Durability

Durability

Alcoa clears one of the framework's five conviction sources — capital intensity — and fails or only partly meets the other four. It produced 3.1 percent of the world's primary aluminum and 6.4 percent of its alumina in 2025, at prices it does not set [1]. Physical output of alumina is down 27 percent in nine years, and the segment sold below cash cost in the first half of 2026 [2].

Sources: derived from the FY2025 Form 10-K Statement of Consolidated Cash Flow and Statement of Consolidated Operations [3] [4], the FY2025 alumina production table [5], and the 2017 Form S-1 selected historical combined data [6].

Conviction Sources

The framework builds year-10 conviction from five specific places. Graded against Alcoa's own record rather than against the aluminum industry in the abstract, one of the five holds cleanly.

Sources: FY2025 Form 10-K, Item 1 Business — Competition [7] and Item 1A Risk Factors [8] [9]; Aluminum Corporation of China FY2025 Annual Report [10] [11]; Q2 2026 earnings presentation [12]; Q4 2025 earnings call [13]; Form S-1A [14].

Market structure. The Business tab establishes what Alcoa does; what matters for year-10 conviction is what share it holds and whether that share is stable. World primary aluminum output was 74.52 million tonnes in 2025, of which China produced 44.23 million — 59.4 percent [15]. World alumina output was 150.49 million tonnes, of which China produced 92.94 million — 61.8 percent [16]. Against those denominators Alcoa's 2,319 kmt of aluminum and 9,640 kmt of alumina work out to 3.1 percent and 6.4 percent of world supply [17] [18]. This is not a monopoly, a duopoly, or an oligopoly.

The strongest version of the opposite case is regional, and Alcoa states it: it is "the largest alumina producer outside of China and the largest supplier of third-party alumina outside of China," with a first-quartile cost position on the global alumina curve in 2025 [19]. On the ex-China denominator (57.55 million tonnes), Alcoa's 9,640 kmt is 16.8 percent — a real position. But the alumina market is not segmented: China exported 2.55 million tonnes in 2025, up 42.5 percent, and net-exported 1.35 million [20]. A regional leadership share that prices off a global index is a cost advantage, not a market structure.

Share has also moved the wrong way. Alcoa's alumina output was 7.1 percent of world supply in 2024 (10,034 of 141,570 kmt) [21] and 6.4 percent in 2025 — the denominator grew 6.0 percent while Alcoa's numerator fell 3.9 percent [22].

Regulatory entry barriers. In the framework's canonical form — a regulator that will not let a garage startup take share — no such regime protects Alcoa. The regulatory regimes that bind here run in the opposite direction: Alcoa's Western Australian bauxite mining requires state and federal environmental approvals it does not yet hold for its next major mine regions [23]. Trade policy is a genuine tailwind, and a reversible one: the Section 232 tariff on Canadian aluminum went from exempt to 25 percent in March 2025 and to 50 percent in June 2025 [24]. Tariff regimes set by proclamation are not entry barriers in the sense the framework means; they are policy positions with a ten-year half-life at best.

Capital intensity. This is where Alcoa's case is strongest and it is genuinely strong. Building a smelter costs $7,500 to $9,100 per tonne of annual capacity in North America or Europe, $4,500 in the Middle East, $2,100 in Indonesia or India, and $1,150 in China [25].

Source: Q2 2026 earnings presentation, expansion capex intensity slide; the North America and Europe bar plots the $7,500–$9,100 range at its midpoint [26].

Management's own behaviour confirms the barrier. Asked in January 2026 about greenfield expansion, the CEO said Alcoa has no greenfield plans for aluminum and has "not found anywhere around the world that provides a sufficiently low energy price for sufficient returns on a greenfield plant," with refining in a similar position [27]. That protects the incumbent asset base. It also says the incumbent cannot profitably grow it, which is the same fact seen from the other side.

Essentialness. Aluminum is essential and its demand is growing: management expects primary aluminum demand outside China to rise by roughly 7 million tonnes and alumina demand by roughly 18 million tonnes over the next decade [28]. What does not follow is that Alcoa's revenue is defensive. Revenue fell 22.2 percent in FY2019 and another 11.0 percent in FY2020, and 15.3 percent in FY2023 — with volumes broadly intact each time. The 10-K is explicit that demand for aluminum "is also highly correlated to economic growth" and that profitability "is subject to significant fluctuation" [29]. The metal is essential to the customer; the revenue line is not defensive for the shareholder.

Operating history. The Alcoa name reaches back to 1888 and the predecessor "pioneered the aluminum industry over 128 years ago" [30]. Alcoa Corporation, the entity being underwritten, became an independent public company on November 1, 2016 [31]. Its capital structure, portfolio, and management are nine years old and have been through one commodity cycle. The framework's 30-to-50-year test is met by the assets and the metallurgy, not by the security.

