We are data aggregators, not forecasters — so this page does one thing: it lays out what the USDA's own reports say about China's grain demand, quoted with a full citation to every source, and keeps our market read clearly separate. USDA's attaché posts and its WASDE-cycle circulars are read together and dated; where our own view diverges — because a fast-moving trade file can outrun a monthly report — that read is walled off in its own box below each subject. Refreshed each WASDE cycle.
China's corn import book keeps shrinking — the August WASDE cut MY2026/27 imports 6.0 → 5.0 MMT and MY2025/26 to 4.0, with sorghum cut 900,000 t on expectations of lower U.S. sorghum exports — and 0 MT of the corn book is U.S. on the books.
The August circular cut China's corn imports for a second consecutive month: MY2026/27 6.0 → 5.0 MMT — landing exactly on the Beijing attaché's July number — with the reason given as "Expectations of lighter demand for corn imports," and MY2025/26 5.0 → 4.0 MMT on trade data [Grain WMT, Aug 2026; WASDE-674 via our ingest]. Domestic feed and residual use was trimmed 243 → 242 MMT for MY2026/27, so this is not a demand collapse — it is import demand being met from somewhere other than the world market.
The new information is sorghum. USDA cut China's MY2026/27 sorghum imports 7.5 → 6.6 MMT on "Expectations of lower U.S. sorghum exports," and cut U.S. sorghum exports 5.2 → 4.3 MMT on a "Smaller crop" in the same table. Sorghum is the one U.S. feed-grain line into China that has actually been shipping while the corn line sits at zero, and both ends of it were reduced this month. Anyone treating U.S. feed-grain exposure to China as a corn question is missing the leg that was carrying the volume.
Global context runs the other way and matters for what the U.S. does with the corn: USDA raised U.S. corn exports 81.0 → 83.0 MMT on "Reduced competition from other exporters," and the EU's import book rose to 23.5 MMT on a 20-year-low crop (eu-grain). The demand that left China's book is being replaced in Europe's.
As of early July 2026, USDA's read on China corn is more domestic supply, less import demand. The Beijing attaché raised MY2026/27 corn production 1 MMT to 306 MMT on better yields and expanding Northeast-China area, and cut Post's MY2026/27 import estimate to 6 MMT — 2 MMT below March — citing continued import restrictions and "Beijing's silence on grain import deals" [CH2026-0085, Jul 3]. MY2026/27 ending stocks are pegged at 164 MMT, 11 MMT below the prior year on lower imports and more state auctions [CH2026-0085]. Feed and residual corn use is held at 241 MMT, with poultry — not hogs — the growth engine [CH2026-0085].
On the official side, the July WASDE-cycle circular corroborates the import-demand weakness: it cut China's MY2025/26 corn imports to 5 MMT (−1 MMT) on trade data [Grain: WM&T, Jul 2026] — the same level the attaché reports. For MY2026/27, that circular listed no change to China corn trade this month (only barley moved, +0.2 MMT to 10.7 MMT on stronger Canadian supply) [Grain: WM&T, Jul 2026].
Trade politics sit underneath all of this: China cut additional tariffs on U.S. corn to 10% on Nov 10, 2025, and leaders met May 14, 2026 — yet the attaché reports 0 MT of outstanding U.S.-origin corn sales for MY2025/26 and no resumption of U.S. purchases. Meanwhile Argentina re-entered the China market after 13 years (~454,000 MT / 12 vessels since April) and Brazil began shipping DDGS [CH2026-0085].
| Attribute | Reference | Current | Source |
|---|---|---|---|
| Corn production MY2026/27 | 305 MMT | 306 MMT (+1) | CH2026-0085 |
| Corn imports MY2026/27 (Post) | 8 MMT | 6 MMT (−2) | CH2026-0085 |
| Corn imports MY2025/26 | 6 MMT | 5 MMT (−1) | CH2026-0085 + Grain WM&T |
| Corn feed & residual MY2026/27 | 241 MMT | 241 MMT (unch.) | CH2026-0085 |
| Corn ending stocks MY2026/27 | — | 164 MMT | CH2026-0085 |
| Corn imports MY2026/27 — Official | 6.00 MMT (July) | 5.00 MMT (August, −1.0) | WASDE-674 / Grain WMT Aug 2026 |
| Corn imports MY2025/26 — Official | 5.00 MMT (July) | 4.00 MMT (August, −1.0) | WASDE-674 |
| Sorghum imports MY2026/27 — Official | 7.50 MMT (July) | 6.60 MMT (August, −0.9) | Grain WMT Aug 2026 |
| Corn feed & residual MY2026/27 — Official | 243 MMT (July) | 242 MMT (August) | WASDE-674 |
| Corn ending stocks MY2026/27 — Official | 165.13 MMT | 165.13 MMT | WASDE-674 |
Both sources predate or coincide with a fast-moving trade file, and that is where we may legitimately diverge from the USDA baseline:
Added August 16, 2026. The July call — that the import number was the most perishable line in the view — verified in the direction of less, not more: Official imports fell another 1 MMT for each marketing year, and the attaché's 6 MMT is now the Official 5. Two months of consecutive cuts with feed use roughly flat says the substitution is domestic and durable (state auctions, aged rice into the feed ration — see china-rice), not a timing effect waiting on a trade announcement.
