On 4 November 2026 the head of Indonesia’s National Nutrition Agency (BGN), Sudaryono, described a rule change he called a policy “miracle” on a sheet of paper: milk served in the free school-meal programme, Makan Bergizi Gratis (MBG), must now contain at least 80% fresh milk, up from a requirement that previously allowed most of it to be imported milk powder. In the same speech he acknowledged that domestic fresh milk covers only about 20% of national need [1]. In April, BGN’s own milk adviser had said that even raising the fresh share from 20% to 30% might only be possible around 2027 or 2028 [5].
The price did not move with the rule. A kitchen pays a set Rp3,000 for a 125 ml portion and Rp4,500 for 200 ml, delivered [2]. When nobody can undercut the price, a new milk supplier has only one thing to sell: milk that actually arrives, meets the rule and is permitted. This article asks whether a regional dairy plant beside Java’s dairy farms could build a business on that.
How a cup of school milk reaches a child
MBG is cooked and handed out by local kitchens called SPPG. In April 2026 about 20,000 operating kitchens served roughly 60 million recipients, about 3,000 each, and milk appeared on the menu twice a week [5]. Each kitchen buys its ingredients with programme funds. For milk, BGN allows only plain full-cream pasteurised, UHT or sterilised milk, with no added sugar, preservatives or flavouring, at least 80% fresh milk and a distribution permit from the food and drug regulator, BPOM. Sweetened condensed milk and milk-flavoured drinks are banned. Pasteurised milk must be kept cold from the plant until a child receives it [2].
Upstream, the milk comes from cows kept mostly by smallholders: about 540,657 dairy cows, more than 90% of them on small farms [6], producing about 808 million litres a year, roughly 2 million litres a day [4][5]. East Java alone supplies almost 60% [15]. Farmers sell through cooperatives, and the processors’ association says 90% of smallholder milk goes to processing companies that turn it into UHT and pasteurised milk [7]. Schools have mostly received UHT cartons: they keep for months and can travel in ordinary trucks without refrigeration, which suits a country of islands [4].
How school milk reaches an MBG kitchen today
- 1. Smallholder farms
About 540,657 dairy cows, over 90% on small farms; about 2 million litres a day - 2. Cooperatives
Collect and chill milk; 90% of smallholder milk goes on to processors - 3. Processors or cooperative rooms
UHT lines shared with commercial runs, or small manual pasteurising rooms - 4. MBG kitchen (SPPG)
Buys permitted plain milk at a set Rp3,000 per 125 ml, delivered - 5. Pupils
Milk twice a week; cold chain required for pasteurised milk
The shortage moved from cartons to cows
Until the new rule, the bottleneck was the carton line. UHT lines also run commercial products, so school milk queued behind them; the shortage was of plain full-cream UHT, not of milk as such [5]. The industry ministry put 2026 demand at 4.8 billion packs against 2.39 billion packs of national capacity in the small sizes [6]. Processors responded: Ultrajaya, a listed dairy company, opened a Rp1.14 trillion plant at Cibitung near Jakarta in December 2025 with three UHT lines for 125 ml and 200 ml packs [8].
The 80% rule moves the constraint upstream. If each pack must itself contain 80% fresh milk, 4.8 billion 125 ml packs need about 480 million litres of fresh milk a year, close to 60% of everything Indonesian cows produce (our arithmetic from [4] and [6]). That reading follows CNA’s report of the product rule [2]. Another report quotes Sudaryono saying that 80% of the supply handed to children must be fresh milk from local farmers [3], which could be counted across a kitchen’s or a supplier’s total rather than per pack. We do not know which counting BGN enforces, and the size of the squeeze depends on it.
The rule also lands on a programme that keeps changing. In July 2026 BGN asked kitchens to stop serving branded packaged milk, and the processors protested [7]. Milk buying was then halted and reopened from 1 November under the new rule [3]. From 2 November, 6,097 kitchens were suspended for failing to file daily spending reports, until they comply [13]. And where milk is hard to get, BGN lets kitchens replace it with other animal protein [5].
Our inference from these facts: kitchens far from dairy farms will mostly substitute or wait for UHT, while kitchens near Java’s dairy clusters sit next to the milk but must compete with processors for it. Those near-farm kitchens are where a new supplier could be useful, if they actually go short.
