In October, Jakarta tofu maker Sudiarto was still buying American soybeans at prices he found painfully high. He remembered fetching local beans from villages years earlier, but also remembered the dirt that came with them. The imported beans were clean, and he knew how they behaved in production. His account, reported by detikFinance, explains why replacing imports takes more than finding a cheaper sack. [6]
A new local supplier would have to lower the cost of the tofu a maker can actually sell. A small reduction in output can swallow a bean-price discount. That leaves a specific business worth investigating in Central Java: buying suitable local lots, preparing them when necessary and combining nearby makers’ orders into economical deliveries.
Why the familiar beans keep the order
Tofu makers extract soy milk from beans, cause it to set and press it into food. Both the beans and the processing affect the result. A 2016 Lampung study, published in 2019, compared four domestic varieties with imported beans across four makers; its abstract gives no basis for assuming that domestic beans automatically produce more tofu. [8]
The owner therefore buys more than kilograms. Cleanliness, acceptable output and dependable replacement supplies affect what the purchase is worth. Sudiarto’s recollection is evidence of his experience, rather than a finding that all local beans are dirty or unavailable today. [6]
The spending pressure is substantial. DetikFinance reported October prices of Rp11,300–11,500 per kilogram, against earlier prices below Rp10,000. For an illustrative workshop using 100 kilograms a day, even a Rp1,300 increase means Rp130,000 more daily spending on beans. In November, the national tofu-and-tempe cooperative association told Liputan6 that exchange-rate pressure was raising costs while customers’ purchasing power limited food-price increases. [6] [1]
There are already several ways to buy. FKS Food and Agri describes domestic and overseas sourcing combined with warehousing and transport. A September 2025 government release named FKS Multi Agro, cooperatives and other traders among participants absorbing local harvests in Pati. Local sourcing is already part of the established trade. [5] [2]
Nor is a shorter route new. A Grobogan study published in 2017 documented collectors supplying processors directly and drying beans before resale. Its historical evidence shows why simply removing a trader does not remove the work that trader performs. Today’s route costs still need checking. [11]
One supplier would take responsibility for the lot
The proposed business, Clean local soybeans on pooled delivery routes, would start with tofu makers near suitable Central Java harvests. A regional trader would ask owners what they buy, inspect their invoices and identify which beans produce food their customers accept. It would then buy compatible lots from farmers or collectors, measure moisture and cleanliness, and pay existing contractors to clean or dry only the lots that need it.
Orders from several nearby workshops would share deliveries. Already-compliant beans could go straight to buyers when orders and collection times align. The company would keep track of each lot and agree in advance how grade shortfalls change the invoice. The owner would pay per accepted kilogram; the supplier would earn the difference between that payment and its purchase and service costs. A promise about the delivered grade is manageable in principle. A blanket promise about tofu output would also expose it to processing decisions it does not control.
The reason to switch would be a lower cost for comparable saleable tofu, with less purchasing work for the owner. The saving would have to survive preparation, lost weight, freight and any change in payment terms. This is an opening to test against existing suppliers’ customers; research has not established a general failure by those suppliers.
Finding the first buyers would mean owner visits, invoice comparisons and small production comparisons, followed by repeat paid orders. Farmers and collectors would need competitive prices and reliable payment. Contractors would need enough paid work to make the route attractive. No cooperative access or partner agreement is assumed. A plausible founding team would combine regional commodity purchasing, food-input quality control and working capital, initially hiring preparation and transport rather than buying a fleet.
A small output loss can erase the discount
Consider an assumed delivered offer of Rp11,000 per kilogram. Bapanas reported some Java supply at Rp11,300 in early October. That is a useful reference for the calculation; the proposed buyers still need same-day quotes. [4]
At those prices, the new beans must produce more than 97.35% as much comparable saleable tofu per kilogram as the alternative merely to retain a bean-cost advantage. A 2.65% output shortfall erases the Rp300 discount before any extra processing expense. The comparison must hold food quality and moisture comparable: extra water in the finished tofu is not an economic improvement.
