Alcoa深度分析：铝业务盈利强劲但氧化铝亏损，管理层举债收购

What the business is and why it earns

Alcoa does three things. It digs bauxite, refines it into alumina and smelts alumina into aluminum. The 10-K says most of its bauxite and alumina are sold under multi-year contracts, and its aluminum under annual or multi-year delivery contracts. I had to read that part twice, the contracts fix volume, not price. Aluminum is priced off the London Metal Exchange price plus regional premiums and although it knows who will take its tonnes, it does not know what it will be paid for them.

Therefore, when metal prices rise, Alcoa earns a great deal as there costs of operation remain relatively stable/fixed regardless. In the June quarter the aluminum segment made $1,073 million of segment EBITDA, against $97 million a year earlier, however some of that was Alcoa’s own work. Restarts at San Ciprián, Alumar and Lista lifted production 5% from the previous quarter, incentivised through higher prices. The restarts have also had a human cost with a worker dying at Alumar in September during the ramp-up.

The one thing the headlines miss

The aluminum segment is carrying the whole company, yet in the same quarter, alumina lost $96 million of segment EBITDA. Alcoa’s refinery cost is about $368 a tonne, and it realised about $334, that half of the business is selling below its own cost. Management wrote the alumina unit’s goodwill down to zero in late 2025, which is a formal admission that it is worth less than its book value.

Trailing owner earnings are about $1.16 billion, while a normal year is about $390 million, therefore around two-thirds of today’s cash earnings comes from the cycle rather than the business. Over ten years the owner-earnings margin has swung from -7% to 7.7%, A thoughtful owner can guess roughly what Alcoa will produce in a decade, but knowing what it will earn is much more harder to say.

The risk, and the counter-case

The risk is the balance sheet in a bad year. My platforms survival test, sets $5.5 billion due within two years against about $2.5 billion of resources available in a bad year. That is a $3.0 billion shortfall before any deal debt, net debt was already 2.8 years of normal owner earnings. Management quotes pro forma leverage of about 1.5 times, but that is measured against peak-cycle EBITDA, the new notes add about $175 million of annual interest. There is also a dependence that is easy to overlook. One state government, Western Australia, decides through ministerial approvals what grade of bauxite feeds the Pinjarra and Wagerup refineries.

However, the other side deserves a fair hearing, the new notes run to 2034 and 2036, which is a real answer to refinancing risk. The bull case is coherent too. Suppose China holds its 45-million-tonne cap, supply from Indonesia and the Middle East stays constrained, and ex-China alumina tightens in 2027 as management has stated they expect. Then something close to today’s $1.16 billion becomes the normal year rather than the peak, and buying alumina now looks shrewd. I personally don’t know which way that goes, nobody does yet in my opinion but don’t quote me on that.

What management is doing with the money

The record gives me pause. Alcoa spent $715 million on buybacks from 2018 to 2024, and our desk reckons $708 million of that value was lost by buying near the cycle top, typical value destruction for shareholders. The all-share buyout of the Alumina Limited minority lifted diluted shares from 178 million to 263 million. Owner earnings per share fell from $5.03 in 2021 to $1.33 in 2025.

AliGroup is the largest bet yet. It costs $3.1 billion in cash plus the shares, plus a contingent payment of up to $750 million that hands the seller much of the price upside. Promised integration savings are $50 million a year, so the deal depends on the assets themselves. Kwinana was closed, at a cost of $856 million in charges. Meanwhile, the Ma’aden stake became $1.36 billion of shares, though they are locked up until July 2028. Asset sales target $500 million to $1 billion by 2030, and Massena East could bring up to $315 million, with $50 to $65 million upfront.

What I am watching

• November 13, 2026: whether the $175 million receivables facility is renewed on similar terms, just as the deal cash goes out.
• Early 2027: the Western Australian decision on Myara North and Holyoake, which management expects within the first few months of next year. The question is whether the conditions restore bauxite grade.
• Alumina against cost: whether the API stays below Alcoa’s roughly $368 a tonne refinery cost into the first half of 2027.
• Two quarters after AliGroup closes: whether net debt falls relative to normal owner earnings, and whether the fall comes from the business or only from the metal price.