We would now change what we watch. The corn line is at zero U.S. and shrinking, so the marginal news there is binary and political. The line with actual U.S. volume in it is sorghum, and USDA just cut it 900,000 t while attributing the cut to expectations about U.S. supply rather than about Chinese demand. That is a supply-side reason for a demand-side number, which is the kind of attribution worth checking against export-sales data rather than accepting. And the corn demand that keeps leaving this book is showing up in Europe's (eu-grain), where U.S. exports were raised 2.0 MMT in the same release — so the U.S. export story this year is a destination substitution, not a volume loss.
USDA cuts China's MY2026/27 cotton crop 6% to 33.5 million bales — but the cut is a subsidy map, not a weather map: area in the Yellow/Yangtze basins falls ~34% against 6% in Xinjiang, which now carries ~90% of national production, while Chinese domestic cotton trades ~32 cents above the A-Index.
The August release left China's crop alone (33.5 M bales, unchanged) and moved demand: domestic use MY2026/27 41.50 → 42.00 M bales and MY2025/26 41.00 → 41.50, with China named among the countries whose higher consumption lifts global use to a six-year-high 122.9 M bales. Imports for MY2025/26 were raised on trade data 7.00 → 7.20 M bales; the MY2026/27 import line is unchanged at 7.00. Ending stocks fell in both years (2026/27 35.49 → 34.69) [Cotton WMT Aug 2026; WASDE-674 via our ingest].
The price gap this view is built on narrowed, and it narrowed from the world side. China's domestic quote rose only 1.0 cent (119.3 → 120.3) between July 7 and August 10 while the A-Index rose 7.7 cents to 95.0, cutting the China premium from about 32 cents to about 25. Nothing in Chinese policy did that.
The August circular also identifies where the premium is actually leaking. Its India feature reports that India's cotton yarn exports to China tripled in the first ten months of the marketing year, taking market share from 7 to 21 percent, on a depreciating rupee and falling yarn prices, and that China's overall yarn imports rose over the same period. That is the China premium being arbitraged as yarn rather than as fiber — a channel that never appears in a cotton import line, and one reason the MY2026/27 import number can sit still at 7.00 M bales while consumption rises. See india-cotton.
USDA forecasts China's MY2026/27 cotton production at 33.5 million bales (7.29 MMT), 6 percent lower year-over-year but 10 percent above the five-year average, on harvested area of 3 million hectares (−2 percent) and yield of 2,431 kg/ha — 5 percent below last year but 9 percent above the five-year average [IPAD CIR, Jul 28 2026]. The decline is attributed to reduced harvested area, and USDA locates that reduction almost entirely outside Xinjiang: area in the Yellow River and Yangtze River basins is expected to fall about 34 percent from last year, against a 6 percent decline in Xinjiang.
The stated mechanism is policy, not weather. Production declines in eastern China and the North China Plain are "largely driven by policy changes and reduced profitability following the withdrawal of national cotton subsidies in those regions," with farmers shifting acreage "to more profitable alternatives, including rice"; area has contracted significantly in the eastern provinces since the 2014 reforms, while the target price-based cotton subsidy for Xinjiang growers remains in force per government reporting. The result is a national crop that is now about 90 percent Xinjiang, at yields "nearly double those in other regions."
On conditions, the report's general finding is favorable: China's summer crops were planted inside the optimum March–May window with "no observed or reported adverse conditions," soil moisture was favorable across the main production regions as of July, and VIIRS NDVI was above average as of July 19. For Xinjiang specifically, satellite observations from April through June "indicate favorable early-season conditions in the region, including strong irrigation and soil moisture levels."
