replacing meat. Swine fever culls in 2025 were 8.4 times the pigs new protein would spare in 2035 (our calculation).
- Only domestic inputs reduce dependence. About 84% of the plant-protein food ingredients reaching Vietnam in 2025 came from China (our calculation). Alternative protein made from them moves the dependence. Fermentation on cassava or sugar, rice and peanut protein, and soy foods from Vietnamese beans reduce it.
- Wild fish is under most pressure. The marine fish stock in 2016 to 2020 was 22% below 2000 to 2005.
- Official food security counts volumes, not imports. The food-security resolution to 2030 has no import or protein indicator. A national protein balance is the first move (Moves for any 2050).
Context and supply: ten findings
The real non-animal protein economy is traditional. Soy milk, tofu and factory-made đồ chay (traditional vegetarian food) dominate. Modern plant-based meat is nearly absent. Every local product we audited with 10 g of protein or more per 100 g ran on imported soy or gluten (Alt protein today).
Cheap, but not unique, and not ready on rules. A regional study scores Vietnam 100 on labour and 0 on regulatory readiness. Labour and power are a small share of fermentation cost. Vietnam exports more than 2 Mt of cassava starch a year, mostly to China, but almost no glucose syrup (Region).
Rich in starch, poor in protein. Vietnam grows about 10 Mt of cassava roots and 43.5 Mt of paddy rice, and has a sugar surplus. Its protein-rich side streams, such as okara (soy pulp), are small. Cassava prices rose 50% to 90% in the year to April 2026 (Asset map).
Fermentation skills exist but are captive. Three amino-acid makers run almost all large fermentation. We found no contract manufacturing offer, no open pilot fermenter, no protein-isolate plant and no domestic textured-protein extruder. Frozen-food makers are ready co-packers (Industrial base).
Research sits next to alternative protein, not inside it. Vietnam is strong in yeast protein expression, stem-cell culture, koji genetics and microalgae. It is near zero in texturisation, fungal protein, precision fermentation and food cell lines. The main gap is shared infrastructure (Knowledge and talent).
Food rules have an open door but no gate. New protein foods are sold through self-declaration. But the law also treats unpermitted "new-technology food" as recallable, and there is no procedure to get permission. The new Food Safety Law, with a vote considered in May 2027, is a one-off window (Rules).
Capital is thin. No disclosed deal in these protein families in Vietnam has exceeded USD 5 million. Agritech venture capital in Vietnam was about USD 1 million in 2025 (Capital).
Capital, not operating cost, decides fermentation economics. Fungal protein made in Vietnam would cost USD 4,050 to 14,700 per tonne of protein, over half of it capital and maintenance. That fits a functional ingredient, not a bulk one. Textured plant protein is cheap to make, but Chinese product enters duty-free at about USD 1.29 to 1.38 per kg (Economics).
One technology family fits now: textured plant protein for food. Local legume and rice proteins and fungal foods such as koji are worth proving. Precision fermentation fits only as high-value products made under contract. Cultivated meat and seafood are research topics for this decade (Technology fit).
The calendar is front-loaded. Almost every relevant rule change lands in 2025 to 2027. The direction of the next decade will be visible within about 18 months (Rules, Outlook).
Demand: who would buy it
Part III (Protein diet to Demand sizing) asks who would buy new protein, in what form and how much. Surveys usually overstate behaviour, so we follow what people and firms actually do.
- A high-meat diet, still rising. Meat per person is projected to rise 25% by 2035, led by poultry. Adults already eat enough protein (Protein diet).
- Cheap protein sets the bar. Eggs and chicken cost VND 34,000 to 45,000 per 100 g of protein. Imported plant-based meat costs 12 to 18 times as much per gram of protein.
- Chay is the incumbent. Existing chay (vegetarian) days already avoid about 2.5% of meat protein (Chay baseline).
- People say yes and buy little. In an urban online survey, 74% would choose plant-based meat at the same price as meat. But only 3% eat it weekly. Plant milk is the one modern plant category at scale (Consumers).
- Blends are the entry route, and soy is the price to beat. Mixing soy into meat products saved money every year from 2019 to 2024. A new extender has to beat hydrated soy, and be named as an ingredient, never as độn (filler) (Frontier demand).
- Kitchens and factories before shelves. Traditional markets take 83% of food retail. Factory canteens serve up to 1.2 billion meals a year. Schools already have plant-protein days, but serve tofu beside meat (Channels).
- The clearest demand is business to business. Plant proteins shipped from China to Vietnam nearly tripled from 2019 to 2025, to about 54,000 t. They pay no duty, so a domestic maker has no tariff shelter (Business buyers).
- Export: a tariff edge, but falling demand for meat analogues. Vietnam enters the EU, the UK and Japan duty-free on lines where Thai and Chinese goods pay 6% to 13%. The export base is wrapped foods such as dumplings (Export demand).
- How much. On a benchmark path, these routes reach about 19,000 t of protein a year by 2035. About 11,000 to 17,000 t would be made in Vietnam, mostly replacing imported ingredients. It would displace about 0.2% of meat protein (Demand sizing).
Futures: 2035 and 2050
Part IV (Outlook to Vision 2050) never presents a scenario as a forecast.
- The pressures point to more imported protein. The working-age population peaks around 2035 to 2040, and African swine fever stays endemic. Agriculture must cut emissions, and the Mekong Delta keeps sinking. No official protein number reaches beyond 2030 (Drivers to 2050).
