Alcoa

Phase A — Understand the business

Lens 1 · Company Overview

Alcoa is the original integrated aluminum company — spun out of the old Alcoa Inc. as an independent public company on November 1, 2016 — and it is a pure upstream play: bauxite mining → alumina refining → aluminum smelting & casting → captive energy. It runs 25 operating locations across eight countries on five continents (primarily Australia, Brazil, Canada, Iceland, Norway, Spain, the US) and reports in two segments: Alumina and Aluminum.

How it actually makes money. Both products are commodities whose price Alcoa does not set:

• Aluminum is priced off the LME daily, plus a regional premium (the US Midwest premium being the one that matters most right now) plus a product premium for shape/alloy. Realized FY2025 price: $3,376/t (up from $2,841/t in 2024).
• Alumina is priced off the Alumina Price Index (API) — a blend of CRU, Platts and FastMarkets spot. ~95% of third-party smelter-grade alumina shipments are sold on an adjusted-API or fixed-spot basis. Realized FY2025 price: $415/t (down from $472/t in 2024, an ~11% API decline).

So Alcoa is a price-taker with two independent commodity exposures. The Alumina segment is its own biggest customer (Alcoa's own smelters took ~34% of total alumina shipments in 2025; ~two-thirds of alumina production is sold to third parties). End markets for the metal: transportation, building & construction, packaging, wire, industrial.

Customers/suppliers/competitors. Customer disclosure is thin — no single 10%+ customer is named, and the customers.csv is empty; sales go to fabricators and alumina/aluminum traders globally. Key suppliers are of caustic soda, calcined petroleum coke, lime, and — critically — energy (gas and power). Competitors: Rio Tinto, Norsk Hydro, South32, Emirates Global Aluminium, Rusal, and Chinese state-linked smelters (China = ~61% of global primary aluminum).

Contract structure. Mostly index-linked (API / LME + premiums), some multi-year offtake, some spot. The take-or-pay exposure runs the wrong way — Alcoa carries take-or-pay power-contract costs at previously closed sites as a recurring restructuring drag.

Lens 2 · Supply Chain

Map the chain with the named stakeholders (this is the lens that fails if it stays generic):

Upstream inputs →

• Bauxite: mostly self-supplied. Alcoa-operated mines produced 33.0 mdmt in 2025; partnership mines added 4.5 mdmt (equity basis) = 37.5 mdmt total; it had access to 43.2 mdmt across its portfolio and offtake. Named external bauxite source: Compagnie des Bauxites de Guinée (CBG) in Guinea, held via a 45% interest in Halco (Mining) Inc. (Halco owns 51% of CBG; Government of Guinea owns 49%).
• Caustic soda, calcined petroleum coke, lime, aluminum fluoride, cathode blocks, liquid pitch — bought competitively from multiple suppliers; caustic soda and calcined coke flagged as price-volatile. Strathcona calciner (Alcoa 39% / Rio Tinto 61%, Alberta) supplies calcined coke.
• Energy — the real chokepoint. Energy = ~24% of alumina refining cost; power = ~24% of primary aluminum cost. Alcoa self-generates only ~11% of its smelter power. Named suppliers: Hydro-Québec (three Québec smelters, contracts to 2029; Baie-Comeau renewal to 2036), Manicouagan Power LP (Alcoa 40% JV w/ Hydro-Québec), New York Power Authority (Massena West, new 10-yr renewable contract from Apr 2026), Landsvirkjun (Iceland, 40-yr contract to 2048), AGL Hydro / Origin / Alinta (Portland, Australia, swaps), Brazilian self-owned hydro (Barra Grande, Estreito, Machadinho, Serra do Facão).

→ Alcoa (refining + smelting) →

• 11.65 Mtpy consolidated alumina refining capacity (Pinjarra 4.7Mt, Wagerup 2.879Mt, São Luís 2.084Mt, San Ciprián 1.6Mt, Poços de Caldas 0.39Mt) — 1.0 Mtpy idle (San Ciprián 0.8Mt + Poços 0.214Mt).
• 2.645 Mtpy consolidated smelting capacity across 12 sites — 0.196 Mtpy idle (San Ciprián, Warrick, Portland, Alumar, Lista).

→ End customers: fabricators in transport/construction/packaging/wire, plus alumina and aluminum traders. Ball Corporation, Nexans, Unilever named as ELYSIS low-carbon-metal buyers.