Who already sells milk to the kitchens
National processors are the default. Ultrajaya runs dedicated school lines and its own herds of about 7,000 cows [8]. Others make milk under someone else’s brand: in 2022 PT Sukanda Djaya, part of Diamond, agreed to produce about 18 million litres of UHT milk branded for Jakarta’s food company Food Station, using milk from West Java cooperatives [10]. Kitchens choose UHT because it is easy to store and needs no cold chain.
Cooperatives and small processors have begun selling directly. Koperasi Margo Makmur Mandiri in Brau, Batu, East Java, delivers pasteurised cups to 10 kitchens in Malang, Batu and Mojokerto, 1,000 to 3,500 cups per kitchen per delivery, one to three times a week. It asks for a 30–40% down payment, and everything from pasteurising to filling cups is still done by hand while it tries to build a modern production house [9]. Sweet Sundae, a small Yogyakarta processor, supplied five Central Java kitchens in 2025 with 12,000 cups of 100 ml per pickup, reported revenue of up to Rp1 billion, and insisted on cash up front after payment delays under the early reimbursement system [11]. In Cimahi, West Java, a farm under Koperasi Jagri had 150 litres for deliveries that needed 370 litres; the raw milk was pasteurised inside the kitchen [12]. KPBS Pangalengan, south of Bandung, runs its own packaging unit [6].
The state is building more of these. The agriculture ministry promotes Dapur Susu Indonesia (DASI) units, small cooperative plants costing under Rp5 billion that can each supply 5 to 10 kitchens; the industry ministry offers to reimburse up to 35% of processing investment [6]; and the cooperatives ministry plans pasteurisation machines and cold chains so that cooperatives can supply kitchens directly [15]. A packaging maker has proposed shared co-packing plants at dairy centres for the same reason [4].
Each alternative leaves something. UHT travels well but competes for line time and now for fresh milk. Village rooms are close to the farms but small, manual and tied to one cooperative’s milk. A DASI unit is sized for a handful of kitchens and is still being rolled out. Between a hand-filled village room and a national factory there is a gap in scale: a plant that can serve dozens of kitchens in one dairy region from several cooperatives’ milk. That gap is our hypothesis, not an established shortage.
A plant sized between a village room and a national factory
The business we examine, the Cluster Milk Plant for MBG Kitchens, is a new regional company with one permitted plant in a Java dairy area, for example around Malang–Batu, Boyolali or south Bandung. Each day it collects chilled milk from several cooperatives, pasteurises or sterilises it, fills plain 125 ml and 200 ml cups and delivers them on fixed routes to kitchens within cold-delivery range on their milk days. With each delivery it hands over the product permit, the fresh-milk share and delivery temperatures, records that help a kitchen show it followed the milk rule at a time when BGN is suspending kitchens over incomplete daily spending reports [13].
The kitchen head chooses the supplier and pays from programme funds at the set price, so the company cannot win on price. It would win on reliability and access: milk that arrives on the menu day in the quantity the kitchen needs, from a supplier large enough not to run out and documented enough to pass the new rule. Sweet Sundae and the Brau cooperative already run small versions of this; the new company’s difference is drawing milk from several cooperatives and running at a scale that neither a single farm nor a hand-run room reaches. It would sell first to kitchens these suppliers cannot cover. Kitchens can be found and visited one by one within a radius; BGN is also building an online marketplace through which kitchens must buy seven commodities, including eggs, chicken, rice and cooking oil, with a full launch targeted for December [3], though milk’s place in it is not confirmed. To protect itself against late payments, it would ask for part payment up front, as the Brau cooperative does [9].
Cooperatives would supply it for the reason they supply anyone: a buyer paying at least what processors pay. Whether it can get milk that processors already have under contract is one of the questions that decide the business.
Where Rp3,000 a cup goes
The one sourced cost is the milk itself. A Sumedang farm said its milk sold for about Rp7,000 a litre before MBG and Rp10,000 after [12]. At those prices, 125 ml of fresh milk costs Rp875–1,250. If cups, processing, permits, chilled delivery and spoilage together cost Rp500–850 per cup, which is our assumption and not a sourced figure, the plant keeps Rp900–1,625 of each Rp3,000 cup, a contribution of 30–54% before overheads. That is an illustration of what must hold, not an estimate.