But the report is not internally consistent on Xinjiang water. The caption to Figure 4 — an Aksu (northwest Xinjiang) comparison of July 2025 against July 2026 imagery — states that "there are significant differences in irrigation water reservoirs and crop conditions leading to year-over-year declines in area and production." That is a materially different statement from the body's "strong irrigation and soil moisture levels," and both appear in the same section. This view carries both without merging them.
From the July WASDE-cycle circular, the price context: China's domestic cotton price rose 2.3 cents to 119.3 cents/lb, against an A-Index of 87.3 and a U.S. price of 74.1 — a domestic market trading roughly 32 cents above the world index — while global 2026/27 production was raised 1.2 million bales to 117.3 million and ending stocks to 71.2 million [Cotton WM&T, Jul 2026].
| Attribute | Reference | Current | Source |
|---|---|---|---|
| China cotton production | ~35.6 M bales (implied) | 33.5 M bales / 7.29 MMT (−6%) | IPAD CIR |
| vs five-year average | — | +10% | IPAD CIR |
| Harvested area | ~3.06 M ha (implied) | 3.0 M ha (−2%) | IPAD CIR |
| Yield | ~2,559 kg/ha (implied) | 2,431 kg/ha (−5% y/y, +9% vs 5-yr) | IPAD CIR |
| Yellow/Yangtze basin area | — | −34% y/y | IPAD CIR |
| Xinjiang area | — | −6% y/y | IPAD CIR |
| Xinjiang share of production | — | ~90% | IPAD CIR |
| China domestic price (c/lb) | 119.3 (7-Jul) | 120.3 (10-Aug, +1.0) | Cotton WMT Aug 2026 |
| A-Index (c/lb) | 87.3 (7-Jul) | 95.0 (10-Aug, +7.7) | Cotton WMT Aug 2026 |
| China premium over A-Index | +32.0 c/lb (7-Jul) | +25.3 c/lb (10-Aug) | our computation from Cotton WMT |
| China domestic use MY2026/27 | 41.50 M bales (July) | 42.00 M bales (August) | WASDE-674 |
| China imports MY2025/26 | 7.00 M bales (July) | 7.20 M bales (August, trade data) | WASDE-674 / Cotton WMT Aug 2026 |
| China imports MY2026/27 | 7.00 M bales | 7.00 M bales (unchanged) | WASDE-674 |
| China ending stocks MY2026/27 | 35.49 M bales (July) | 34.69 M bales (August) | WASDE-674 |
The durable fact in this report is not the 6% cut — it is what the 2014 subsidy reform finally looks like twelve years on. A −34% area collapse in the Yellow and Yangtze basins against −6% in Xinjiang says the reform did not shrink Chinese cotton so much as relocate it to one province, which now carries ~90% of production at roughly double the national yield. The practical consequence for anyone reading a China cotton number: national production has become a Xinjiang water-and-weather number, with essentially no geographic diversification left to absorb a bad year there. The distribution of outcomes around the Chinese crop is now narrower in the middle and fatter in the tails than the smooth −6% headline suggests.
Which is why we treat the Figure 4 caption as the most important sentence in the document. If the Aksu reservoir comparison is the truer read of Xinjiang water than the body's "strong irrigation," the risk to 33.5 million bales is to the downside, and it is an irrigation-storage risk rather than a rainfall risk — a distinction that matters because a rain-driven model will not see it. This is structurally the same mechanism as Australian rice in australia-grain (water storage, not precipitation, as the binding constraint) and the same one we frame elsewhere as abstraction risk rather than as crop loss. It is also the specific thing worth watching before the northern-hemisphere harvest: an irrigated crop's water constraint is set months before harvest and is observable, unlike weather.
The second thread is that the eastern acreage that left cotton went to rice, per USDA's own sentence — so china-rice and this view are two halves of one land-allocation decision, and a slightly larger Chinese rice crop is partly made of the cotton crop it replaced. That is the rare case where two commodity views should be read together or not at all.
On price, a domestic market at 119.3 cents against an 87.3-cent A-Index is a ~37% premium, and it is the standing reason China's internal cotton economics do not transmit to the world price the way a 6% production cut in an open market would. We would not read the production cut as directly bullish ICE without an import or reserve-policy signal to carry it there — none appears in either source. For our own cotton work, the useful addition from this cycle is the Xinjiang-concentration framing and the irrigation question, not the headline tonnage.