The Shrinking Physical Base

Revenue rose 37.7 percent between FY2016 and FY2025 while the physical base shrank. Alumina production fell from 13,251 kmt to 9,640 kmt, down 27.3 percent; bauxite from 45.0 to 37.5 million dry metric tons, down 16.7 percent; primary aluminum production was roughly flat at 2,368 to 2,319 kmt, while total aluminum shipments fell from 2,953 to 2,522 kmt.

Source: reported segment production data for the full FY2016–FY2025 series, of which FY2016–FY2019 and FY2022 come from 10-Ks not held in this corpus; FY2020–FY2021 alumina production per the FY2021 Form 10-K segment table [32] and FY2024–FY2025 per the FY2025 Form 10-K segment tables [33] [34].

Capacity tells the same story with less noise. Consolidated smelting capacity was 3,401,000 metric tons per year at December 31, 2015 [35], 2,962,000 at the end of 2021 [36], and 2,645,000 at the end of 2025, of which 196,000 sat idle [37]. That is 22 percent of smelting capacity retired in a decade. Refining capacity fell from 13,843 kmt at the start of 2025 [38] to 11,653 kmt at its end, when the Kwinana refinery — fully curtailed since June 2024 — was permanently closed, taking $856 million of restructuring charges and leaving roughly $525 million of cash outlays to run through 2031 [39].

Nine years of revenue growth on a shrinking asset base means the growth came from price. Alcoa's realized third-party aluminum price was $3,376 per tonne in FY2025 against $2,841 in FY2024 [40]; the same measure on the pre-separation carve-out was $2,669 in 2011 [41]. Fourteen years for a 26 percent nominal price gain, before any allowance for input-cost inflation over the same span.

Structural Threats

The alumina cost curve is inverting under Alcoa. This is the named, quantified threat, and it is live rather than prospective. Alcoa's Alumina segment realized $472 per tonne in FY2024 and $415 in FY2025 against adjusted operating costs of $309 and $317 [42]. In the six months to June 30, 2026 it realized $329 against a cost of $352 — production sold below cash cost — and the segment posted adjusted EBITDA of negative $136 million against positive $803 million a year earlier, with the Alumina Price Index down 38 percent year on year [43].

Sources: FY2025 Form 10-K segment table [44]; Q2 FY2026 Form 10-Q segment table [45].

The cause is supply, not demand. Alcoa attributes the decline to "refinery expansions in China and Indonesia" and recorded a $144 million goodwill impairment in Q4 2025 that took the Alumina reporting unit's goodwill to zero, citing declining alumina prices, higher capital expenditure on Australian mine moves and reclamation, and a higher discount rate [46]. Writing a reporting unit's goodwill to zero is management's own statement that the long-run value of that unit fell, and the Alumina segment is 11,653 kmt of the group's capacity against 2,645 kmt of smelting [47].

Sizing it for year 10: at FY2025 volumes of roughly 9,650 kmt of produced alumina shipments, each $50 per tonne of sustained margin compression is about $480 million of annual EBITDA. The move from FY2024's $163 per tonne spread to 1H FY2026's negative $23 is a $186 per tonne swing — on the order of $1.8 billion annualised — against FY2025 group adjusted EBITDA contributions of $882 million from Alumina and $1,058 million from Aluminum [48] [49]. A commodity whose price can move a segment's entire earnings contribution inside eighteen months is not a base from which a year-10 floor can be underwritten.

Bauxite quality and the Australian approvals. Alcoa's Australian refineries run on lower-grade bauxite than they were designed for, which "has caused increased production costs," and the company expects grades to stay at recent levels because mining in the next major mine regions — Myara North and Holyoake — will begin "no earlier than 2029" [50]. The 10-K states plainly that the grade issue could move the Alumina segment from the first to the second cost quartile until new mine regions are accessed [51]. Alcoa operates under a federal national-interest exemption granted for 18 months while a strategic assessment covering its mine lease to 2045 is completed by August 2027, and under a state exemption pending the WA EPA assessment [52]. Management says delays through the first half of 2027 are not expected to materially affect bauxite supply, quality, or operating costs, and that longer delays could be mitigated by modifying mining activities and refinery operating rates [53]. The exposure is that the cost position of the group's largest segment depends on a permitting outcome that is not in the company's control and that, at best, resolves the grade problem three years from now.

Section 45X expiry. The advanced-manufacturing credit contributed $63 million to FY2025 cost of goods sold at the Massena West and Warrick smelters. The One Big Beautiful Bill Act, enacted July 4, 2025, phases the credit out from 2031 and eliminates it entirely from 2034 [54]. That is a legislated, dated withdrawal of roughly 6 percent of FY2025 Aluminum segment adjusted EBITDA, landing inside the year-10 window.

