Tearing apart the aluminium backwardation

Aluminium.

The wires this morning are running variations on a theme: aluminium "rallied on supply concerns," which is the kind of phrase a journalist writes when they don't want to do the reading. There are no concerns. There is a backwardation of $95.50 a tonne on the cash-to-three-month spread. That is not a concern. That is the physical market grabbing the futures market by the lapels.

$95.50 per tonne is the tightest the spread has been since 2007. Not since the Russian invasion in 2022. Not since the energy crisis. 2007. That is the year the global aluminium market was last this physically squeezed, and almost nobody in your feed has mentioned the number.

A backwardation is the market's confession. It means somebody, somewhere, will pay almost a hundred dollars more for a tonne of metal today than they will pay for the same tonne delivered ninety days from now. They are saying: I cannot wait three months. The factory will be cold. The order will be late. Give me the metal now.

When that signal flares, you stop reading the price chart and start reading the loading docks. The chart is a lagging confession. The docks are the truth.

Story off. Numbers on.

Three things are happening simultaneously, and that's what nobody is putting on the same page.

First, the Gulf. Qatalum and Aluminium Bahrain — two of the largest non-Chinese smelters on the planet — declared force majeure last month. They cannot ship through the Strait of Hormuz at any reliable cadence. Together they normally feed something close to 2.5 million tonnes a year into the seaborne market. Right now they are feeding it whatever they can get past the disruption, which is materially less.

Second, sanctions. Since the LME stopped accepting Russian metal produced after April 2024, the warranted stockpile has been quietly bifurcating. Roughly half of what's still on warrant is Russian brand. The other half — the "deliverable" half that anyone outside Asia actually wants — has been getting picked off systematically. There have been no significant fresh non-Russian deliveries onto warrant since March.

Third, the smelters that could in theory help — the idled European and US capacity — can't. They went dark during the 2022 energy shock, and with power prices what they are post-Hormuz, they are not coming back. Restarting an aluminium smelter is not flipping a switch. It uses as much electricity as a mid-sized city. Nobody is paying that bill on a maybe.

It's a three-input squeeze with no relief valve. The LME is the pressure gauge, and it's reading red.

Now look at where the speculators are sitting.

Let that sit for a moment. The cash-to-three-month spread is at its tightest level since 2007. Two of the world's largest non-Chinese smelters are in force majeure. Stocks are at a three-year low. And the speculative community is, in aggregate, half a thousand contracts long. They might as well not be in the room.

Part of this is the LME warehouse game muddying the picture. Port Klang has been a stocks-finance theme park for years — Trafigura-style trades that warrant huge tonnages purely to capture rent-share economics, then cancel them again the moment the math flips. To anyone watching from screens, it looks like inventory is sloshing back and forth. To anyone calling shipping brokers in Valparaíso or Singapore, the deliverable pile is just getting smaller.

I have watched this exact pattern set up twice before — once in nickel in 2022, once in copper last year. Both times the speculative community was caught long-after-the-fact, chasing a price that had already moved. In aluminium, they aren't even short. They're just absent. Which means when the move comes, there's nobody to provide the first wave of buying. The physical hands have to do all the work themselves, and they will, and it will be ugly.

The way this resolves is mechanical. There are only so many ways a backwardation this tight unwinds, and "gently" is not on the list.

A prompt date passes. Somebody — a fabricator in Germany, a can-sheet plant in Tennessee — needs metal and discovers the warehouse they were planning to draw from has nothing usable left. They lift the offer. The next physical buyer behind them sees that and lifts higher. The CTAs, who are sitting on essentially zero exposure, watch their trend-following models flip from neutral to long inside a single session and pile in on top.

That's the move. Not a slow grind. A two-day step-change while everyone re-prices what they thought was abundance into what is actually a hostage negotiation.

Where the edge isn't: the broad mining ETFs, the diversified majors. Half their EBITDA is iron ore and coal, and they hedge their aluminium book a year out — which means they will sit out the spot move while collecting whatever they locked in last summer. Not interesting.

Where it is: the unhedged North American primary smelters and the secondary recyclers whose entire P&L moves with the Midwest premium. Companies like Kaiser. Century Aluminum. Real Alloy on the secondary side. Their revenue is a near-perfect spot proxy, and their cost base — power contracts, scrap input — is locked. Every dollar of premium expansion lands in the operating margin.

Physical layer says shortage. Paper layer says nothing. When those two disagree, the physical layer wins. It always has. The only question is whether the unwind happens this week, when the May prompt date hits, or in a month when somebody finally tries to deliver against a contract and finds the cupboard bare. Either way, $0.5K of net longs is not the position you want to be wrong from.