Single-source / chokepoint dependencies:

1. Western Australia mining permits. The entire Pinjarra+Wagerup refining base (the cash core) depends on continued bauxite from the Huntly and Willowdale mines. Alcoa is operating under an 18-month national-interest exemption (granted Feb 2026) while a federal EPBC strategic assessment runs to August 2027; new mine regions (Myara North, Holyoake) won't start before 2029. Bauxite grade is expected to stay similar but the permitting overhang is real.
2. European gas. San Ciprián (Spain) refinery+smelter is a structural sore — gas now TTF-linked, smelter mid-restart, loss-making, 25% owned by Trento EQT.
3. JV partner Rio Tinto is omnipresent (Bécancour/ABI smelter, Alumar, Strathcona, ELYSIS) — a partner and a competitor at once.

Lens 3 · Competitive Advantages (moats)

Be honest: a commodity producer's moat is its cost curve, not its brand. Alcoa's durable advantages:

1. Vertical integration + scale. One of the world's largest bauxite/alumina producers (more so after buying out Alumina Ltd in Aug 2024 to take AWAC to 100%). Self-supplied bauxite and ~two-thirds-external alumina give it optionality most pure smelters lack. This is a real but modest moat — it smooths input cost, it doesn't confer pricing power.
2. Long-life, low-cost hydro power in Canada/Brazil/Iceland/Norway. Power is the swing cost in smelting; Alcoa's hydro-heavy footprint (Manicouagan, Brazilian consortia, Landsvirkjun to 2048) is a genuine cost-curve and carbon advantage as CBAM/Section-232-style carbon pricing arrives.
3. ELYSIS — the inert-anode, zero-carbon smelting JV with Rio Tinto + Québec. In Nov 2025 it started the first 450 kA inert-anode cell at Rio Tinto's Alma smelter; first production targeted 2027; Alcoa can take up to 40% of demonstration metal. This is the one potential step-change moat — process IP that, if it commercializes, resets the industry's carbon cost curve. It is also a decade-out option, not a 2026 earnings lever.
4. Captive Brazilian hydro + WA bauxite resource life to 2045 = durable, hard-to-replicate physical assets.

Bargaining power: weak over customers (price = LME/API, set by the market and China), moderate over suppliers for everything except energy and permits, where Alcoa is the supplicant (Hydro-Québec, NYPA, the Australian federal government, Spanish/Norwegian carbon-compensation regimes). Net: this is a cost-position moat, not a franchise moat. [Grounding: filings + positioning.md missing → web for industry structure.]

Lens 4 · Segments

segments.csv is empty, so all segment figures are from the 10-K MD&A tables.

Alumina segment (FY2025 vs FY2024):

Alumina EBITDA fell $526M (−37%) — lower API (−11%), Poços de Caldas ARO charges, higher caustic soda/lime, plus the Kwinana curtailment. Decelerating, price-driven.

Aluminum segment (FY2025 vs FY2024):

Aluminum EBITDA rose $401M (+61%) — realized price +$535/t driven by LME +9% ($2,614/t on a 15-day lag) and the Midwest premium +211% on Section 232; carbon compensation in Spain/Norway; cheaper Brazilian hydro — partly offset by Canada tariff costs and higher alumina input cost. Accelerating, premium-driven.

The structural read: the company's two segments are now counter-cyclical to each other on alumina price — when alumina is high, the Alumina segment wins and the Aluminum segment eats input cost; when alumina is low (as in 2025/26), the reverse. Total Segment Adj EBITDA was nearly flat ($1,940M 2025 vs $2,065M 2024) despite the violent mix shift. By product division, FY-equivalent revenue mix (Q1-26): Aluminum ~$2,582M, Alumina $521M, Bauxite $118M, Energy $77M. Geography: USD-denominated sales against AUD/BRL/EUR/CAD/NOK/ISK cost bases — a structural FX exposure.

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print: Q1 2026)

The print:

• Sales $3,193M (vs $3,369M Q1-25, −5% YoY)
• Net income attributable to Alcoa $425M (vs $548M Q1-25)
• GAAP diluted EPS $1.60 (vs $2.07 Q1-25)
• Total Segment Adjusted EBITDA $654M (Alumina −$40M; Aluminum $694M)

vs consensus: revenue MISSED ($3.19B vs $3.30B est) and adjusted EPS missed ($1.40 vs $1.47–1.48 est). Note the GAAP vs adjusted gap: GAAP diluted was $1.60, but adjusted (the number the Street tracks) was ~$1.40 — the GAAP figure was flattered by an $88M mark-to-market gain on the Ma'aden shares in Other income. Do not read the $1.60 as an operational beat.