Substitution and technology. Alcoa names steel, titanium, plastics, composites, ceramics, and glass as competing materials whose use "could reduce the demand for aluminum products" [55]. On the evidence in this corpus, material substitution runs toward aluminum rather than away from it over the next decade, so this is not where the year-10 risk sits. The technology question is inert-anode smelting: Alcoa's ELYSIS joint venture with Rio Tinto holds the patents [56], and management says the earliest Alcoa implementation is after 2030 [57]. Alcoa co-owns that development rather than facing it. On the framework's "your margin is my opportunity" test, no software or platform business can take share in bauxite mining or smelting; what defends the margin is the cost of building the physical plant, and nothing else.

Customer concentration — checked, and absent. Alcoa's 10-K discloses no single-customer dependence; its largest alumina customer is its own smelter system, at roughly 34 percent of total alumina shipments — 4,471 kmt of intersegment shipments against 13,300 kmt shipped in all [58]. The contrast in the peer set is instructive: Century Aluminum derived 54.0 percent of consolidated net sales from Glencore in 2025 [59]. Concentration is a genuine threat for parts of this industry; it is not one for Alcoa.

China dependence — checked. Alcoa is US-listed and US-incorporated with no Chinese revenue dependence disclosed. The China exposure here is structural and runs through price, not through customers: Chinese policy on capacity limits and environmental enforcement sets 59 to 62 percent of world supply in both products [60] [61].

The Disqualifier Check

The framework's numeric disqualifier is revenue declining high-single-digit for three consecutive fiscal years. It does not fire. fit_features.revenue_trajectory records consecutive_decline_years: 0 and three_year_hsd_decline: false.

Source: fit_features.revenue_trajectory, tied to the FY2025 Form 10-K Statement of Consolidated Operations [62].

The declines are severe but never consecutive past two years: −22.2 percent in FY2019, −11.0 percent in FY2020, then +30.9 percent in FY2021; −15.3 percent in FY2023, then +12.7 percent and +7.9 percent. Year-on-year change ranges from −22.2 to +30.9 percent across nine transitions, a 53-point spread around a mean of 5.1 percent.

The longer record, on the S-1's pre-separation carve-out basis, reads: $14,709 million in 2011, $13,060 million in 2012, $12,573 million in 2013, $13,147 million in 2014, $11,199 million in 2015 [63]. FY2025 revenue of $12,831 million is 12.8 percent below the 2011 figure. Two things are true at once and both belong on the page: the framework's specific structural-decline test is not met, and fifteen years of record show a sawtooth around a flat-to-falling line rather than a growing business.

Cash Flow Consistency

The framework's consistency test runs on adjusted FCF — reported FCF less stock-based compensation less the five-year average of acquisition spend. That series is not computable here. fit_features.not_computable records adjusted_fcf: missing SBC for FY 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024, 2025; no complete consecutive five-year acquisition window with SBC, and fcf_stability: fewer than five consecutive adjusted-FCF years. The rolling five-year average of adjusted FCF, the coefficient of variation, and the negative-year list are all absent from the feature file, and this tab does not improvise them.

What can be shown is the reported FCF series the feature file does carry, with its rolling five-year average computed here and labelled as such.

Source: reported FCF per fit_features.adjusted_fcf.series[].fcf, tied to the FY2025 Form 10-K Statement of Consolidated Cash Flow [64]; the rolling five-year average is derived here and is not the framework's adjusted-FCF series, which is not computable.

The rolling five-year average of reported FCF runs $100.2 million (FY2016–20), $349.2 million, $253.8 million, $156.0 million, $103.0 million, and $208.2 million (FY2021–25) — a mean of $195.1 million with a standard deviation of $87.9 million, a coefficient of variation of 0.45. The framework tolerates volatile annual FCF and rejects unpredictable five-year averages; a 45 percent coefficient of variation on the smoothed series is the second thing, not the first.

The negative years are FY2016 (−$715 million) and FY2023 (−$440 million). Neither is business-model-inherent in the sense the framework accepts. An insurer's underwriting cycle produces a loss year because premiums are collected before claims are known and the pricing corrects; the mechanism is identifiable and the cadence is 5 to 8 years. Alcoa's negative years are price years — FY2023 combined a 15.3 percent revenue decline with $531 million of capex [65], and the segment record shows Alumina adjusted EBITDA falling to $273 million from $788 million [66]. There is no repricing mechanism that restores the margin on a schedule, because Alcoa does not set the price.

