China Aluminium Stocks Fall Below 800kt, the Floor Has an Expiry

Chinese aluminium social inventory fell 63,000 tonnes in a single week, dropping the national total below the 800,000-tonne level for the first time this cycle.

That pace of decline is what makes this notable, not the direction of travel. Weekly draws had been running at a third of that rate only a fortnight earlier, and the acceleration points to two forces arriving at once: constrained outbound shipments from Xinjiang colliding with concentrated pre-holiday restocking ahead of China’s National Day break.

The distinction between those two forces matters more than the headline number. One is a genuine physical tightening; the other is a delivery delay with a built-in reversal. Here is a clear-eyed read on whether the current tightness is real or temporary, and what that tells you about how long Shanghai Futures Exchange (SHFE) price support can hold before the picture flips.

What the 63,000-tonne drawdown actually shows

The story is in the sequence, not the snapshot. Shanghai Metals Market (SMM) data through September 2026 traces a destocking phase that started steady and then sharpened.

Early September opened with social inventory at 815,000 tonnes, already down 37,000 tonnes week-on-week. By 10 September, primary inventory sat at 796,000 tonnes, a 19,000-tonne weekly draw that quietly breached the psychologically important 800,000-tonne threshold. Four days later, SMM’s weekly update put stocks at 776,000 tonnes.

Then the pace tripled. The week of 20 September recorded a 63,000-tonne decline, with 43,000 tonnes of that fall landing between Monday and Thursday alone.

Billet inventory tells a quieter counter-story. SMM logged a 142,000-tonne billet build in early September, with that overhang easing only marginally by mid-week of 20 September, down just 500 tonnes. The billet excess persists but is not yet transmitting aggressively into the ingot draw, which sharpens the question of what is actually pulling ingots off the shelves.

The jump from a 19,000-tonne weekly draw to a 63,000-tonne one tells you at least one structural force has intensified rather than simply persisting. That is why the SHFE price floor here reflects real near-term physical tightness, not speculative positioning.

China’s aluminium export dynamics in 2026 have added a further dimension to domestic inventory behaviour: when Gulf-driven supply disruptions pull Chinese metal seaward, the resulting outbound flows reduce the ingot volumes available for domestic warehouse replenishment, compressing the inventory baseline from which pre-holiday draws are measured.

Two forces, one outcome

Two drivers are working at the same time, and they are distinguishable in character even though they produce the identical observed effect on the inventory line.

The first is a logistics bottleneck in Xinjiang, where outbound rail shipments have been hindered. The second is tactical pre-holiday restocking as buyers stock up before the National Day shutdown. Each deserves its own examination, because the trade implications diverge sharply depending on which one is dominant.

Xinjiang’s logistics bottleneck and why it hits inventory so hard

To understand why a regional shipping snag moves the national inventory figure so violently, start with how much of China’s aluminium comes out of one remote province.

Xinjiang is not a marginal player. It is the single heaviest weight in China’s aluminium supply system, and its geographic isolation turns any transport friction into an outsized national effect.

• Installed capacity: 6.18 to 6.53 million tonnes
• Utilisation rate: 99.97%
• Share of China’s total capacity: 13.7% to 15.6%
• Share of national output: more than 15% to 20%
• Share of global production: approximately 9%
• Major smelters: eight
• Haul distance to southeast China ports: 4,000 to 5,000 km
• Transport cost premium: RMB 400 to 600 per tonne

Near-zero idle capacity At 99.97% utilisation, Xinjiang runs with essentially no spare buffer. There is no slack in the system to absorb shocks, which means the region’s output moves entirely at the mercy of its outbound logistics chain.

China’s aluminium supply ceiling, driven by the 45-million-tonne statutory cap on domestic smelting capacity, means that Xinjiang’s near-zero idle rate is not an anomaly specific to one province but a feature of the entire Chinese production system operating without meaningful headroom.

That logistics chain is the vulnerability. Xinjiang’s smelters depend almost entirely on long-distance rail to move ingots to the coastal consumption hubs where they clear into warehouses. When outbound rail slows, arrivals at those hubs fall almost immediately, and social inventory drops as a direct consequence rather than because anyone is consuming more metal.