USDA's official production assessment puts China's MY2026/27 rice crop at 147 MMT (+1%) on favorable satellite conditions and area partly won from abandoned eastern cotton — while the Beijing attaché reports China resumed aged-rice auctions for feed use in late May, the first since 2023, making rice a feed-grain story as much as a food one.
USDA forecasts China's MY2026/27 rice production at 147 million metric tons, up about 1 percent from last year and from the five-year average, on harvested area of 29 million hectares (unchanged year-over-year but slightly above the five-year average) and yield of 7.24 tons per hectare, up approximately 1 percent — a rise driven by "increases in both harvested area and yield" [IPAD CIR, Jul 28 2026].
Conditions are reported clean. China's summer crops were planted within the optimum March–May window with "no observed or reported adverse conditions"; soil moisture was favorable across the main production regions as of July, and VIIRS NDVI was above average as of July 19. For rice specifically, "satellite data indicates favorable conditions for early, intermediate, and late rice crops, supported by adequate rainfall and temperature patterns. No adverse precipitation or vegetation indicators have been observed." The report's own definition of the adverse case is worth carrying: "heavy, long-lasting rains and high humidity, resulting in waterlogging, crop lodging, and prolonged planting or replanting times beyond the optimum planting window."
Crop stage as of late July: early double-crop rice is in advanced maturity and harvest, single-crop rice is entering heading, and late double-crop planting is underway. The output split is ~16 percent early double crop, ~66 percent single crop, ~18 percent late double crop — so at the date of this report roughly a fifth of the crop was not yet established. Major producing regions: Heilongjiang, Hunan, Jiangxi, Hubei, Jiangsu, Sichuan, Anhui, Guangxi, Guangdong, Jilin and Yunnan.
Where the rice area came from is stated in the same report's cotton section: eastern-province farmers who lost the national cotton subsidy after the 2014 reforms "have shifted acreage to more profitable alternatives, including rice," with Yellow River and Yangtze River basin cotton area falling about 34 percent this year alone [IPAD CIR].
On the demand side, the July 3 Beijing attaché update supplies the piece IPAD does not cover: "China resumed targeted auctions of aged rice stocks for feed use in late May, the first since 2023, prompting Post to raise MY2026/27 rice consumption" [CH2026-0085]. Post also raised China's total grain feed and residual use to 293 MMT, "led by stronger-than-expected old stock rice releases for feed and continued poultry sector growth" — placing rice directly in the feed ration alongside corn, whose import book Post cut to 6 MMT for MY2026/27.
| Attribute | Reference | Current | Source |
|---|---|---|---|
| Rice production | ~145.5 MMT (implied) | 147 MMT (+~1%) | IPAD CIR, Jul 28 2026 |
| vs five-year average | — | +~1% | IPAD CIR |
| Harvested area | ~29 M ha | 29 M ha (unchanged, slightly above 5-yr avg) | IPAD CIR |
| Yield | ~7.17 t/ha (implied) | 7.24 t/ha (+~1%) | IPAD CIR |
| Rice consumption | — | raised on aged-stock feed auctions | CH2026-0085 (attaché) |
| Total grain feed & residual | — | 293 MMT (revised up) | CH2026-0085 (attaché) |
| Aged-stock feed auctions | none since 2023 | resumed late May 2026 | CH2026-0085 (attaché) |
A 1% rise in a 147 MMT crop is not itself news. What makes this view worth carrying is that two independent things are pushing Chinese rice toward the feed complex at the same time, and neither shows up in a rice balance sheet read on its own.
The first is on the supply side: the eastern acreage USDA says left cotton went partly to rice, so part of the 2026 rice crop is the physical residue of a subsidy decision made in 2014 (see china-cotton — the two views are one land-allocation story split across two commodities). The second is on the demand side: aged-stock auctions for feed use resumed in late May for the first time since 2023, and Post explicitly credits those releases for raising total grain feed and residual use to 293 MMT. Rice is being used to substitute in the feed ration in a year when China is buying 5–6 MMT of corn and zero U.S. corn — that is the connection to china-corn, and it is a plausible part of why the corn import book can stay collapsed without a visible feed-supply problem.
The honest limitation is that we have a direction and no quantity. Nobody publishes the auctioned volume, so we cannot size how much corn-equivalent the rice releases displace, and we would not put a number on it. What we would watch: further auction announcements, the corn-vs-rice feed price relationship in Chinese domestic markets, and whether the August cycle moves China's rice consumption line rather than its production line.