Two structural drags sit between EBITDA and free cash flow for the whole of the year-10 window. Asset retirement obligations stood at $1,405 million at December 31, 2025, with $285 million of cash outflows scheduled for 2026 and $914 million for 2027–2030 [67]. Environmental remediation reserves stood at $282 million [68]. That is roughly $1.2 billion of legacy closure cash committed over the next five years, against ten-year cumulative reported FCF of $1,542 million [69].

Stock-based compensation, the first adjustment the framework applies, is small and disclosed for the recent years: $35 million in 2023, $36 million in 2024, and $41 million in 2025 [70]. Cash acquisition spend has been immaterial through FY2025 — the August 2024 Alumina Limited transaction was settled in stock. On those two adjustments alone, FY2025 adjusted FCF would compute to roughly $520 million against reported $567 million. That is an illustration, not the feature file's number, and the earlier years cannot be built the same way.

The Year-10 Case

The case that year-10 revenue and adjusted FCF are higher. Aluminum's supply side has genuinely tightened. China's operating electrolytic aluminum capacity reached 44.83 million tonnes per year at the end of 2025, "basically hitting the ceiling," with an increase of only 270,000 tonnes over the year [71]. Ex-China demand is expected to grow by roughly 7 million tonnes of aluminum and 18 million tonnes of alumina over the next decade while new supply is expensive to build [72]. Alcoa's Aluminum segment delivered record adjusted EBITDA of $1,073 million in Q2 2026 at a 32 percent margin, on an LME averaging $3,585 per tonne and rising regional premiums [73]. And Alcoa is adding scale rather than losing it: the June 30, 2026 agreement to acquire South32's upstream assets adds roughly 5.2 million tonnes of alumina capacity, a 53 percent increase, and roughly 900,000 tonnes of aluminum capacity, a 37 percent increase, at a valuation management describes as well below replacement cost [74] [75]. On volume arithmetic alone, a 2036 revenue line above 2025's $12,831 million is more likely than not.

The case against. Every element of that argument is a price forecast or an acquisition that has not closed. The consideration is $3,100 million of cash plus approximately 17 million Alcoa shares valued at about $1,000 million, plus a ticking fee at 5 percent per annum on the cash and up to $750 million of contingent payments over four years, funded initially by a $3,100 million bridge, with closing expected in the first half of 2027 [76]. Under the framework's own definition of adjusted FCF, $3,100 million of cash acquisition spend enters the five-year average at $620 million a year for 2027 through 2031. Alcoa's average reported FCF over FY2021–FY2025 was $208 million a year; it has cleared $620 million once in ten years, in FY2017. Adding roughly $40 million of SBC, reported FCF would need to run above roughly $660 million a year through 2031 simply for the framework's adjusted FCF to stay positive — before the acquired assets' own contribution, which the corpus does not disclose on a stand-alone basis.

On the revenue leg, the doubt is not about aluminum; it is about who captures the value. Alcoa is a 3.1 percent producer of a commodity priced daily on the LME and a 6.4 percent producer of one priced off an index it publishes but does not influence [77]. The larger segment by capacity is currently selling below cash cost, its goodwill has been written to zero, and the fix for its cost position is a mining approval that produces no ore before 2029 [78] [79] [80]. Fifteen years of history show revenue 12.8 percent lower than 2011 and cumulative free cash flow of $1,542 million against a market capitalization near $11.8 billion.

The read. The gate is not met. The framework asks for very high conviction that both year-10 revenue and year-10 adjusted FCF exceed today's, and treats any proper doubt as failure; here the doubt is specific and quantified rather than atmospheric — the larger segment is selling below cash cost today, its cost fix is a permitting outcome no earlier than 2029, and the framework's own adjusted-FCF definition would carry a $620 million annual acquisition charge from 2027 against a five-year average reported FCF of $208 million. The strongest fact against this read sits directly alongside it: China's production ceiling is binding, ex-China aluminum demand is expected to grow 7 million tonnes over the decade, greenfield capacity in the West costs seven times what it costs in China, and Alcoa's Aluminum segment has just printed its best quarter — so a higher 2036 revenue line is entirely plausible. The framework does not accept plausible in place of very high conviction, which is what separates this from a judgement about whether the shares are cheap.

What would change this read: a sustained recovery in the Alumina Price Index that restores the segment's spread above $100 per tonne for four consecutive quarters, ministerial approval of the Myara North and Holyoake mine regions on the 2026 timetable, and post-close disclosure showing the acquired assets generating free cash flow above the $620 million annual acquisition charge. The framework's own falsifier applies in reverse: three consecutive years of revenue decline, or FCF and EBITDA sliding from here, would remove the remaining ambiguity. On the balance sheet and repurchase mechanics that determine whether any of this reaches per-share value, see Self-Help; on the yield the current price implies, see Yield.