The reversal mechanism is already known and already dangerous. When these backlogs clear, delayed ingots do not trickle back in; they arrive in bulk. Late December 2025 offered a concrete precedent, when a resolved bottleneck triggered a sharp short-term spike in domestic stocks as the queued volume flooded through at once.

For your positioning, this reframes the entire tightness. The current draw is not the product of smelter shutdowns or capacity exits, both of which would sustain a price floor. It is a delivery delay that will eventually clear. That makes the timing of resolution, not the size of the disruption, the variable to track. It is the difference between reading this as a structural supply cut and reading it as a temporary window before reversal.

Pre-holiday restocking: genuine demand signal or tactical noise?

The second force is harder to pin down, because the same inventory drop supports two genuinely competing readings.

The supply-driven case argues that high SHFE prices have kept downstream buyers on a strict just-in-time footing throughout this period. Under this reading, the rapid destocking owes more to low arrival rates than to any surge in end-use consumption. SMM’s own holiday-period analyses have repeatedly attributed destocking phases primarily to constrained inflows rather than demand strength.

SMM’s aluminium rod inventory analysis from mid-September 2026 supports this picture, showing that rod stocks remained elevated and downstream operating rates were suppressed by oversupply conditions, a dynamic that runs counter to any broad-based demand recovery narrative.

The demand-driven case points to seasonality. Pre-holiday downstream restocking is a well-established pattern, the autumn cousin of the “Golden March/Silver April” buying dynamics, and the approach of National Day has concentrated procurement into the current week.

The complication is that SMM noted in early September that genuine peak-season end-user demand had not yet fully materialised. That single observation tempers the demand interpretation considerably. Buying can be brisk without consumption being strong; those are not the same thing.

Price context reinforces the caution. SHFE aluminium has traded in the mid-RMB 24,000 per tonne range, with the AL2610 contract moving between RMB 24,250 and 24,430 per tonne in mid-September. For historical scale, the Q1 2026 post-New Year build added 65,000 tonnes to reach 703,000 tonnes, while a mid-April 2025 single-week draw of 55,000 tonnes shows these swings are seasonally routine. The persistent 142,000-tonne billet build, meanwhile, signals that downstream is not restocking uniformly across product types.

Before committing to either reading, three questions sharpen the analysis:

1. Are downstream operating rates improving, or merely flat?
2. Are ingot arrivals genuinely disrupted, or simply delayed in transit?
3. Will post-holiday inflows absorb into real consumption, or rebuild inventory?

The honest read for a commercial position is that the demand signal is real but concentrated and tactical. It supports prices in the near term without confirming the structural demand recovery that a medium-term bullish stance would require. Getting this distinction wrong in either direction produces a mispriced entry or a mistimed exit.

What downstream operating rates are telling you

The clearest real-time test sits in the operating-rate data. During the week of 20 September, the share of liquid aluminium usage declined 0.02 percentage points week-on-week, reflecting select downstream production curtailments.

That runs directly counter to a clean demand-surge narrative. If consumption were genuinely accelerating, liquid aluminium usage would rise, not slip. This is the single most important figure for testing whether the restocking is broad-based or concentrated in a handful of product segments, and right now it leans toward the latter.

Price trajectory and the post-holiday reversal risk

The current price bands reflect a market held between a floor and a ceiling, and both are worth naming precisely.

SHFE aluminium is projected to trade between RMB 23,800 and RMB 24,500 per tonne in the week following 20 September, with the inventory drawdown providing the floor and macro uncertainty capping the top. On the London Metal Exchange (LME), SMM forecasts a US$3,200 to US$3,350 per tonne range over the same week.

The counterforce is the post-holiday rebuild, and it is not hypothetical. When pre-holiday restocking ends and delayed Xinjiang ingots arrive in bulk, inventory shifts from drawdown to rapid accumulation, a pattern that has repeated across prior holiday cycles.

The reversal risk in one figure During a comparable pre-holiday phase in early 2025, analysts projected Q1 inventory could peak at 850,000 to 900,000 tonnes once the restocking catalyst expired. The same forces creating today’s draw are the ones that reverse it.