The supplier faces a second narrow allowance. In an illustration, selling at Rp11,000 and retaining Rp400 per kilogram before fixed overhead leaves Rp10,600 for all variable costs. Bapanas reported a national dry local-bean farmgate average of Rp9,006 in September 2025. With an assumed 2% weight loss, that becomes roughly Rp9,190 per delivered kilogram, leaving Rp1,410 for preparation, bags, transport, finance, selling and other variable costs. These different-date references describe a budget to test, not a buying-and-selling profit available today. [2]
The same release reported a producer reference of Rp10,775. At that purchase level and the same assumed loss, beans alone cost about Rp10,995 per delivered kilogram. The illustrative margin has gone before cleaning or delivery. The opportunity cannot rest on assuming farmers will always sell at a depressed harvest price. [2]
For scale, the Agriculture Ministry’s publication summary records 230,210 tonnes of national soybean output in 2024, with 30.80% from Central Java: about 70,905 tonnes. If 10–20% were suitable, available to buy and close enough to serve, and a supplier won 10–20% of that flow, annual sales would be roughly 709–2,836 tonnes. At the assumed price, that is Rp7.8 billion–31.2 billion in resale revenue. Those shares are assumptions about historical supply, not evidence of current orders. [9]
Revenue would mostly pay for beans. At the illustrative Rp400 left per kilogram after variable costs, the range leaves about Rp284 million–1.13 billion before fixed overhead. An assumed Rp600 million annual overhead needs 1,500 tonnes to break even: the low-volume case fails. A regional business is conceivable at sufficient retained volume; repeating successful routes into other harvest-linked clusters offers a growth path, provided each route keeps its economics.
Get the competing quotes before buying equipment
The strongest objection is straightforward: an existing collector may already do this cheaply, and FKS or a local cooperative could match the grade, schedule or price. Better-filled routes and trusted relationships might improve repeat business, but they offer no demonstrated protection from imitation. Off-season purchases, rejected lots, late customer payments or half-empty deliveries could consume the spread.
Owners can also buy directly. Their own time, transport, preparation and stock losses belong in that comparison. Where publicly supported supply is available, it belongs there too: Bapanas has reported freight assistance and describes price intervention as conditional. A private supplier must beat the owner’s cheapest adequate option on comparable terms. [3] [4]
Investigate the pooled-route supplier first. Collect current, grade-matched supplier and contractor quotes alongside owners’ invoices and a short record of the work involved in self-buying. Only if those figures leave room for both parties should a small paired tofu batch test the output. Stop if no repeatable price gives the maker a worthwhile saving and the supplier enough contribution at attainable volume.
The documented finding is that makers face costly inputs while existing suppliers already perform useful quality and delivery work. Our inference is that some nearby local routes could perform that work more economically. Whether they can do so repeatedly, without reducing saleable tofu output, remains the decisive unknown. Owners willing to share invoices and batch records, and collectors or hauliers able to quote actual routes, could answer it.
Research Notes
The sources, calculations and evidence records behind this investigation. Numbered citations in the article link here; ↩ returns to the text.
Sources and provenance
[1] Ujian Ketahanan Industri Nasional di Tengah Gejolak Rupiah ↩
Original source. Liputan6, November 2026 reporting, including interviews dated 4 November. Saved article body read in full; image captions and recommendations excluded from the argument. Supports attributed cost and pass-through pressure. Exchange-rate and aggregate import figures were not independently checked and are not used here.
[2] Bantu Petani Kedelai Lokal, Pemerintah Bareng Swasta dan Asosiasi Serap Hasil Panen Produksi Dalam Negeri ↩
Original source. Bapanas, 17 September 2025; full official library text read. Reports the 16 September national dry-local-bean price, producer reference, harvest season and named Pati absorption participants. These historical national prices are not procurement quotes for the proposed route. Reference-price discussion is attributed reporting, not a new legal assessment.
[3] Harga Kedelai Terkendali, Distributor dan Importir Harus Patuhi Harga Acuan ↩
Original source. Bapanas, 14 April 2026; full official library text read. Reports regional prices, 2025 freight assistance of 120,800 kg and projected 2026 consumption of 2.74 million tonnes. Forecast consumption is used only to value the broad customer job. Support availability for any proposed buyer is unknown.
[4] Harga Kedelai di Tingkat Perajin Tahu dan Tempe Terpantau Stabil, SPHP Kedelai Dilakukan secara Terukur ↩
Original source. Bapanas, 8 October 2026; saved official release read. Cites Gakoptindo prices for 7 October, including some Java beans at Rp11,300/kg, and explains conditional intervention. The regional low is not a named cooperative’s quotation or a matched current offer; payment, grade and delivery details are missing.
