China’s Aluminium Inventory Cycle: Record Draw, Elevated Stocks

China’s aluminium ingot stockpiles shed 118,000 tonnes in just two weeks heading into the National Day holiday, the steepest pre-holiday reduction for this seasonal window in recent years. That figure alone reads as unambiguously bullish. The context complicates it.

Metals investors already track the National Day inventory cycle as a repeatable seasonal event. What sets 2026 apart is that the record-style drawdown was achieved from the highest aggregate stock level since 2020, meaning the numbers arrived unusual on both sides of the ledger.

The China aluminium inventory cycle around Golden Week is one of the most-watched seasonal patterns in base metals, and today’s SMM data release puts the 2026 figures firmly in view. This piece breaks down the pattern into a working framework.

Here is what the data actually tells you, and what it does not: what the pre-holiday destocking confirms, what it cannot confirm, and the specific variables to monitor once the holiday ends.

Six years of evidence: how China’s National Day holiday reshapes aluminium stocks

Across China’s major consumption regions, aluminium inventories follow a predictable three-phase rhythm around the National Day holiday. Stocks draw down before the holiday, accumulate through the holiday week itself, then reverse afterward. According to SMM data published via AL Circle, this sequence has repeated across all six holiday windows from 2020 to 2025.

SMM’s IOSCO-aligned benchmark methodology, which includes external audit verification since 2020, underpins the credibility of the inventory and price data series that analysts use to track this seasonal pattern across years.

The holiday-week accumulation is where the pattern shows its teeth. During the seven-day break, ingot stocks rose between 4% and 17.3%, billet stocks expanded between 12.7% and 66%, and combined stocks grew between 6.9% and 21.2% across those six years.

Those ranges are wide, and that matters for how you use them. The pattern reliably sets a direction, but the spread of outcomes means it cannot set a magnitude. Any investor leaning on this framework should treat the historical range as a cone of uncertainty rather than a single point estimate.

The inventory dynamics playing out in China’s consumption zones are one input into a larger structural picture: the global supply deficit projected for 2026 means that pre-holiday destocking events carry more price-relevant weight than they did in years when the global balance was closer to equilibrium.

The post-holiday phase is remarkably consistent. By the second week after the holiday, billets shifted into drawdown across all six years tracked, while ingots did not re-accumulate in that window except in 2021.

Why billets and ingots move differently

Billets accumulate faster during the holiday because downstream processors, extrusion plants and fabricators, stockpile more aggressively ahead of the closure than ingot consumers do. They are laying in raw material to keep lines running through the break.

That same dynamic runs in reverse afterward. Once operations resume, fabricators work through their holiday-period billet stock quickly, which is why billets draw down more sharply than ingots by week two. Keep that asymmetry in mind: it is the key to reading the 2026 billet numbers correctly.

What the September 2026 data actually shows

The 2026 story starts on 10 September, when combined aluminium stock across China’s major consumption zones reached 949,000 tonnes, split between 796,000 tonnes of ingots and 153,000 tonnes of billets. That aggregate was the highest for that calendar point since 2020.

From there, the ingot drawdown accelerated. In the week to 17 September, ingot stocks fell by 63,000 tonnes, a 7.9% reduction. The following week to 24 September brought a further 55,000-tonne draw, down 7.5%.

Combined, that is 118,000 tonnes of ingot inventory erased in two weeks, the steepest seasonal reduction for this window in recent years.

The billet picture ran a different course. Billet inventory peaked at 157,500 tonnes around 14 September, a four-year seasonal high, then eased to 149,000 tonnes by 24 September. Even after that decline, the level remained the highest for the period in three years and roughly 61,000 tonnes above 2023 levels.

The secondary and alloy segments split further still:

• Secondary aluminium ingot inventory across three major regions sat at 12,403 mt on 23 September, down 283 mt day-on-day.
• Secondary aluminium alloy ingot inventory measured 12,947 mt on 15 September, down 205 mt day-on-day.
• Cast aluminium alloy inventories, by contrast, recorded multiple consecutive weekly builds through September.

Here is the tension the reader should sit with. The steepest pre-holiday ingot drawdown in recent memory is being achieved from the highest starting stock level in six years. The net position heading into the holiday is more ambiguous than either the drawdown or the stock level suggests on its own.

Three drivers that made 2026 different, and why two of them will expire

The unusually steep 2026 destocking was not a single force. Three drivers stacked on top of one another, and the useful analytical work is separating the ones that persist from the ones that vanish with the holiday.

1. Xinjiang logistics disruption: A regional bottleneck curtailed arrivals into major consumption regions, forcing buyers to draw on existing stock. Expiry condition: fades once logistics normalise.
2. Pre-holiday restocking: Downstream buyers front-loaded procurement ahead of Golden Week closures. Expiry condition: completed once the holiday passes.
3. Price-triggered billet demand: A RMB 380/mt decline in the SMM A00 spot price pulled billet stocks off their four-year high by prompting real purchasing. Expiry condition: durable only if prices stay attractive.