What drove it: the Alumina segment swung to −$40M EBITDA (from +$664M a year earlier) on the collapsed API — this is the single most important line in the print. Aluminum carried the quarter at +$694M on premium strength.

Balance-sheet flags:

• Operating cash flow was NEGATIVE −$179M in Q1-26 (vs +$75M Q1-25) — working-capital build (receivables −$117M, inventories −$183M, payables −$195M). Q1 is seasonally weak, but the inventory build bears watching.
• Cash $1,353M (down from $1,597M at YE25); LT debt essentially flat at $2,442M.
• Inventories rose to $2,297M.

Market reaction: muted-to-positive — shares +0.5% after-hours to ~$70.71 on a stronger Q2 guide (BMO: "explainable" miss, no change to shipment targets, premium uplift to drive a "significant" Q2 earnings increase). The tape rewarded the forward (Q2 premium tailwind) and forgave the backward (alumina-driven miss).

Unusual vs own history: Alumina segment posting negative EBITDA is rare and signals the segment is near cash-cost at current API — the cyclical floor is being tested on the alumina side even as aluminum prints near highs.

Lens 6 · Earnings Calls (sentiment trend)

transcripts/ is empty, so this is web-sourced and necessarily lighter.

Tonal arc across the last several calls:

• 2024 (Kwinana curtailment, Alumina Ltd close): defensive — "portfolio optimization," cost-out, balance-sheet repair. Management in triage.
• FY2025 (full-year, Jan 2026): confident operational tone — "five smelters and one refinery set annual production records," "met the high end of adjusted net debt target," "strong operational performance." Pivot from survival to stability.
• Q1 2026: "resilience despite forecast miss," leaning hard on the Q2 premium tailwind and reaffirming 2026 shipment guidance.

Recurring phrases: "operational stability," "disciplined capital allocation," "strong balance sheet / low debt," "Alcoa Business System," "value-creating growth." Things they've stopped saying: the heavy 2023-24 curtailment/restructuring language has receded — the restructuring is now executed (Kwinana closed), not pending. Net sentiment: improving, operationally credible, but conspicuously quiet on what they'll do with the recovering free cash flow (no buyback talk — see Lens 9).

Lens 7 · Comps

EV/EBITDA derivation: market cap ~$17.5B + net debt ~$0.84B (LT debt $2,439M − cash $1,597M) − Ma'aden stake $1,397M ≈ EV ~$16.9B; ÷ FY2025 Segment Adj EBITDA $1,940M ≈ ~8.7×. On 2024 EBITDA ($2,065M) ≈ ~8.2×. On consensus 2026 net income ~$1.78B, the forward P/E at a ~$17.5B cap is ~9.8×.

Read: Alcoa screens expensive on trailing EV/EBITDA vs Norsk Hydro (even allowing the 2.9× looks too low to be clean) but cheap on forward P/E (~10×) if you believe the consensus 2026 net-income jump. The whole comp set is hostage to where you mark the cycle. Kaiser is a downstream specialty fabricator and a poor structural comp; Century is the closest US upstream peer but tiny. The honest comp is "Alcoa vs the LME forward curve," not "Alcoa vs a P/E."

Lens 8 · Stock-Price Catalysts (>5% moves, last ~5 years)

Mostly ``, labeled. Pattern recognition is the point.

• All-time-high $90.80 (Mar 24, 2022) — the post-COVID commodity supercycle + Russia/Ukraine aluminum-supply fear. Peak euphoria.
• 2022 H2 collapse — Q3-22 net loss of −$746M on falling LME/alumina + European energy-cost spike (San Ciprián gas). Stock roundtripped. Lesson: energy and price, not volume.
• 2023 trough — full-year net loss −$651M; the survival low.
• Aug 2024 — Alumina Ltd acquisition closes (all-stock, ~$2.7B), taking AWAC to 100%. Structural, not a pop.
• Late 2024 — alumina API hit an all-time high on supply disruptions; the Alumina segment carried 2024 earnings.
• Mar–Jun 2025 — Section 232 tariff escalation (Canada exemption removed; 25%→50% on Jun 4, 2025), driving Midwest premium +211%. This is the dominant 2025-26 catalyst. The stock is up ~184% over the trailing year to ~$72-77, near a 52-week high $75.70.
• Apr 2025 — ATO transfer-pricing case (>$800M in tax/interest/penalties) resolved entirely in Alcoa's favor — removed a tail risk.
• Mar 2026 — speculative "$68 target on Middle East energy" pieces — noise, not signal.