Where Rp3,000 a cup could go in a regional plant
| Item | Per 125 ml cup | Status |
|---|---|---|
| Set delivered price | Rp3,000 | Evidence |
| Fresh milk at Rp7,000–10,000 a litre | Rp875–1,250 | Evidence (one farm) |
| Cup, processing, permits, chilled delivery, spoilage | Rp500–850 | Assumption |
| Left to cover plant, staff and profit | Rp900–1,625 (30–54%) | Illustration |
The deciding relationship is between the milk price and the fixed portion price. Every Rp1,000 a litre the plant must pay above the farm price takes Rp125 off each cup. If competition with processors pushed milk to Rp14,000 a litre, the contribution in our illustration would fall to Rp400–750 a cup.
The buyer’s cheapest alternative costs the same: national UHT at the same Rp3,000, which spares the kitchen any cold storage. Before the permit rule, some kitchens pasteurised raw farm milk themselves [12]. The requirement for a BPOM-permitted product [2] appears to rule that out, though we have not confirmed how it is applied. A cooperative could own its own capacity through a subsidised DASI unit at under Rp5 billion for 5 to 10 kitchens [6], so the plant must serve many more kitchens per rupiah of plant than that to justify existing.
How big is the first market? West, Central and East Java had 15,653 operating kitchens in June 2026 [14]. If suspensions like November’s removed about a fifth, and a quarter to a third of the rest lie within cold-delivery reach of a dairy cluster (both our assumptions), some 3,000–4,200 kitchens could be served, buying perhaps Rp0.9–2.3 trillion of milk a year. One plant serving 30 to 60 of them would earn roughly Rp9–32 billion a year in revenue. That could support a dedicated regional company, with further plants in other dairy areas, and sales to boarding schools, the armed forces or retailers to keep the plant running when the programme pauses. The working is in the Research Notes.
The subsidised neighbour and the single payer
The strongest case against the plant is that the government wants the same kitchens supplied by cooperatives. Officials have said they want pasteurised school milk produced by farmers through cooperatives [15], and DASI units arrive with subsidies the new company would not get [6]. If BGN steers kitchens to cooperative units, a private plant may be left with only overflow. One defence is to make cooperatives partners rather than rivals: buy their milk and offer to process for them. That is the second business in this investigation, Cooperative Toll Processing, described in the notes.
Processors are the other threat. They already buy 90% of smallholder milk [7], so they could tie up the milk the plant needs, and the industry ministry wants them to mentor cooperatives into processing [6]. If national UHT reliably reaches kitchens near the farms after 1 November, a kitchen gains nothing by switching to a pasteurised cup that needs a refrigerator.
Then there is the single payer. Within four months milk was discouraged, halted, reopened and its rules rewritten, and thousands of kitchens were suspended [3][7][13]. A plant bought for Rp5 billion or more cannot sit idle for months. What the company could hold, if it gets going, is milk contracts with several cooperatives and dense delivery routes in one region. Both are repeatable in another dairy area, but neither keeps out a well-funded processor that decides to compete for the same kitchens.
A four-week test before anyone builds a plant
The question that decides everything is cheap to answer: do kitchens near a dairy cluster actually go without compliant milk after 1 November? Take 20 to 40 kitchen heads within about 50 km of one Java cluster, record for four weeks what milk arrived on each milk day and from whom, and then offer the kitchens that report gaps a same-price trial supply, initially from an existing permitted producer. If fewer than one kitchen in five reports missed or substituted milk days, or fewer than one in ten accepts the trial, the opening is not there. A second question needs only a letter: how BGN counts the 80%, per pack or across a kitchen’s supply.
What is documented: the set price, the product rule, the doubled fresh-milk requirement against limited domestic milk, the policy upheavals and the small suppliers already selling to kitchens. What we infer: that the squeeze falls hardest near the farms, where processors and kitchens compete for the same milk. What is unknown and decisive: whether kitchens there go short, and whether a plant’s cost per cup leaves a margin. Kitchen heads, cooperative managers and anyone who has run a mid-sized dairy plant could answer these quickly.