SMM’s billet commentary described the destocking as occurring “on the resonance of supply disruptions and pre-holiday stockpiling,” and analysts have characterised the first two drivers as carrying built-in expiry dates. That framing matters. Two of the three forces behind the record drawdown are scheduled to disappear.

The read you should take is to weight the third component, genuine demand, more heavily than the headline 118,000-tonne figure when forming a view on post-holiday direction.

The RMB 380/mt price decline that triggered genuine billet purchasing fits a broader pattern: industrial demand signals in 2026 have repeatedly shown that price-sensitive fabricators re-enter the market when spot aluminium retreats, creating a demand floor that has limited the depth of previous corrections.

What “demand-led” actually means when production held steady

Analysts designate the ingot draw as demand-led for a specific reason: weekly primary aluminium production held near 874,700 tonnes throughout the period, with no major output cuts. The inventory fell because material was being consumed, not because supply was throttled.

A daily draw of 26,000 tonnes of primary ingot across three key markets was cited as evidence of genuine physical uptake. That production-stability context is what makes this a cleaner demand signal than a drawdown accompanied by output cuts would be.

One caveat tempers the read-through. Despite the destocking, downstream procurement remained largely just-in-time because of high prevailing prices, which limits how far you can stretch this into a claim of structural demand improvement.

The bull and bear cases heading into the post-holiday window

The 2026 data supports two defensible interpretations, and both deserve equal weight before you decide which is playing out.

Bull case:

• Destocking broadened across both ingots and billets, not just one segment.
• The ingot draw was demand-led, occurring without output cuts.
• The breadth supports a firmer price floor into Q4 2026.

Bear and cautionary case:

• The Xinjiang disruption and pre-holiday restocking both fade with the holiday.
• The billet overhang remains around 61,000 tonnes above 2023 levels.
• Cast alloy inventories kept building through September, signalling uneven demand.
• A sharp post-holiday price rebound risks re-triggering just-in-time hesitancy and renewed accumulation.

The bull case has a named anchor. JPMorgan Global Research, in its 21 August 2026 Metals Outlook, forecast a Q4 2026 aluminium price consistent with a firmer floor.

JPMorgan’s Q4 2026 aluminium forecast: $3,700/t.

JPMorgan’s $3,700/t Q4 target sits within a wider range of analyst views: the aluminium price forecast consensus for 2026 reflects divergent assumptions about Chinese restocking velocity, production curtailment risk, and the pace of energy cost normalisation across primary smelters.

Treat the six-year pattern as probabilistic, not deterministic. The record says inventories accumulate through the holiday week and billets draw down by week two, but the magnitude range is wide, and 2026’s elevated starting stocks widen it further.

Three variables to watch in the two weeks after Golden Week

1. The first post-holiday SMM inventory print (mid-October): The rate of holiday-week accumulation will confirm whether 2026 tracks the upper or lower end of the historical 4% to 17.3% ingot range.
2. Billet trajectory at the two-week mark: Historical precedent is unambiguous. Billets drew down by this point in all six prior years. A failure to do so in 2026 would be a meaningful deviation given the elevated starting stock.
3. Cast aluminium alloy trend: If builds continue post-holiday, the demand divergence across product segments is structural rather than seasonal, which weakens the broader bull case.

What the data can and cannot tell investors about the months ahead

The confirmed signal is real. The 2026 pre-holiday destocking is genuine, demand-led in its ingot component, and historically unprecedented in its two-week magnitude of 118,000 tonnes, making it a positive input for Q4 aluminium sentiment.

The confirmed limit is equally real. Starting stocks of 949,000 tonnes were the highest since 2020, billet inventory at 149,000 mt sits at a three-year seasonal high, and two of the three main drivers expire with the holiday itself.

The domestic Chinese inventory picture sits alongside a separate but related global signal: LME aluminium stocks reached their lowest level this century in 2026, a development that provides the international supply context against which China’s 949,000-tonne aggregate should be read.

What this leaves you with:

• The pre-holiday data sets a more constructive floor than the aggregate stock level alone suggested.
• The mid-October SMM print is the actual confirmation event, not the pre-holiday drawdown.
• Historical precedent says billets draw down by week two in all six prior years; watch whether 2026 follows.
• JPMorgan’s $3,700/t Q4 2026 forecast serves as the professional consensus anchor for the bull case.

The practical calculus is straightforward. The setup is more constructive than the headline stock levels suggested, but the six-year pattern says the real test of underlying demand comes in the second week after the holiday, not during it. Position sizing and timing for any aluminium exposure should wait for that October print.

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.