H1 2026: The Aluminum Market Doesn't Care Why You Built It

The First Half Nobody Priced In: Everyone Is Right. That's the Problem.

Here's the thing that's easy to miss if you look at H1 2026 region by region instead of all at once: Century and EGA breaking ground in Oklahoma, Magnitude 7 restarting New Madrid, Adani and IHC writing an $11.5 billion check in Odisha, Alcoa buying South32, Indonesia's entire Chinese-backed pipeline, none of these are the same decision wearing a different jersey. Indonesia's projects trace back to the bauxite export ban and downstreaming policy from years ago. Inola is a Section 232 and US industrial-policy story. Adani's bet is about India's long-standing aluminum deficit. Alcoa buying South32 is consolidation, not new supply. Australia's government packages exist to stop capacity from disappearing, not to add any. Every one of these makes complete sense on its own terms: resource security, tariff arbitrage, downstream integration, decarbonization, long-term demand, regional resilience. Nobody here is being reckless.

Six months ago, the swing factor everyone named for 2026 was Indonesia: a lot of new capacity landing fast, threatening to cap the price rally everyone wanted going into Q1. That call was directionally right. Nobody had "regional conflict knocks out two Gulf smelters and shuts the Strait of Hormuz for a quarter" on the board. H1 became a story of the conflict yanking supply out of the Middle East while a set of separately-motivated capacity decisions elsewhere kept moving on their own timelines, timelines the price spike mostly sped up rather than created.

That's exactly what makes this worth flagging. The Gulf war and the price spike it caused didn't invent most of these projects, for most players it mostly accelerated decisions and improved board confidence on plans that were already moving. But it's not true across the board. China's own 45 million tonne cap has been pushing capital out of the country regardless of any war, and the high margins this year's price spike created gave that capital a genuinely new reason to move now rather than later, new Chinese-backed smelter commitments, in Indonesia and elsewhere, that weren't on anyone's list before 2026. So, the picture is a mix: some capacity is a war-accelerated version of an old plan, some of it is a brand-new decision the price spike itself created. Either way, when a dozen separate, individually rational strategies, each answering a different regional question, all land supply in the same 18-24 month window, you get a classic coordination problem. Economists call it the fallacy of composition. Commodity markets have a simpler name for it: the cure for high prices. Nobody has to misread the market for the market to still end up oversupplied. The irony is that nobody has to be wrong for the industry to collectively create its next problem.

The price: $2,800 to $3,546 and back down to $3,100

LME aluminum was drifting toward $2,800-3,000 in January-February. Then March happened: Qatalum's controlled shutdown, Alba cutting 19% of capacity, and by March 19 a four-year high near $3,546. On March 28, attacks struck Al Taweelah and Alba's Bahrain site directly, and the metal gapped up again, with CRU floating an upside scenario north of $4,000. Roughly 3.5 million tonnes of Gulf capacity, about 5% of global output, became a question mark overnight.

Then came the ceasefire, and the market treated it like a done deal. It wasn't. The US and Iran signed an MOU on June 17, aluminum fell 16% in June, the steepest monthly drop since 2008, landing around $3,090-3,100 by early July. But the strait's status stayed genuinely contested through late June, with conflicting reports on whether it was actually open, and intermittent clashes kept happening. This is a truce being tested week to week, not peace, shipping volumes remain well below pre-conflict levels, and the UAE's own state oil company says full flows might not return until 2027. If this cracks again, there's more room to reprice upward than $3,100 suggests.

One more wrinkle: the delivered price tells a different story than the LME headline. Japanese buyers just settled Q3 MJP premiums at $395/mt, an 11-year high, up 12-13% from Q2. The US Midwest premium, which broke $1.00/lb in January, now sits around 110.25 cents/lb. Stack either premium on the LME base and the "aluminum crashed 16%" story looks a lot less true for anyone actually taking delivery. Regional premiums have stopped moving together, and Section 232, the Gulf disruption, and CBAM are why.

