Aluminium Bahrain B.S.C. (ALBH)

Primary aluminium is a global commodity with one dominant producer nation and a hard ceiling on its output. China holds roughly 60% of world smelting and is capped by policy at 45 million tonnes a year, a limit it has now reached. Incremental demand must therefore be met outside China, where the marginal tonne is expensive.

Two policy changes matter. The EU's Carbon Border Adjustment Mechanism entered its definitive phase in January 2026, pricing embedded carbon in imported metal and separating low-carbon from high-carbon supply. And US tariffs lifted the Midwest premium to unusual levels. Alba benefits from both: certified under the Aluminium Stewardship Initiative, platinum EcoVadis rated, selling a low-carbon line branded EternAl, and its US operation was around US$20m net positive across 2025 as premium gains exceeded tariff cost.

Alba's peers are mostly unlisted, state-owned, or diversified across bauxite and downstream. Pure-play listed primary smelters are rare, which is part of why Alba's multiple is low. Among GCC producers, Alba came through the 2026 conflict curtailed but physically intact, while EGA's Al Taweelah suffered damage guided at up to a year.

Moat verdict: narrow, and contractual rather than structural

Alba has no brand, no switching costs and no network effect: it sells an exchange-priced commodity. What it does have is a real low-cost position resting on three things. Scale, since one 1.6mt site spreads fixed cost further than any competitor's. Self-generated power, with PS5 Block 4 lifting the complex to 2,481 MW and cutting emissions intensity by 0.5 tonnes of CO2 per tonne. And a ten-year fixed-price gas contract with Bapco Upstream.

Narrow rather than wide, because the most valuable of the three, the gas price, is not a competitive achievement. It is a contract with a government-owned counterparty under the same ultimate ownership as the 69% shareholder, in a state the IMF has urged to cut subsidies and levy corporate tax. Management confirmed in February 2026 that no price-increase notification had been received: reassurance about the present, not about renewal.

Alba spent five years deleveraging and finished the job in 2025. Net debt fell from US$2,218m in 2021 to US$564m, cash rose to US$409m, net debt to EBITDA reached 0.52x and the equity ratio reached 80% of total assets. On 31 December 2025 this was one of the strongest balance sheets in the sector.

And then the company decided to spend it

The US$2.2bn Aluminium Dunkerque acquisition is to be fully financed by a consortium of Alba's banking partners, with Bpifrance contributing EUR 100m for 6% of the target rather than of Alba. On the last verified balance sheet that takes net debt from US$564m to roughly US$2.76bn, about 2.56x FY2025 EBITDA. Five years of deleveraging is committed in one transaction, and the debt quantum, tenor, pricing and covenants are undisclosed.

Alba does not publish a debt maturity schedule and none could be verified this run. What is visible: total assets rose 9% to BHD 2,867.2m in the half while equity attributable to owners rose 4% to BHD 2,163.5m, so total liabilities, taken as the difference, rose from BHD 537.7m to BHD 703.7m, up 31% in six months.

Distribution, and whether it survives the acquisition

Alba pays semi-annually against a stated 35% payout ratio, described by the chief financial officer as a benchmark for the smelting industry. The 2025 final dividend was 43.51 fils, BHD 61.6m or US$163.9m, paid on 8 April 2026. Cash returned during 2025 was US$140m against US$765m of free cash flow, covered 5.5 times. On capacity the dividend is not at risk.

The risk is priority. Asked on the FY2025 call about a higher or special dividend, the chief financial officer said Alba had projects ahead and would assess the situation. Two weeks later it entered exclusivity on Dunkerque. Anyone buying this for yield should treat 35% as a ceiling while the acquisition and the New Replacement Line compete for the same cash.

This is the cleanest part of the story. No dilution, no convertible, no scrip, no buyback. The corollary: Dunkerque is funded entirely with debt, protecting per-share earnings while transferring risk to the balance sheet. For a minority holder in a cyclical, that is a trade, not a gift.

Framework: mid-cycle earnings power on a mid-cycle multiple. A discounted cash flow is the wrong tool here. Alba's cash flow is a levered function of one exchange price that moved 46% year on year, and discounting a forecast built off a war-inflated spot produces a spot-price bet with a decimal point on it.

Two observations. A company earning 10.9% on equity and trading at 0.61x book is priced as though that return will not persist, which for a cyclical is reasonable. And the trailing 4.2x is flattered by peak prices and curtailment-driven inventory drawdown, so the honest comparison is 5.7x normalised, not 4.2x trailing.

The bear case, argued properly

The peace dividend is the problem. Q2 earnings were made by a US$3,576/t LME that existed because Gulf supply was missing. Spot has already fallen from US$3,855/t on 2 June to US$3,260/t on 3 August. If Hormuz normalises and EGA and Qatalum restart, roughly 9% of world supply returns to a market also absorbing Indonesian commissioning. Alba gets its volume back into a price that no longer pays for it. Committing US$2.2bn of bank debt at the top of a price cycle, for capacity in the highest-power-cost region on earth, is the textbook cyclical error, and strip out fixed-price gas and Alba is a mid-cost smelter with no upstream integration.

The bull case, argued properly

The deficit outlasts the war. China is capped at 45m tonnes and has reached it. CBAM now structurally favours certified low-carbon metal, which Alba sells, and demand is growing from grid, transport and data centres. At 3.8x trailing EV/EBITDA and 0.61x book, the price already discounts a substantial reversion, and the bear case lands at roughly half today's price rather than at zero. Dunkerque could be the re-rating rather than the risk: two continents, European customers and a French state co-investor resolve the concentration that most caps the multiple.

Invalidation: what tells a holder they are wrong

• LME cash closes a month below US$2,600/t without Gulf supply having returned, which would say the deficit was thinner than believed.
• Dunkerque closes with debt above US$2.4bn, a tenor under five years, or a covenant tighter than 3.5x.
• No confirmed Lines 1 to 3 restart by the Q3 results on 10 November 2026.
• Any notification of a gas price increase, which removes the only durable advantage in this note.
• A 2026 interim dividend below the 35% payout policy.

Built from Alba's Q2 and H1 2026 release (4 August 2026), the Annual Report 2025, the FY2025 investor presentation and webcast transcript, the Q1 2026 release and Dunkerque announcement, Bahrain Bourse major shareholder disclosure, S&P Global Commodity Insights and AL Circle, with market data taken on 6 August 2026. Alba reports under IFRS with a 31 December year end. Conversions use the Central Bank of Bahrain peg of BHD 0.376 per US dollar. Multiples, per-share values, the fair value range, the sensitivity grid and the scenario returns are our own calculations.

What desk research cannot answer: the quantum, tenor, pricing and covenants of the Dunkerque financing; Aluminium Dunkerque's standalone earnings and power cost structure; the plant's physical condition after the March attack and the cost of restarting Lines 1 to 3; the renewal terms of the Bapco Upstream gas contract; and the average daily traded value of the free float. That is where the remaining edge sits.