Research Notes
The sources, calculations and evidence records behind this investigation. Numbered citations in the article link here; ↩ returns to the text.
1. Agrina — “Bukan Sekadar Nutrisi, Program MBG Jadi Instrumen Strategis Intervensi Pasar” (4 November 2026) ↩
Full text read. Source article. Supports: BGN head’s statement that MBG milk must contain at least 80% fresh milk (previously 30% local, 70% imported powder in this report’s wording) and that domestic fresh milk meets about 20% of national need; BGN’s online purchasing for eggs. Limitation: a speech report; “about 20%” is the official’s figure.
2. CNA Indonesia — “Susu kental manis dan minuman rasa susu dilarang masuk menu MBG” (September 2026) ↩
Full text read. Supports: allowed products (plain full-cream pasteurised, UHT, sterilised), at least 80% fresh milk per product, no added sugar or additives, BPOM distribution permit, cold chain for pasteurised milk, set prices Rp3,000 (minimum 125 ml) and Rp4,500 (200 ml) delivered to the kitchen; milk as a supplement to other protein. A photo caption refers to a mass poisoning in Gemolong, Sragen, attributed to an MBG menu; it does not say milk was involved, so the article does not use it.
3. JournalArta — “Susu MBG kembali 1 November, 80 persen wajib lokal” (24 October 2026) ↩
Full text read. Supports: milk buying halted and reopened from 1 November by an internal memo; Sudaryono’s wording that 80% of the supply must be fresh milk from local farmers; BGN marketplace for seven commodities, soft launch planned and full launch targeted for December. Limitation: the supply-share wording differs from source 2’s per-product rule.
4. Sonora — “Susu MBG, Kemasan Aseptik, dan Peluang Kolaborasi” (March 2026) ↩
Full text read. Supports: national production about 808 million litres a year (BPS 2024 cow numbers, cited by Prof. Epi Taufik); UHT used because of long shelf life without cold chain; the earlier 20% fresh-milk specification; LamiPak’s proposal for shared co-packing plants at dairy centres. LamiPak is a packaging supplier with an interest in the proposal.
5. RMOL — “Indonesia Butuh Tambahan 144 Juta Liter Susu Segar per Tahun buat MBG” (19 April 2026) ↩
Full text read. Supports: domestic production 808,352,840 kg a year (about 2 million litres a day); 20,000 kitchens serving about 60 million recipients; milk twice a week; 20% minimum fresh share then, with 30% possible around 2027–2028; UHT line queues and capacity of 4.32 billion packs; substitution permitted where milk is unavailable.
6. Katadata — “MBG Butuh 4,8 Miliar Kemasan Susu pada 2026” (2 June 2026) ↩
Full text read. Supports: 4.8 billion packs needed in 2026 against 2.39 billion of capacity; up to 35% investment reimbursement; 540,657 dairy cows, over 90% on smallholdings; DASI units under Rp5 billion for 5–10 kitchens; KPBS Pangalengan’s packaging production unit (photo caption).
7. Kontan — “AIPS: Larangan Susu Kemasan Bermerek di MBG Berisiko Ganggu Serapan Susu Peternak” (24 July 2026) ↩
Full text read. Supports: 90% of smallholder milk goes to processors (association’s statement); BGN’s July request to stop branded packaged milk.
8. Detik — “Susu Ultra Gelontorkan Rp 1,14 Triliun buat MBG” (24 January 2026) ↩
Full text read. Supports: Ultrajaya’s Rp1.14 trillion Cibitung plant from 8 December 2025 with three UHT lines for 125 ml and 200 ml; about 7,000 cows on two farms. No margin figure appears; a margin figure seen in search results was not used.
9. JatimTIMES — “Koperasi Susu Brau Suplai Ribuan Cup Produk Pasteurisasi ke SPPG Lintas Kota” (June 2026) ↩
Full text read. Supports: 10 kitchens supplied; 1,000–3,500 cups per kitchen per delivery, one to three times a week; 30–40% down payment; manual production; a planned modern production house.