[5] Distribution and Logistics — FKS Food and Agri ↩
Original source. FKS Food and Agri, undated corporate distribution-and-logistics page, saved text read. Describes local/global procurement, wholesalers and small enterprises, warehousing and land transport. This establishes the advertised model; it is not an independent audit of service or margins.
[6] Harga Kedelai Sangat Mahal, Kenapa Perajin Tahu Masih Impor dari AS? ↩
Original source. Ignacio Geordi Oswaldo, detikFinance, 5 October 2026; full article body read. Reports prices and Sudiarto’s Jakarta purchasing experience. His memories of local supply and cleanliness are attributed personal recollection. His colloquial explanation in terms of starch is not adopted as food science or a national quality finding.
[7] Kedelai Impor Naik, Perajin Tempe di Grobogan Pilih Perkecil Ukuran
Original source. Nur Soli, MetroTV, 10 April 2026; saved article body read. Reports Grobogan tempe makers reducing portion sizes amid bean and packaging increases. It supports limited pass-through in those cases, not tofu-output performance. This is related evidence for the spending problem, rather than a second target market already validated.
[8] Evaluasi Mutu Tahu dari Beberapa Varietas Unggul Kedelai Dibuat oleh Beberapa Pengrajin ↩
Original source. Alvi Yani and Joko Susilo Utomo, Buletin Palawija 17(1), 2019; 2016 Lampung research. Abstract and introductory/process passages reopened in the indexed original PDF during writing after a direct request failed. Supports the processing explanation and the need to compare bean and maker effects; no result-table figures are used.
[9] Analisis Kinerja Perdagangan Kedelai Tahun 2025 ↩
Original source. Agriculture Ministry, Analisis Kinerja Perdagangan Kedelai Tahun 2025, publication summary dated 29 December 2025. Summary read; the 60-page report was not. Supplies 2024 dry-bean output and Central Java share. It does not establish present free stock, compatible grade, route access or demand.
[10] Kedelai Lokal — Kedelai Indonesia / PT Attempe Healthy Natural
Original source. Kedelai Indonesia / PT Attempe Healthy Natural, product listing posted 19 May 2025, saved during evaluation on 12 November 2026. Advertises local organic/non-GMO beans at Rp22,500/kg, minimum 25 kg. Delivery coverage and current contracted terms are unverified; product and output claims are the seller’s.
[11] Kristanti and Almuntaha (2017), Penentuan Saluran Pemasaran terhadap Tingkat Harga pada Rantai Pasok Kedelai di Kabupaten Grobogan ↩
Original source. Novita Erma Kristanti and Iqbal Saimima Almuntaha, Agritech 37(4), 2017; abstract and chain descriptions on pp.446–447 reopened. Historical fieldwork documents collector-to-processor sales and upstream drying. It establishes precedent, not current costs or continuity. Historical prices are excluded from the economics.
Competitive comparison
| Alternative and channel | Price and reach | Outcome and remaining question |
|---|---|---|
| FKS Multi Agro / FKS Food and Agri: sourcing and logistics; wholesale and small-business sales. | Indonesia, including Java; matched customer price unknown. | Integrated procurement, holding and transport advertised; no general inefficiency demonstrated. Must obtain a route-specific quote. [2] [5] |
| GAKOPTINDO / local KOPTI: cooperative purchasing and local-harvest participation. | Local processor networks; individual price and credit terms unknown. Association regional prices are context. | Existing purchasing relationships; a new supplier must improve cost, grade or convenience after switching costs. Historical channel evidence is dated. [2] [4] [11] |
| PT Gerbang Cahaya Utama and PT Putra Permata Pasifik: local-harvest buyers. | Pati participation reported in 2025; current price, capacity and customer channel unknown. | Further evidence that local absorption already exists; no unsupported comparison of service quality. [2] |
| Kedelai Indonesia: premium local beans through product enquiries. | Rp22,500/kg advertised, minimum 25 kg; Sleman office, delivery coverage unknown. | Local product already marketed for tofu/tempe. Premium claims do not supply a low-cost mass-market price anchor. [10] |
| Local collectors and owner direct buying: village contacts and resale. | Historical Grobogan channels and owner recollection; present quote unknown. | Can avoid extra resale steps. Own buying includes time, preparation, freight, losses and funding. [6] [11] |
| FDP freight support / SPHP intervention: public implementing partners. | Targeted routes/conditions; no universal customer price. | Include the supported offer wherever eligible; private economics cannot assume support away. [2] [3] [4] |
| Retain familiar imported beans and adjust food portions. | Current supplier spending continues; Grobogan tempe response reported. | Protects the nominal retail price by changing the product. It is not a saving per comparable unit of food. [7] |
Evidence ledger and competing explanations
Observed: reported input-price pressure, one owner’s historical purchasing experience, existing domestic sourcing, premium local supply and historical short channels. Inferred: preparation, continuity, relationships and funding help explain why a local kilogram does not automatically displace imports. Hypothesis: pooled orders and selective preparation create a net advantage on particular present-day routes. Annual production and past low farm prices do not establish that advantage. Alternative explanations include an already-efficient collector, superior incumbent terms or unsuitable local batches. No customer, supplier or partner has been recruited and no production test has been performed.