What the market actually reacts to for AA: (1) the Midwest premium / Section 232 policy (the current king variable), (2) LME aluminum and the API, (3) energy-cost shocks (European gas), and (4) portfolio/balance-sheet events (M&A, debt, the Ma'aden monetization). It does not react much to volume/shipment beats. This is a macro/policy stock wearing a company ticker.

---

Phase C — Judge people & books

Lens 9 · Management

CEO: William F. ("Bill") Oplinger — President & CEO since Sep 24, 2023. The archetype is unambiguous: a finance-and-operations lifer, not a founder. EVP & CFO of Alcoa Inc./Alcoa Corp from 2013–Feb 2023 (through the 2016 spin), then EVP & COO Feb–Sep 2023, then CEO. He owned the balance sheet for a decade before he owned the company.

1. Track record (quantified): On his watch the company has executed a credible turnaround off the 2023 trough — FY2023 NI −$651M → FY2024 $60M → FY2025 $1,157M; total debt reduced $147M in 2025, the 2027/2028 Notes tendered/redeemed (~$1.03B) and refinanced into 2030/2032 Notes; the company "met the high end of its adjusted net debt target" at YE25; the Alumina Ltd buy-in (his deal) simplified the structure to 100% AWAC; and the Ma'aden monetization turned a 25.1% Saudi JV stake into $1,350M of shares+cash. This is competent, disciplined portfolio surgery.
2. Tenure & skin in the game: ~2.7 years as CEO, ~13 years as a senior officer. Insider ownership: n/a (insider-transactions.csv absent). For a professionally-managed large-cap this is normal-low; do not expect founder-level alignment.
3. Capital-allocation history: The stated framework is balance sheet first, then sustaining capex, then "return cash in competition with growth". In practice 2025 capital went to debt paydown and the dividend ($105M) — not buybacks. The $500M repurchase program announced in 2021 appears dormant: Item 5 of the FY2025 10-K discloses only the quarterly dividend and makes no mention of 2025 repurchases. ROIC/ROE trend: negative (2023) → ~breakeven (2024) → positive (2025); a real inflection, but cycle-driven more than structurally engineered.
4. Red flags: None of the governance kind — no related-party self-dealing, no promotional behavior, no strategy whiplash. The honest critique is the opposite of promotion: management is conservative to a fault on capital return, sitting on a recovering FCF and a $1.4B Ma'aden stake with no articulated plan to return capital while the stock trades near highs.
5. Founder vs professional manager: Professional, finance-pedigree. Implication: expect disciplined, un-flashy stewardship — good for surviving the cycle, less likely to take the bold swing (or the aggressive buyback at the bottom) a founder-operator might.

Lens 10 · Forensic Red Flags

Acting as a forensic analyst. Across the three statements, grounded in filings:

• Revenue recognition: clean and low-risk — index-priced commodity sales, point-in-time transfer. Note the segment "purchases alumina/aluminum to satisfy customer commitments and bears risk of loss" — a trading book that adds gross-up volume; watch "Other segment items" (trading) which is sizeable (~$2.6B in Alumina) and lower-quality than production margin.
• Earnings vs cash quality — THE key flag. FY2025 net income $1,157M but a chunk is non-cash/non-operating: a $786M gain on the Saudi JV sale + $197M Ma'aden MtM gain ran through Other income (−$1,057M net). Strip those and underlying operating earnings are far smaller — FY2025 cash from operations was $1,185M, but it includes the working-capital swings and excludes the JV gain (correctly, as investing). The takeaway: GAAP net income materially overstates operating earnings in 2025, and again in Q1-26 (the $88M Ma'aden MtM). Always work from Segment Adjusted EBITDA ($1,940M) and operating cash flow, not the headline.
• Restructuring as a recurring item: "Restructuring and other charges" has been large and persistent — $184M (2023), $341M (2024), $918M (2025, incl. $856M Kwinana closure + $144M goodwill impairment). When "one-time" charges recur every year, they're an operating cost. The Kwinana closure carries ~$525M of further cash outlays through 2031 (~$120M in 2026).
• Balance-sheet liabilities to watch: Asset retirement obligations $1,120M and environmental remediation $206M (YE25) — both rising (ARO up from $691M in 2024 on Kwinana). These are long-dated, real, and growing as assets close. Also net derivative liability (power/aluminum hedges) of roughly −$1.5B net (FV deriv liabilities $567M current + $916M noncurrent vs $164M assets) — a large mark against equity via OCI.
• Pension/OPEB: $248M + $421M accrued (Q1-26) — manageable, declining.
• SBC: small ($13M Q1-26) — not flattering non-GAAP materially.
• Goodwill: $144M impairment in 2025 — already taken; residual goodwill risk modest for an asset-heavy miner.
• Working capital: Q1-26 inventory build (+$183M) and negative operating cash flow — seasonal but monitor for a demand-softening tell.