- People will eat more meat, and the imports behind it will grow. On trend, meat protein eaten rises from about 1.01 Mt in 2025 to 1.45 Mt in 2050 (our model). The imported soybean meal behind it rises from 7.2 Mt to about 10.4 Mt (Protein balance 2050).
- New protein adds a second source; it does not replace meat. Replacing 1 t of meat protein avoids about 4.5 t of soybean meal and 9.5 t of maize imports (our calculation). But at the scale demand supports, new protein displaces about 0.4% of meat protein in 2050. Better feed efficiency would save about twice as much soybean meal.
- Frontier technology works, but costs fall slowly. Fermentation costs fall 13% to 20% each time output doubles, far slower than solar power. Protein from renewable power made in Vietnam would still cost USD 3.3 to 6.5 per kg in 2050 (our estimate) (Frontier technology).
- Two questions decide most. Will imported protein come under chronic stress? Will Asia's fermentation market stay open, or fall under Chinese price dominance? Crossing them gives four worlds for 2050 (Scenarios 2050).
What to do
Seven plays; three lead under almost every weighting. A play is a business or programme someone could start, with a named product, buyer and partner (Plays):
- T1. Domestic textured plant protein for chay brands and frozen-food makers, replacing imports from China (0 to 2 years). Best owned by a chay maker, co-packer or distributor, and financed with debt.
- T4. A hub that turns cassava starch into fermentation sugar, next to starch factories (1 to 3 years; needs a committed fermenter).
- T5. Shared pilot and tolling fermentation, where firms rent time on someone else's fermenter (2 to 4 years). Most fermentation plays need it first.
Koji and fungal foods (T9) and mung bean and rice protein for plant drinks (T10) are worth proving. Precision fermentation fits only as contract-made, high-value ingredients (T7). Cultivated seafood is research for this decade (T8).
We also checked 170 published solutions, including the Good Food Institute's solutions database. None displaces T1, T4 or T5.
First public steps. The three highest-ranked policy options need no new agency (Policy options):
- A one-line clause in the new Food Safety Law letting the Government regulate "new foods and new food ingredients".
- A joint note from the agriculture and health ministries on the genetically modified organism (GMO) status of purified precision-fermented proteins.
- A proportionate novel-food procedure in the law's implementing decree.
The larger outlays are a shared food-grade pilot fermentation line and a protein-quality laboratory.
Product targets. Ten product profiles say what a product has to be for the demand that exists. Twenty cheap demand moves each name who would pay (Demand to frontier).
Seven moves pay off in all four worlds. Most are public, cheap and can start in 2026 to 2028 (Moves for any 2050): a national protein balance; a new-food route with deadlines; shared food-grade pilot capacity; a protein-quality laboratory and price series; rules for crop residues and a by-product atlas; climate-proof siting; and process-engineering skills.
Checked against the actors
We turned the stated priorities of 80 named actors into 118 decision questions. Follow-up research then changed these points (Actor check):
- Schools fill their plant-protein days with tofu, and menu software still sets an animal-protein floor of 48%.
- Textured protein lines are cheap (Chinese lines from USD 11,000 to 100,000). The soy-flour price decides the margin.
- Blends: buyers punish hidden filler, not a declared ingredient.
- Funders: only 4 of 19 demand moves have a funder whose rules fit and that accepts Vietnam.
- Firms: none of the four Vietnam-based meat-analogue makers we tracked is active.
For each reader
- Investors: size ingredient businesses on the imports they can replace, not on consumer forecasts. If a thesis rests on food security, check that the inputs are domestic (Investors).
- Policy makers: act on the Food Safety Law window, and share capital costs rather than grant tax holidays. Publish a national protein balance (Policy makers).
- Startups: start where the route is clear (textured protein, koji), and sell first as an ingredient, store-brand or co-packing supplier to incumbents (Startups).
- Manufacturers: you are the first customer for textured protein and the natural owner of a textured-protein line. Test extenders against your own soy recipe (Food manufacturers).
- Researchers: the most useful gaps are protein-quality data, duckweed safety, heat-tolerant strains on cassava, and how much of the soy in tofu is Vietnamese (Research).
- International bodies: support novel-food drafting, a food-grade pilot line and finance for first plants. Do not fund alternative protein as pandemic prevention (International).
What would change our view
- The Food Safety Law passes without a new-food route, or with one that has no deadlines.
- Chinese textured soy or microbial protein lands in Vietnam below the cost of Vietnamese production.
- An incumbent fermenter offers tolling.
- A Vietnamese study shows repeat buying of a modern analogue or blend at a premium over meat.
- Food-grade soy flour is quoted below about USD 700 per tonne delivered, or three chay makers sign conditional orders.
- A study measures lower disease or antimicrobial-resistance risk from replacing meat. We found none.
The eighteen unknowns that matter most, and who can answer each, are in Unknowns.
About this report. AltProtein Vietnam, version 1.0, November 2026. It rests on thirteen rounds of desk research (2,669 sources), a retail audit in Nha Trang and Ho Chi Minh City, reproducible protein balance and demand models, and a check against 80 named actors. Each claim carries an evidence tag and a confidence level (How to read). AltProtein Vietnam used Claude Opus 5.5, an AI model made by Anthropic, for the research, drafting and calculations, and is responsible for the content. Parker Vu of AltProtein Vietnam led the project. Matt Tom of the Global Forum for Plant-Based and Alternative Proteins mentored it, under the Sentient Futures Project Incubator 2026. This is scoping research, not investment, legal or engineering advice.