Economics: cost per comparable output
Let P be the delivered bean price and Y the comparable saleable tofu output per kg of beans. Before differences in other processing costs, bean cost per food unit is P/Y. The proposed offer wins only when Y(local)/Y(alternative) exceeds 11,000/11,300 = 0.973451. The implied maximum relative shortfall is 2.6549%. Rp11,000 is an assumed offer; Rp11,300 is a dated regional reference. Actual extra labour, energy, rejection and payment terms must be added. Neither a matched quote nor measured output has established a saving. [4]
Illustrative supplier budget: Rp11,000 revenue minus Rp400 contribution = Rp10,600/kg maximum total variable cost. Procurement at the historical Rp9,006/kg divided by assumed 98% retained weight is Rp9,189.80; residual budget Rp1,410.20. At 95% retention, procurement becomes Rp9,480 and the residual Rp1,120. At the historical Rp10,775 producer reference and 98% retention, procurement alone is Rp10,994.90. All retention rates are assumptions. The reference is policy context reported in 2025, not a verified present legal minimum or purchase offer. [2]
The residual must cover contracted cleaning/drying, bags, freight, short holding, finance, expected loss/bad debt and allocated acquisition. No route cost is measured. Acquisition illustration: assumed Rp1.5 million selling, samples and travel per retained owner divided by 100 kg/day × 300 days = Rp50/kg in the first year, included within the Rp10,600 ceiling. Lower purchases or failed retention increase this allocation; it is not a measured acquisition cost.
Self-buying breakeven: Q = fixed trip/setup cost ÷ (supplier’s delivered usable-kg price − own usable-kg procurement and variable cost), only where the denominator is positive. Numeric breakeven cannot be established without actual quotes and a purchasing diary. Owned truck/dryer comparisons must include depreciation, labour, utilisation and funding; owning equipment is not free. Compare credit-equivalent terms and food output, including eligible supported supply.
Market size and overhead workings
First regional market, conditional calculation: 230,210 national tonnes in 2024 × 30.80% = 70,904.68 Central Java tonnes. Assume 10–20% compatible, uncommitted and route-accessible: 7,090.468–14,180.936 tonnes. At assumed Rp11,000/kg, potential resale value is Rp78.0 billion–156.0 billion. Capture an assumed 10–20% of those respective bounds: 709.0468–2,836.1872 tonnes; annual gross resale Rp7.80 billion–31.20 billion. The weak inputs are accessible supply and buyer adoption; this is a supply-constrained estimate, not a customer count or current stock survey. [9]
At assumed Rp400/kg contribution, those volumes leave Rp283.6 million–1,134.5 million before fixed costs. Assumed Rp600 million annual fixed overhead requires 600,000,000/400 = 1.5 million kg, or 1,500 tonnes. The range then yields approximately minus Rp316.4 million to plus Rp534.5 million before tax. At Rp300–500/kg contribution, the same overhead requires 2,000–1,200 tonnes. Revenue and profit are distinct; no observed margin is claimed.