10. Kontan — “Food Station jalin kemitraan produksi susu UHT dengan Sukanda Djaya” (April 2022) ↩
Full text read. Historical: Sukanda Djaya (Diamond) to make about 18 million litres of FS-branded UHT milk for Food Station, using milk linked to KPSBU and GKSI West Java. Shows contract manufacturing under a buyer’s brand; not MBG-specific.
11. Detik — “UMKM asal Yogya raup Rp 1 M dari pasok susu program MBG” (21 August 2025) ↩
Full text read. Supports: Sweet Sundae supplying five Central Java kitchens, 12,000 cups of 100 ml per pickup twice a week, eight more kitchens planned, revenue up to Rp1 billion, cash up front, letters of intent and five-year agreements. Its stated 4.5 tonnes a day from 97 cows looks implausible and is not used.
12. Koran Jakarta / Antara — “Dampak Program MBG, Peternak Sapi Rasakan Peningkatan Keuntungan” (17 April 2025) ↩
Full text read. Supports: milk price about Rp7,000 a litre before MBG and Rp10,000 after at Nusa Dairy, Sumedang; 150 litres available against 370 litres needed per delivery for 3,500 students; raw milk pasteurised in the kitchen; sales through Koperasi Jagri to Cimahi kitchens. One farm’s price, 2025.
13. iNews / RCTI+ — “BGN Suspend 6.097 SPPG Mulai 2 November” (31 October 2026) ↩
Full text read. Supports: 6,097 kitchens suspended from 2 November for not filing daily spending reports (LPDH), reinstated once compliant.
14. GoodStats — “10 Provinsi dengan SPPG Terbanyak Juni 2026” (BGN data, 9 June 2026) ↩
Full text read. Supports: operating kitchens in West Java 6,721, Central Java 4,592, East Java 4,340 (total 15,653).
15. RRI — “Pemerintah Perankan Peternak Rakyat Jadi Rantai Pasok Program MBG” (30 July 2026) ↩
Full text read. Supports: cooperatives ministry plans to integrate dairy cooperatives into MBG with pasteurisation machines and cold chains; East Java’s almost 60% share of national milk (cooperative chair’s statement).
Competitor and substitute table
| Alternative | Offering and coverage | Price and terms | What it leaves | Source |
|---|---|---|---|---|
| Ultrajaya | School UHT 125/200 ml, national, Cibitung plant, own herds | Set Rp3,000/125 ml delivered | Lines shared with commercial runs; needs fresh milk under the new rule | [2], [5], [8] |
| Sukanda Djaya (Diamond) and other processors | UHT, including production under a buyer’s brand | Not found | Same line and milk constraints | [10] |
| Koperasi Margo Makmur Mandiri (Brau) | Pasteurised cups, 10 kitchens in Malang Raya and Mojokerto | 30–40% down payment; delivery fee or pickup | Manual production; one cooperative’s milk | [9] |
| Sweet Sundae | Plain 100 ml milk, five Central Java kitchens (2025) | Cash up front; five-year agreements | Small scale; selective about kitchens | [11] |
| Koperasi Jagri / Nusa Dairy | Raw milk to Cimahi kitchens (2025) | Rp10,000/L farm price | 150 L against 370 L needed; kitchen pasteurisation | [12] |
| KPBS Pangalengan | Cooperative with a packaging unit | Not found | Kitchen count not verified (a search result reported about 50) | [6] |
| DASI and ministry-equipped cooperatives | Subsidised small plants for 5–10 kitchens each | Under Rp5 bn per unit; up to 35% reimbursement | Prototype stage; small per unit | [6], [15] |
| No milk / other protein | Kitchen substitutes where milk is unavailable | No milk cost | Children miss milk | [5] |
Evidence ledger
Observed: set price and product rule [2]; 80% requirement and 20% domestic coverage [1]; supply-share wording [3]; halt, reopening, suspension and the July stop [3], [7], [13]; small suppliers’ scale and terms [9], [11], [12]. Inferred: that near-farm kitchens face competition for fresh milk; that BPOM permits exclude in-kitchen pasteurised raw milk. Speculative: that kitchens near clusters go short after 1 November and would switch at the same price. Competing explanation: national UHT and subsidised units may cover these kitchens adequately, and BGN may count the 80% loosely.