The macro backdrop frames the near-term dynamic without overriding it. The market broadly expects the US Federal Reserve to ease toward roughly a 3% policy rate across 2026, softening the dollar and providing genuine medium-term support to metal prices. Pulling the other way, Chinese industrial demand flatlined in the first half of 2026, with Citi’s end-use tracker down 0.4% year-on-year, capping any decisive breakout.

Geopolitics has already faded as a driver. The Middle East war premium that pushed LME three-month prices to US$3,545 to US$3,787.50 per tonne between March and June 2026 had largely dissipated by July 2026, easing prices back toward US$3,170 per tonne.

The takeaway on timing is direct: the current floor is inventory-driven and time-limited. Once the holiday passes and Xinjiang backlogs clear, the drawdown reverses. If you have not planned an exit before that inflection, you are exposed to a rapid move the other way. The window of inventory-supported strength is measurable in weeks, not months, and the rebuild is the most reliable exit signal to watch.

The market’s near-term floor is real, but post-holiday price support has a hard expiry date: once Xinjiang backlogs clear and restocking demand exhausts itself, the same forces that drove the drawdown rotate into a rebuild that erodes that support rapidly.

Three variables that determine whether the price floor holds

Turning the analysis into something usable means naming the specific signals that will confirm or break the current floor into and beyond the National Day period.

1. Xinjiang outbound rail normalisation. A continuation signal is persistent shipment friction keeping arrivals suppressed; a deterioration signal is the first bulk arrivals of delayed ingots, which mark the start of the rebuild.
2. Downstream operating rates post-holiday. A continuation signal is rising liquid aluminium usage as real consumption absorbs inflows; a deterioration signal is flat or falling operating rates, meaning the metal simply rebuilds warehouse stocks.
3. Post-holiday inventory rebuild pace versus historical analogues. A continuation signal is a slow, shallow build; a deterioration signal is an accumulation approaching the prior 850,000 to 900,000-tonne peak template.

The observation cadence differs across these. SHFE warrant data and SMM weekly inventory updates give you a near-real-time read on variables one and three. Downstream operating-rate surveys, the truest test of variable two, carry a one-to-two week lag, so the demand confirmation always arrives after the price move has begun.

LME stock levels at century lows have tightened the global supply cushion that would ordinarily absorb a Chinese inventory rebuild, meaning the post-holiday reversal risk identified here could carry further price consequences internationally than a comparable drawdown-to-rebuild cycle would have produced in prior years.

The RMB 23,800 per tonne floor of the projected range is the level to watch as the practical marker of whether inventory support is still holding.

The billet overhang as an amplifier

The 142,000-tonne billet build from early September, still barely moved after a 500-tonne weekly decline, is a secondary risk layer worth flagging. Billet excess does not transmit to ingot prices in a straight line; the two products clear through different channels.

The danger is coincidence. If a post-holiday ingot rebuild lands at the same time the billet overhang persists, the combined pressure on ingot prices could run beyond what the historical analogues alone would suggest.

What the data tells you before the holiday, and what it does not

Two forces are confirmed, and both point the same way in the near term. Xinjiang’s logistics bottleneck and concentrated pre-holiday restocking are together pulling inventory below 800,000 tonnes and supporting SHFE prices within the RMB 23,800 to 24,500 per tonne band. Neither, on the evidence, amounts to a structural demand recovery.

Here is the split worth holding in mind:

What the data confirms

• Physical tightness is real and inventory-driven right now
• The Fed easing path toward a 3% target rate provides genuine medium-term price support

What remains uncertain

• Whether post-holiday operating rates recover enough to absorb the inevitable rebuild
• Whether the flat H1 2026 demand backdrop (down 0.4% year-on-year, Citi) lifts at all

The central unknown is that first uncertainty. It is the variable separating a brief price floor from a sustained one, and it will not resolve until operating-rate surveys land after the holiday.

The decision context, then, is one that rewards short-duration, tactically timed positioning rather than structural long exposure. The post-holiday rebuild is the exit signal; the monitoring variables above are how you see it coming before consensus does.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions. Past performance does not guarantee future results. Financial projections are subject to market conditions and various risk factors, and forward-looking statements are speculative and subject to change based on market developments.