The aluminium illusion: how tariffs made America weaker in a metal it needs most

ALUMINIUM IS the quietest metal in the Pentagon inventory and, as of this week, its loudest liability. Last summer some 80 government officials and industry executives gathered in Washington to run a war simulation. The exercise asked how the American aluminium supply chain would hold up in a major conflict. High-purity aluminium - an ultra-refined grade, stripped of iron and silicon, essential for fighter-jet airframes, armour plate and rocket casings - emerged as the weakest link. Seven months later, the United States attacked Iran. In March, Iranian drones struck Emirates Global Aluminium's smelter in Abu Dhabi and Aluminium Bahrain's complex south of the Gulf. The war game's warning has come home.

The structural irony is that the vulnerability was not merely neglected but deepened by policy. American tariffs on aluminium, first imposed in 2018 and hiked to 50% in 2025 after Canadian exemptions were revoked, were designed to protect domestic production. They did the opposite. Major Canadian producers, long integrated into the American defence supply chain, diverted exports to Europe. Aluminerie Alouette shifted its European sales from 4% of output to 57% in a matter of months. The United States, having pushed away its most reliable supplier, turned instead to the Middle East. The UAE and Bahrain together came to account for nearly a quarter of American unwrought aluminium imports. It is hard to construct a more precise self-sabotage.

The numbers are stark. The sole large-scale American producer of high-purity aluminium - Century Aluminium's Hawesville smelter in Kentucky - shut down in 2022, succumbing to soaring energy costs. Hawesville had the demonstrated capacity to produce at least 100,000 tonnes a year. The Commerce Department's Section 232 investigation put defence requirements at roughly 25,000 tonnes annually, or about 10% of total US demand for the grade, which it pegged at 250,000 tonnes a year. Defence needs therefore represent a far heavier share of the limited domestic supply of high-purity aluminium than of overall aluminium consumption. With Hawesville dark, the gap fell to imports. Emirates Global Aluminium, the dominant supplier, provides some 75,000 to 85,000 tonnes a year for military purposes, according to people with direct knowledge. The UAE supplies roughly 90% of American imports of that grade.

Iran's strikes on March 28th forced EGA's Al Taweelah site - one of the world's largest aluminium complexes, with a capacity of 1.6m tonnes - into emergency shutdown. Restoring it requires rebuilding each of 1,262 reduction cells individually. As of early July, only 89 had been restarted. EGA says full recovery could take a year. Aluminium Bahrain, already running at reduced capacity, also declared force majeure. LME aluminium prices climbed to a four-year high of $3,687 a tonne in April. The peace deal in mid-June eased them to around $3,380, but they remain well above pre-war levels. LME warehouse stocks have fallen roughly 60% since May, to 418,675 tonnes. The Strait of Hormuz closure complicated outbound logistics for months. Though it has since reopened, the physical damage to Gulf smelters cannot be negotiated away.

To be sure, the Pentagon has not sat idle. In 2023 the Department of Defence awarded Arconic, an Apollo Global Management-backed supplier, $45.5m under the Defence Production Act to expand high-purity aluminium production in Iowa. The $57.5m project, completed last autumn, doubled domestic capacity at Arconic's Davenport Works to roughly 22,000 tonnes a year. That is progress. It is also too little, too late to cover a structural shortfall that runs far larger than a single facility can fill, particularly while the broader smelter base has been eroding. The Pentagon invested in Constellium's Muscle Shoals, Alabama plant in 2024 for aerospace castings, a similar gesture of targeted support. Both are real. Both are marginal against the scale of the dependency.

The deeper problem is not the absence of warnings but the mismatch between rhetoric and mechanism. The Trump administration's tariff wall was built on the logic that trade barriers would rebuild American industrial capacity. They instead created a constellation of second-order effects: higher domestic prices that penalise downstream manufacturers from automotive to food packaging; alienated allied suppliers whose cheap hydroelectric power makes them genuinely competitive; and a new, geopolitically fragile concentration of imports precisely where conflict was most likely. The Midwest premium - the surcharge American buyers pay above the LME benchmark - more than doubled after the 50% tariff took effect. American firms now pay substantially more for aluminium than competitors abroad, while receiving less secure supply. That is a bad bargain.

Three companies stand in the frame. Kaiser Aluminium, Constellium and Arconic are the top three providers of high-purity metal to defence contractors including Lockheed Martin and Boeing. Their exposure to Gulf supply, their capacity to pivot to alternative sources, and the extent of Pentagon demand guarantees on their books will determine who benefits from the scramble that follows. Arconic's DPA backing gives it a structural advantage. Constellium, a French firm with US investment from the Pentagon, holds a similar position. Kaiser's Tennessee operations are the wildcard. For investors, the relevant question is not whether these companies will sell more aluminium - they will. It is whether margins expand sustainably or whether the war premium is a one-off cost pass-through.

The broader lesson is institutional. A defence industrial base that cannot supply the base metal of modern weapons is not a base at all; it is a shopping list. The Pentagon's war game identified the vulnerability seven months before the actual crisis. The analysis of the simulation, published only this week, confirms that officials saw the problem and did not act with anything approaching the urgency it warranted. That is not a story about military incompetence. It is a story about political incentives. Tariffs are popular with the voters who care about the aluminium industry because they appear to protect American jobs. They are unpopular in the places that actually matter for aluminium - the foundries, extrusion plants and aerospace fabricators downstream - precisely because the costs are diffuse and the beneficiaries are loud. The politics rewarded the illusion of self-sufficiency while the supply chain grew thinner.

The better answer is not more tariffs. It is an off-ramp from the current framework. Restoring tariff exemptions for Canadian aluminium would immediately reduce reliance on the Gulf without requiring the years needed to build new smelters. It would also repair an alliance that the 2025 policy treated as an adversary. On the domestic front, the single operating alumina refinery in the United States - the intermediate product from which primary aluminium is smelted - represents a single point of failure that no amount of war gaming can fix. Building a few more smelters is the right long-run goal, but smelters take a decade and require guaranteed, low-cost electricity. Energy policy, not trade policy, is the binding constraint.

The war in Iran has exposed a defect that was structural, not situational. The danger is not that the United States will run out of aluminium tomorrow. It is that the political system has learned to mistake the appearance of protection for the reality of it. Consumers have paid the tariff. The defence industrial base has paid the tariff. The only thing the tariff has not achieved is the thing it promised. Better to start correcting that before the next war game turns out to be accurate again.