Economics workings
Price: Rp3,000 per 125 ml delivered [2] (evidence). Milk: 0.125 L × Rp7,000–10,000/L [12] = Rp875–1,250 (evidence for one farm’s price; the plant may pay more, weakest input). Cups, processing, permits, chilled delivery and spoilage: Rp500–850 per cup (assumption). Cost per cup: Rp1,375–2,100. Contribution: Rp900–1,625 (30–54%) before plant capital, staff and overheads (illustration). Sensitivity: milk at Rp14,000/L gives Rp1,750 milk cost and Rp400–750 contribution.
Fresh-milk demand under the per-pack reading: 4.8 billion packs [6] × 0.125 L × 80% = 480 million L, against about 808 million L produced [4], [5] ≈ 59%.
Beachhead: 15,653 kitchens in three Java provinces [14] (base) → about 12,000 after removing about a fifth for suspensions (assumption informed by 6,097 suspended nationally [13]) → 25–35% within cold-delivery reach of a dairy cluster (assumption) = 3,000–4,200 kitchens → × 100,000–180,000 cups a year each (assumption: 1,400–2,250 cups per milk day, consistent with 1,000–3,500 per delivery at Brau [9] and about 3,000 recipients per kitchen [5], over 72–80 milk days) × Rp3,000 = Rp0.9–2.3 trillion a year. One plant serving 30–60 kitchens (assumption): 3.0–10.8 million cups = Rp9.0–32.4 billion revenue; illustrative contribution Rp2.7–17.6 billion. Volume 375,000–1,350,000 L a year, roughly 1,900–6,800 L per operating day over about 200 days.
Strategy details
Cluster Milk Plant for MBG Kitchens (featured). Provider: a new regional company with a permitted plant (BPOM, veterinary hygiene certificate NKV, halal). Partners: several cooperatives (milk at or above processor prices), cup suppliers. Resources: plant (benchmark: DASI under Rp5 billion for 5–10 kitchens [6]; a larger plant would cost more, not sourced), chilled fleet, working capital for payment delays and programme pauses. Decisive assumption: kitchens within reach of a Java cluster lack reliable compliant milk after 1 November and would switch at the same price. Test (matters): four-week milk-day log at 20–40 kitchens within 50 km of one cluster, then a same-price trial; falsified if fewer than 1 in 5 report gaps or under 10% accept.
Cooperative Toll Processing (peer). A permitted plant processes and packs mid-sized cooperatives’ milk for a fee per litre, and the cooperatives sell to kitchens under their own name, keeping more of the Rp3,000 than a raw-milk sale at Rp7,000–10,000 a litre without building a plant. Contract manufacturing under a buyer’s brand exists in Indonesian dairy [10], and a packaging maker has proposed co-packing at dairy centres [4]; no independent toll processor serving MBG cooperatives was found, and no fee is sourced. Competes with subsidised DASI units and processors’ mentoring [6]. Decisive assumption: cooperatives without plants would pay a toll fee rather than wait for subsidised machines. Test (business): written toll terms to five West or East Java cooperatives; falsified if none commits volume at a fee covering conversion cost. It can also become a revenue line of the featured plant.
Routes considered
Merged into Cooperative Toll Processing: tolling alliances, toll networks, modular toll units, permit-backed toll networks, a digital tolling orchestrator, slotted pasteurisation pods. Combined into the featured plant: milk-day scheduling and proof of delivery, contracted rooms as overflow capacity, pooled milk buying, chilling hubs, compliance logs and route planning, other institutional buyers, staggered collection windows. Not pursued: mobile pasteurisation vans (no permit route or cost), hub pickup and thermal lockers (shift cold-chain risk or add unfunded assets), payment ledgers, gateways, escrow and advances (BGN is building its own purchasing system, and lending against a payer that suspends kitchens has no evidenced demand), freshness-linked pricing (conflicts with the set price), cooperative secretary franchises and credit clearinghouses (no payer).
Open questions
How BGN counts the 80% (BGN’s milk team). Whether near-farm kitchens miss milk days after 1 November (kitchen heads). What milk cooperatives can sell outside processor contracts, and at what price (cooperative managers). Cost per cup of a mid-sized plant, and whether milk is included in BGN’s marketplace (former dairy plant operators; BGN).
Leave a Reply