Indonesia: a policy story the war just sped up

Indonesia's build-out predates all of this by years. It runs straight back to the bauxite export ban and the downstreaming mandates that followed it, the same policy architecture that reshaped nickel. Inalum's existing Kuala Tanjung smelter runs around 275-300 kt, held back mainly by power constraints, not ambition. The bigger new smelter, 600,000 tonnes in West Kalimantan next to the Mempawah alumina refinery, is still years out: FID expected late 2025 into 2026, roughly three years of construction, commissioning targeted for 2028-2029. That's a genuinely different timeline than "Indonesia floods the market this year," and the near-term wave is coming from elsewhere in the country's pipeline.

That elsewhere is the Chinese-backed greenfield wave. Huachin runs about 500 kt against a potential 1 million. Adaro's Kaltara is targeting full 500 kt/yr by October. Tsingshan-Xinfa's Juwan missed its Q1 2026 target and was still only "approaching operational status" as of April. Several other Chinese-backed potlines are moving through commissioning at various stages across the same window. SMM's own numbers capture how fast this is moving: the firm projects Indonesia's operating aluminum capacity rising to around 2.51 million tonnes in 2026, up from about 870,000 tonnes in 2025, a near-tripling in a single year. SMM has also had to keep revising its own estimates upward mid-year, its May forecast for new primary aluminum capacity coming online outside China from 2026 onward was raised 61.3% from its January call, to 17.1 million tonnes from 10.6 million, as high prices pulled forward construction and commissioning schedules across the board. China's own producers have run out of room to grow at home, full-year 2025 output hit 45.02 million tonnes, technically over the 2017 cap, and April 2026 daily output set a record at 129,000 tonnes, so Chinese capital has been heading to Indonesia regardless of what the LME was doing in any given month.

None of that is new or reactive. What is worth naming is where it points: my own read is that this still ends in oversupply, not because Indonesia's strategy is wrong, downstreaming a resource you have in abundance is a completely sound national play, but because Indonesia isn't the only country running its own sound strategy on the same clock. SMM's own scenario work, presented at its ICM 2026 conference, captures the turn well: under all three of its geopolitical scenarios, the global primary aluminum market stays in deficit through 2026, by 1.06 to 1.34 million tonnes, before flipping into surplus in 2027, by 690,000 tonnes to as much as 1.42 million, as the wave of new capacity outside China lands. SMM's own price scenarios tell the same story from the other direction: H2 2026 LME centers around $3,600-3,850/t across its three cases, easing to $2,900-3,200/t in 2027 as the shortage premium gives way to surplus pressure. This isn't a demand-collapse story, consumption keeps growing. It's a timing story: several countries independently decided this was their moment to build, for reasons that have nothing to do with each other, and the sum of those decisions is set to flip the market from deficit to surplus in the space of about twelve months. MIND ID's own internal modeling reportedly flags oversupply risk further out too, by 2033 if Indonesia pushes its more aggressive targets, which is the country's own state miner acknowledging the composition problem from the inside.

The Middle East: the shock, and the actual restart numbers

The Gulf produces about 8-9% of world primary aluminum, nearly 80% exported through one chokepoint. EGA's Al Taweelah and Alba's Bahrain plant, roughly two-thirds of regional output, both took direct damage on March 28. Ma'aden played emergency alumina supplier just to keep the neighborhood's furnaces fed.

There's real data on the recovery now. On July 2, EGA reported 89 of Al Taweelah's 1,262 reduction cells restarted, ahead of its own schedule, first cell back May 26. The casthouse resumed May 4, the recycling plant restarted in early May, and first alumina production is expected early Q3. Even so, EGA is holding its own estimate that full recovery could take up to a year, a good reminder that rebuilding 1,262 individual electrolytic cells is slow and mechanical no matter how fast the diplomatic headlines move. Qatalum (never directly hit, running at 60% on a gas-supply scare) and Alba (directly struck, but with a track record of fast "crash-start" recoveries) are both effectively aiming for Q4 2026 full output. Alba didn't sit still while rebuilding either, in June it agreed to buy Aluminium Dunkerque, the EU's largest smelter, for roughly $2.2 billion.

