Transpacific Bound

Transpacific Business

Indonesia Promised to Buy More U.S. Soy. How Much Actually Shipped?

Through September, U.S. soybeans to Indonesia sat at 54% of the annual tonnage target. Wheat was under a quarter, soymeal about 4%. Line those three up, then read the pace before you treat the pledge as booked demand.

Thomas ReedNovember 20, 20267 min
Grain barges tied along the Mississippi at the Cargill Westwego export elevator near New Orleans
Barges at the Cargill-Westwego grain elevator on the Mississippi near New Orleans, where U.S. grain moves toward export. Census shipment tallies show how much of Indonesia’s farm-volume pledge has actually cleared. Photo: USDA / Lance Cheung (public domain)

What the Census numbers show through September

A Reuters analysis published today works through U.S. Census Bureau trade data on the farm side of Washington’s Asia trade deals. For a supplier who sells soy, wheat, or meal into Southeast Asia, Indonesia is the case with named annual tonnages beside the dollar headline.

Indonesia agreed to facilitate about $4.5 billion a year in U.S. farm purchases. In the first nine months of 2026, U.S. agricultural exports to Indonesia came in at less than $2.4 billion, roughly where the same stretch of the year has sat for much of the past decade. The value line alone tells you the package has not rewritten the trade account.

The product split is sharper. Indonesia’s annual U.S. soybean target is 3.5 million tons. Through September, Census data put shipments at 54 percent of that goal, a little under two million tons if the percentage holds against the headline tonnage. Wheat exports were under 450,000 tons, less than a quarter of a 2-million-ton annual commitment. Soybean meal, or soymeal, sat around 150,000 tons, about 4 percent of a 3.8-million-ton target.

Nine months is three-quarters of a calendar year. If cargo moved evenly, you would want roughly 75 percent of an annual tonnage target by then. Soybeans are behind that pace. Wheat and meal are not even close.

What the deal text actually committed

The Agreement on Reciprocal Trade, published by the Office of the U.S. Trade Representative after the February signing with President Prabowo Subianto, does not read like a purchase order. Article 6.4 says Indonesia shall facilitate purchases by Indonesian companies of U.S. originating goods as set out in Annex IV. The farm chapter asks Jakarta to support and facilitate commercial arrangements toward that $4.5 billion indicative farm total, including annual volumes for five years: at least 3.5 million metric tons of soybeans, 3.8 million of soybean meal, and 2 million of wheat.

After those five years, the floors drop: more than 2.5 million tons of soybeans, more than 200,000 tons of meal, more than 1.3 million tons of wheat. The near-term numbers are the stretch targets. The later floors sit closer to an existing U.S. bean trade and a much thinner ask for meal.

Annex IV also includes a clause suppliers should keep in view. If the annual import volumes miss the listed amounts, and the United States determines that no Indonesian trade barrier limited or prevented the imports, Washington is not supposed to treat Indonesia as in breach of those volume commitments. Missed tons without a finding of a barrier are a commercial shortfall in the deal’s own design. They do not, by themselves, open a remedy.

Why soybeans, wheat, and soymeal are different stories

Soybeans were always the least speculative of the three. Indonesia already buys most of its food-grade beans from the United States for tofu and tempeh. Akindo, the country’s soybean importers association, has put annual consumption near 2.7 to 2.9 million tons. The 3.5-million-ton pledge sits above that use. Hitting it would mean stock-building, diversion into crush, or a larger food-grade pull than kitchens usually absorb. The 54 percent shipment share through September looks soft against a flat calendar. It still looks like a market that knows how to buy U.S. beans.

Wheat sits in the middle. Flour millers had already been lifting more U.S. cargo. Traders told Reuters in February that 2025 U.S. wheat purchases reached about 1.1 million tons, up from roughly 750,000 the year before, and that 1.25 to 1.3 million tons looked like a realistic 2026 ceiling. That private estimate sits under the 2-million-ton Agreement figure. Separately, the Indonesian Flour Mill Association (APTINDO) and U.S. Wheat Associates signed a July 2025 memorandum for at least 1 million tons a year of U.S. wheat from 2026 through 2030. Census shipments under 450,000 tons through September sit below even that softer miller track.

Soymeal is the hard one. U.S. meal sales into Indonesia were still measured in the low hundreds of thousands of tons before the deal, around 216,000 tons in the 2025 framing traders used when the pledge landed. Getting anywhere near 3.8 million tons would mean taking a large share of Indonesian feed protein that still comes from South America. Jakarta also routed large feed-mill meal volumes through state channels, including assignments to Berdikari, a state-owned enterprise under the food holding, and tightened import licensing for meal and feed wheat in 2026. USDA’s Foreign Agricultural Service has flagged longer permit chains and transition friction. A 4 percent shipment share through September matches a channel that has not opened at pledged scale.

What shipment data can tell a supplier about demand

Census export tallies answer a narrow question: how much U.S. product cleared for Indonesia in the period. They skip letters of intent, association MoUs, and White House dollar headlines. For a commercial team, that narrowness is useful.

Start with the calendar. Fifty-four percent of an annual soybean target with a quarter of the year left is a soft miss you can still close if fourth-quarter loads accelerate. Less than a quarter on wheat, and 4 percent on meal, means the remaining months would need a surge Indonesian crushers, millers, and feed plants have not shown in the first three quarters. Price, freight, and competing origins still sit between a facilitation annex and a booked vessel.

Then rank the paper in front of you. A confirmed purchase order, an import license, a state buyer tender, or a vessel fixture is demand you can staff against. An APTINDO-style MoU is an industry coordination signal. Annex IV facilitation language is a political ceiling and a monitoring metric. The Agreement’s own breach clause says missed volumes without a barrier finding do not, by themselves, put Jakarta out of compliance. Shipment data tells you what buyers already paid to move.

The product split still matters on the desk. A soybean seller watching 54 percent still has customers who already prefer U.S. beans. A meal seller watching 4 percent is waiting on import channels and buyer economics, not on another speech about $4.5 billion.

What to watch before you staff the chase

The next Census monthly releases for Indonesia-bound soybeans, wheat, and meal through year-end are the cleanest pace check. Look for miller and feed-buyer notices that clear the APTINDO-scale wheat track, or that open meal volumes beyond the thin 2025 baseline. Check whether Berdikari or other designated importers name U.S. origins in sizable feed-meal tenders, and whether import-license timing shortens enough for private traders to book November and December cargo.

Until those show up, treat the three products as three businesses under one headline. Soybeans are the only line near a workable pace. Wheat is lagging a target that already sat above what millers privately said they could absorb. Soymeal remains a policy-and-channel story. The shipment numbers already tell you which of those stories buyers have paid to move.

Related stories

Related global reads