Net forensic read: accounting is conservative and transparent (clean auditor history since 2015; no aggressive rev-rec). The single discipline that matters is never anchoring on GAAP net income — 2025's $1,157M is ~$1B inflated by asset-sale and mark-to-market gains. Use EBITDA + OCF.

Regulatory findings (required):

• SEC Litigation Releases / AAERs: None. Verified via SEC EDGAR EFTS (LR + AAER) for 2021-06-18→2026-06-18: total_sec_findings: 0.
• 10-K Item 3 (Legal Proceedings): Routine for a heavy industrial — CERCLA/environmental remediation at multiple sites (disclosure threshold $1M), and asbestos premises litigation (subsidiaries as premises owners; "not expected to be material," covered by insurance). No securities/fraud actions.
• Tax — resolved favorably: the Australian ATO transfer-pricing dispute (>$800M claimed) was decided in Alcoa's favor (Apr 2025), not appealed, withdrawn.
• Non-SEC enforcement (web): No material FTC/DOJ/FDA/CFPB action found. The live regulatory exposure is environmental/permitting (Australian EPBC mining approvals; the two DCCEEW enforceable undertakings — A$55M/$36M in offsets, fully accrued; Poços de Caldas impoundment-stability remediation A$/$42M) — these are environmental-compliance costs, not enforcement scandals.
• Conclusion: No material regulatory or accounting-fraud findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-K Item 3 as of 2026-06-18. The genuine "regulatory" risk for Alcoa is environmental permitting (Western Australia), not securities law.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 / FY2027 / FY2028 EPS)

Built bottom-up from FY2025 actuals + 2026 company guidance, then triangulated against consensus. Output ``; share count ~262M.

Company 2026 operating guidance:

• Alumina production 9.7–9.9 Mt; shipments 11.8–12.0 Mt.
• Aluminum production 2.4–2.6 Mt; shipments 2.6–2.8 Mt.
• Capex $750M ($675M sustaining, $75M return-seeking).
• Full-year tariff cost on Canadian imports rising (Q2 +$35M sequential) but fully covered by Midwest premium at current pricing.

Macro inputs:

• LME aluminum: H1-26 testing ~$3,000/t, moderating to $2,700–2,800/t H2 as Indonesian capacity arrives; 2026 avg ~$2,945/t.
• Alumina API: ~$310–330/t, soft on Indonesian/Indian refinery additions.
• China at the 45.5Mt capacity cap (expected hit by Q2-26) → structural supply ceiling supporting LME.

Base case (FY2026): Aluminum segment carries the year on a high realized price (LME + an inflated Midwest premium) and full smelter restarts (San Ciprián, Alumar, Lista) adding volume; Alumina segment recovers modestly off the Q1 −$40M trough but stays soft on a weak API. Segment Adj EBITDA ~$2.4–2.7B. Net of ~$620M D&A, ~$160M interest, ~$200M restructuring/other, and a normalized ~20% tax → base EPS ~$6.30–6.90. Consensus net income ~$1.78B → ~$6.78 EPS on 262M sh — my base brackets it.

Bull case (FY2026): LME holds >$2,900, Midwest premium stays elevated (Section 232 intact), alumina firms back toward $400 → EBITDA ~$3.0B+, EPS ~$8.00+.

Bear case (FY2026): Section 232 is rolled back / Canada re-exempted → Midwest premium collapses ($/t premium reverts toward pre-tariff ~$0.20/lb), and LME fades to $2,600 on the Indonesian surplus, and alumina stays ~$310 → Aluminum EBITDA halves, EPS ~$3.00–4.00. This is the asymmetric downside and it is a policy variable, not a demand variable.