TAM, broad annual supplier revenue: 2.74 million tonnes projected national consumption × 1,000 × Rp10,500–12,000/kg = Rp28.77 trillion–32.88 trillion. The price range uses reported price/reference context; it is a valuation estimate across alternatives, not measured sector revenue. [3] [4]
SAM, where this local-supply model could apply: historical national production 230,210 tonnes × assumed 30–50% compatible, uncommitted and serviceable supply × 1,000 × assumed Rp10,500–11,400/kg = Rp725.16 billion–1,312.20 billion annually. This national operating pool extends beyond the first province but excludes no incumbent-tied volume by an invented count. The availability share remains unverified. [9]
Year-five SOM: assume 3–5% of the respective serviceable-market bounds = Rp21.75 billion–65.61 billion annual gross resale, or approximately 2,072–5,755 tonnes. Both bounds use the same expansion path: replicate workable short domestic-harvest routes into additional processor clusters. Shares are assumptions about sales taken from existing traders, cooperatives, collectors or self-buying. No planting target or premium segment is added. Tempe is a possible later adjacency only after a separate compatibility test. The inventory alternative below overlaps this market and cannot be added to it.
Alternative business and proposed tests
Seasonal local soybean stock and scheduled delivery would buy suitable harvest lots, pay for secure storage and release small deliveries through harvest gaps. The owner pays per delivered kg instead of funding a season’s beans. Collectors gain prompt purchase and agreed offtake; a warehouse and lender earn rent/interest on acceptable terms. None has agreed. An experienced stockist would need inventory capital, loss monitoring, traceability, suitable storage and credible drawdown demand. Concentrated harvest timing and reported local-supply history support investigating continuity, not a proven seasonal trading margin. [2] [6]
Its decisive assumption is that the harvest discount exceeds incremental holding cost while leaving a buyer saving against stockists and own storage. Illustration only: Rp10,000/kg stock × 12% annual simple funding × 90/365 = Rp296/kg; 2% storage loss adds 10,000/0.98 − 10,000 = Rp204/kg purchase uplift; assumed storage/pest/insurance Rp250/kg produces about Rp750/kg extra holding. A route otherwise costing Rp10,600 excluding long holding would reach about Rp11,350, above the assumed Rp11,000 offer. A discount greater than Rp750 merely restores the original contribution. Avoid counting ordinary holding twice. All inputs are assumptions, and future sale prices are unknown.
Test that inventory business on paper first: obtain matched harvest, future-delivery and own-storage offers, itemised funding/storage/loss budgets and owners’ priced drawdown commitments. Measure cost per comparable delivered input/output, cash tied up and committed volume. Reject it if no discount survives full holding costs with savings for both parties. Do not lease a warehouse to answer a question that quotes can settle.
For the featured pooled-route supplier, first compare current invoices with contemporaneous collector, preparation, freight and competing delivered offers, plus an owner’s self-buying diary. Hold grade, timing and payment terms comparable. If a price band survives, use existing equipment for a small paired production batch, weighing saleable tofu at comparable moisture and quality and counting extra labour and rejection. Repeat procurement quotes across harvest and off-season conditions before scaling. Measure buyer cost per comparable food unit, supplier contribution, retained orders and cash days. This tests a business, not merely whether local beans can be made into tofu. No beneficial price band, or inadequate repeat volume, falsifies the commercial hypothesis.
Routes considered and open questions
Harvest matching, pooled delivery, measured moisture/grade, selective preparation, direct dispatch of compliant lots, defined invoice adjustments and suitable return-load quotes contribute to the featured business only where they lower a real cost. Short holding satellites and delivery frequency are operating variants. They create no additive market or verified saving. Unconditional raw delivery transfers preparation to owners; mobile fleets add capital before an advantage is known; automated intake assumes a bottleneck not established here; processor pre-financing cannot also remove the processor’s cash burden. These routes were not treated as separate businesses.
The pooled-route supplier is first to investigate because its quote-and-batch comparison is cheaper than taking seasonal inventory risk. It also has the clearest current expansion case, though neither route has demonstrated sales or profit. The outstanding commercial evidence is current compatible free supply, matched output, owner retention, working-capital terms and achievable route volume. Processor owners, collectors, contractors, hauliers and regional traders could supply it.
Editorial verification limitation: published-repeat and recent-heading checks remain incomplete. Three bounded public site searches found no results; that does not establish absence of an earlier investigation. The saved uncertainty is retained for editorial handling.
Leave a Reply