One condition caveat before this view gets quoted forward. USDA's clean bill of health is dated July 19–28 on a crop where ~18% (the late double crop) had not been established yet, and USDA's own definition of the adverse case is heavy prolonged rain and waterlogging — a late-summer risk in the southern double-cropping provinces, not a spring one. We track exactly that failure mode in our own China work, where the live 2026 concern is harvest-period rain records in Hebei and Shandong. "No adverse conditions observed" as of late July is a statement about the past, and the part of the Chinese rice crop most exposed to what follows is the part that was not yet in the ground when it was written.
China imports ~115 MMT of soybeans (~61% of world trade) — but the tonnage routes to Brazil and Argentina, not the U.S.
Where corn shows a collapsed import market, soybeans show a large and still-growing one. USDA's July balance sheet carries China soybean imports at 115 MMT for 2026/27 (raised 1 MMT this month) and 113 MMT for 2025/26 (also +1) — against domestic production of only ~21 MMT, so China imports roughly 85% of the beans it crushes (crush 111 MMT) [WASDE, Jul 2026]. At ~115 of ~189 MMT of world soybean trade, China is about 61% of all global soybean imports — the single buyer the whole trade is organized around.
But the tonnage is structurally South American, and the July cycle shows it plainly. The oilseeds circular raised China's imports "in line with prior year" and, in the same table, raised Brazil's soybean exports on "higher China demand"; U.S. soybean prices, by contrast, only "rallied over the past week on news of China soybean purchases" [Oilseeds WM&T, Jul 2026]. The attaché reporting fills in the destinations: Brazil is finishing a record 184 MMT export-oriented crop [BR2026-0029], and Argentina's single largest whole-bean destination is China at 2.63 MMT [AR2026-0009]. Total U.S. soybean exports to all destinations are ~45 MMT — under half of what China alone imports [WASDE, Jul 2026].
So the picture is not "China isn't buying" (it is, enormously) but "China's incremental bean is a Brazilian bean." The news moves the U.S. board; the cargo clears a South American port.
The domestic crop is not the variable. USDA's July 28 satellite-based production assessment puts China's own MY2026/27 soybean crop at 21 MMT, up 1 percent year-over-year and 6 percent above the five-year average, on harvested area of 10.3 million hectares — unchanged from last year — with the entire gain coming from yield (2.04 t/ha, +1%). Conditions are clean: the crop was planted inside the optimum March–May window, and "precipitation indicators and assessments from Normalized Difference Vegetation Indices (NDVI) indicate favorable early season conditions with no reported weather-related disruptions to planting or crop establishment," with soil moisture favorable as of July and NDVI above average as of July 19 [IPAD CIR, Jul 28 2026]. Key producing provinces are Heilongjiang, Nei Mongol, Anhui, Henan and Jilin. Note this is an official USDA production line (IPAD is the division behind USDA's world production forecasts), consistent with the 21 MMT carried in the July WASDE. Against ~115 MMT of imports, a good domestic crop and a bad one differ by roughly a rounding error in the import book.
| Attribute | Reference | Current | Source |
|---|---|---|---|
| China soybean imports | 114 MMT | 115 MMT (+1) | WASDE |
| China soybean crush | 110 MMT | 111 MMT (+1) | WASDE |
| China soybean production | 21 MMT | 21 MMT (unch.) | WASDE |
| China soybean area | — | 10.3 M ha (unch. y/y) | IPAD CIR, Jul 28 |
| China soybean yield | — | 2.04 t/ha (+~1%) | IPAD CIR, Jul 28 |
| China ending stocks | 44.27 MMT | 44.27 MMT (unch.) | WASDE |
| Brazil soybean exports | 117.5 MMT | 118.0 MMT (+0.5, "higher China demand") | Oilseeds WM&T |
| U.S. soybean exports (all dest.) | ~44.4 MMT | ~45.2 MMT (+0.8) | WASDE / Oilseeds WM&T |
Read next to china-corn, soybeans complete the pattern: China buys from everyone but the United States — in corn by buying almost nothing (and 0 MT from the U.S.), in soybeans by buying enormous amounts but sourcing them from South America. Same destination for the U.S. bushel; two different roads there.
SoftSignal aggregates the USDA WASDE, FAS World Markets & Trade circulars, and country attaché reports into one queryable layer — alongside COT positioning, drought, and export-pace data, readable by you and by AI.
Explore the Data