FY2027/FY2028: Consensus has net income plateauing/fading — ~$1.68B (2027) and ~$1.65B (2028), i.e. EPS ~$6.40 / ~$6.30 — implying the Street expects the 2026 premium tailwind to be the high-water mark and alumina to stay structurally soft. I concur directionally: 2026 is likely peak-of-cycle earnings unless ELYSIS or a supply shock changes the picture. EPS path: 2026 ~$6.6 / 2027 ~$6.0 / 2028 ~$5.5 (slightly below consensus out-years on premium normalization).

Per the watchlist unattended rules, no Brier forecast logged (forecast.ts create skipped).

Lens 12 · Bull vs Bear

Bull case. Alcoa is a deleveraged, simplified, vertically-integrated pure-play at a point in the cycle where (a) China's 45.5Mt cap structurally ceilings the world's largest producer, (b) Western supply is constrained and inventories are historically low, (c) Section 232 hands US-exposed producers a fat regional premium, and (d) the balance sheet is finally clean (net debt ~$0.84B, hit the high end of the net-debt target, ATO tail risk gone). On top of the cycle sit two free options: the $1.4B Ma'aden stake (monetizable from 2028) and ELYSIS zero-carbon smelting (first production 2027). At ~10× forward earnings with FCF inflecting, you're paying a cyclical multiple for a balance-sheet-repaired survivor with carbon-advantaged hydro assets and embedded optionality.

Bear case (2–3 ways it permanently impairs / de-rates).

1. Tariff dependency. ~60% of 2026 US aluminum volume is priced to a Midwest premium that is +211% because of a tariff. ~945kt is duty-paid-MWP-exposed. If Section 232 is softened, litigated away, or Canada re-exempted, a huge slice of 2025-26 incremental EBITDA vanishes overnight — and it's a stroke-of-the-pen risk, not a demand cycle you can model.
2. Alumina structural surplus. Indonesian + Indian + Chinese refinery additions are pushing API to ~$310-330 and the Alumina segment to negative EBITDA at the margin (Q1-26 −$40M). Half the company is in a structural-oversupply commodity.
3. No pricing power, ever. It's a price-taker on two commodities set by China and the LME, with a USD-revenue/local-cost FX mismatch and rising ARO/environmental liabilities ($1.1B+ ARO, growing). A commodity producer cannot out-earn its cost curve through a down-cycle.

Pre-mortem (18 months out, thesis broke — what happened?): Section 232 was rolled back or carved out for Canada in a 2026/27 trade deal; the Midwest premium normalized; the Indonesian smelter wave + China demand softness pushed LME back to ~$2,500; alumina stayed in surplus; FCF that looked like $2B in 2026 prints closer to $0.7B in 2027; the stock that ran +184% to near-highs gave back half. No fraud, no blow-up — just the policy and the cycle both turned, and a price-taker has no defense.

Are multiples too high? On trailing EV/EBITDA (~8.7×) for a cyclical at a cyclical peak, yes — that's a rich entry. On forward (~10× P/E) it looks fair-to-cheap, but only if you trust 2026 as a clean run-rate, which the bear case says you shouldn't. The multiple is fine; the E is the problem — it's borrowed from a tariff.