Australia, the US, and India: three different problems, one crowded calendar

Australia's H1 wasn't about adding capacity, it was about stopping existing capacity from disappearing. All three coal-powered smelters have been negotiating their way out of expiring power contracts. Boyne landed a A$2 billion, 10-year government commitment; Tomago got its own federal-NSW rescue after a bumpy negotiation. These are decarbonization-and-survival deals, not growth bets, and they belong in a different category from everything else in this piece.

The US story is Section 232 doing what it was designed to do. Century restarted 50,000+ tonnes at Mt. Holly. Magnitude 7 announced a restart at its Marston, Missouri smelter (better known as New Madrid), adding 75,000 tonnes by year-end, though nobody's confirmed how it'll be powered yet, and the county had some of the worst sulfur dioxide readings in the country before it idled. The big one is EGA and Century's 750,000 tonne/year joint venture in Inola, Oklahoma, first new US primary smelter since 1980, a tariff-and-industrial-policy play more than a price-spike reaction. It hit real friction at the end of H1, an AG lawsuit, a 60-day construction moratorium over fluoride concerns, and still needs a power contract signed. Tariffs now split cleanly by origin: 50% standard, 25%/15% preferential for the UK, 200% flat for anything with Russian-smelted content, unchanged since 2023.

India is its own long-running story, not a new one. On July 2-3, Adani and Abu Dhabi's IHC signed an MoU with Odisha for an $11.5 billion complex, the biggest FDI India's metals sector has ever seen. This is about India's structural aluminum deficit, per-capita consumption still a third of the global average against a government demand projection that nearly triples by 2047, and Odisha's own bauxite reserves, not a reaction to a Middle East war. Nothing's broken ground yet.

And Alcoa chose consolidation over construction: a $4.1 billion deal for South32's bauxite, alumina, and aluminum assets across Australia, South Africa, and Brazil (Mozal excluded), signed June 30, closing expected H1 2027. It adds zero new tonnes to the market, just changes who owns the existing ones.

Where that leaves the second half

H1 2026 didn't create this capacity race. It simply made every existing investment case easier to defend. Indonesia kept building because of industrial policy. The US because of tariffs. India because of demand. Australia because of power. The Gulf because it had to rebuild. Different motives. Same outcome: more tonnes.

My base case is that the market tips into surplus over the next one to two years, and this time demand actually is part of the story, not disappearing, but growing slower than earlier expected in 2026 and 2027 as high prices themselves work against downstream buying, arriving right as the supply wave lands.

Every board approved a project based on its own balance sheet. The aluminum market, unfortunately, settles everyone's balance sheet together.

Every region solved its own problem. The global aluminum market inherits all of the solutions.

For Indonesia specifically, the second half is the real test of whether that call is right. The country has spent 2026 proving it can bring capacity online on a schedule that actually matters to the global balance, something it's talked about since the 1970s INALUM smelter and only now looks capable of pulling off. Whether it's remembered as a well-timed industrialization wave or as one strong link in a chain of separately-justified decisions that collectively overbuilt is probably the single most consequential open question heading into H2.

What to actually watch in H2

• Middle East restart pace. If EGA keeps beating its own schedule with 89 of 1,262 cells back as of July 2, the Gulf's return happens faster than the market is pricing.
• Whether Tsingshan-Xinfa actually hit their targets. One of them has already missed Q1. If the other smelter Q3 startup slips too, Indonesia's H2 wave is smaller than the headline pipeline suggests.
• Whether the Hormuz ceasefire survives Q3. The strait's status is still contested. A break would reprice the LME upward fast.
• The Inola power contract with PSO. Nothing about that project is real until this is signed, moratorium or not.
• China's inventory trend. Domestic stock is already at a six-year high. If it keeps building while output stays near record daily rates, that's an oversupply signal showing up before the LME reflects it.

The irony of H1 2026 is that everyone may have made the right decision. The market will decide whether they all made it at the same time.

The world isn't executing one aluminum strategy. It's executing five different strategies that all happen to produce aluminum.