Contrarian view (what the market is refusing to see): The bulls are extrapolating the Midwest-premium windfall as if it were operational improvement. It isn't — it's a transfer payment from US aluminum consumers to US-exposed producers, courtesy of trade policy, and it is the most mean-reverting line in the model. The durable story isn't the premium — it's carbon-advantaged hydro smelting + ELYSIS as the EU CBAM and carbon pricing make dirty Chinese metal uncompetitive in Western markets over the decade. The market is paying for the wrong tailwind.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• What structurally breaks the money machine: Repeal/erosion of Section 232 on Canadian aluminum. That single policy is doing the heavy lifting in 2025-26 EBITDA. It is not a moat — it's a subsidy, and subsidies get traded away in the next negotiation. A US-Canada/USMCA deal that re-exempts Canadian metal would gut the Midwest premium and a large chunk of Aluminum-segment margin.
• Revenue concentration / what shifts: Concentration isn't by customer (no 10% name) — it's by commodity and by geography. ~60% of aluminum volume keys off a US premium; the alumina book keys off a Chinese-supply-set API. Two independent macro bets, both of which can move against Alcoa at once (exactly the 2027 pre-mortem).
• Why the moat is weaker than bulls think: "Vertical integration" doesn't stop the LME from falling. The genuine edge (hydro power) is real but partial — Warrick burns coal, Spain burns gas, and ~89% of smelter power is purchased and exposed to regional power markets and contract roll (Hydro-Québec to 2029, Portland swaps to 2026/expiring). The cost-curve advantage is mid-pack, not top-decile.
• Most dangerous competitor bulls underestimate: Chinese state-linked smelters at the 45.5Mt cap — bulls treat the cap as bullish (supply ceiling) but it's also a reminder that China is 61% of the market and sets the marginal price; any Chinese demand wobble (property, exports rerouted via Indonesia) floods the seaborne market. Also Indonesia, adding cheap new capacity that exports 70% to China — the surplus that's already crushing alumina.
• Worst capital-allocation choices: Hard to find malfeasance — but the dormant $500M buyback while sitting on recovering FCF + a $1.4B Ma'aden stake near a 52-week high is a passivity critique: management is hoarding optionality instead of returning capital when the stock is strong, which may look foolish if the cycle turns before they act.
• Assumptions that must hold for today's price (~$72): Section 232 intact through 2026-27; LME ≥$2,800; alumina recovering off the floor; San Ciprián restart completing without further European energy shocks; WA mining exemption extended past Aug 2027. Break any two and the thesis cracks.
• What if growth disappoints 20–30%: A ~25% haircut to 2026 EBITDA (premium fades + LME softens) takes EBITDA from ~$2.5B to ~$1.9B and EPS from ~$6.6 to ~$3.5–4.0; at an unchanged ~10× the stock is a $40-45 stock, ~40% downside.
• Single scenario that permanently impairs: A simultaneous Section 232 rollback + sustained alumina surplus + LME mean-reversion to $2,400-2,500 — not catastrophic (Alcoa survives; balance sheet is clean) but it re-rates the equity back to mid-cycle trough multiples on trough earnings. Plausibility: moderate-to-high over 2-3 years — the tariff is the explicit swing factor and trade policy is volatile.

Lens 14 · Management Questions (ordered by information value)

1. What is your explicit base/bull/bear assumption for the US Midwest premium and Section 232 in 2026-27, and how much of guided EBITDA evaporates if Canadian metal is re-exempted? (The whole thesis.)
2. With FCF inflecting, the balance sheet at the net-debt target, and the stock near a 52-week high — why no buyback, and what specifically would trigger reactivating the $500M program vs. paying down more debt?
3. When and how do you monetize the $1.4B Ma'aden stake (transferable in thirds from 2028), and what's the intended use of proceeds — debt, buyback, growth?
4. What is your sustaining cost per tonne trajectory for alumina and aluminum, and where do you sit on the global cost curve by quartile after Kwinana's closure?
5. What is the realistic commercialization timeline and capital requirement for ELYSIS beyond the 2027 first-production milestone — when does it move EBITDA, not just the carbon story?
6. Walk through the Western Australia EPBC strategic-assessment risk: what's your contingency if Ministerial approvals for Myara North/Holyoake slip past 2029, and what happens to bauxite grade/cost?
7. How much of the alumina segment is below cash cost at $310-330 API, and at what API do you curtail further refining capacity?
8. What's the all-in cash cost and completion risk on the San Ciprián restart, and at what European power/gas price does the complex become structurally uneconomic again?
9. How should we normalize 2025 GAAP earnings — what is your own "underlying" net income excluding the Saudi gain and Ma'aden marks?
10. What is the expected cadence of ARO/environmental cash outlays (Kwinana ~$525M to 2031, Poços, Suriname/Point Comfort legacy) over the next five years?
11. How do you think about acquisitions vs. organic growth now that the portfolio is simplified — any appetite, and what return hurdle?
12. What's your hedging philosophy on LME, the Midwest premium, and FX (USD revenue / AUD-BRL-EUR-CAD-NOK cost), and how much of 2026 is hedged?
13. How exposed is the cost base to caustic soda and calcined petroleum coke price spikes, and are those single-source anywhere?
14. What does EU CBAM full implementation (Jan 2026) do to your competitive position in European markets vs. Chinese/Russian metal?
15. What's the succession depth below a finance-pedigree CEO — who runs operations, and how deep is the bench for the next down-cycle?

---