Chapter 3

The Operator's Word

The buy-not-build strategy left a visible footprint: Alcoa's diluted share count rose roughly 46% in two years, from 178 million at the end of 2023 to 261 million at the end of 2025. That is the largest single fact in the company's recent capital story, and it does not appear on the line most investors check first — the cash-flow statement reports zero cash spent on acquisitions in every year of the period. To understand how Alcoa got here, and whether management's word has held, the deployment has to be read as one ledger rather than as separate line items. Read that way, it tells two consistent stories at once: a team that reliably delivers what it promises operationally, and a team that has quietly deferred the shareholder returns it announced at the last cycle peak.

The Promises the Team Kept

The record being judged is mostly this management's own. Molly Beerman became chief financial officer in February 2023 and William Oplinger — a two-decade Alcoa insider, previously chief financial and then chief operating officer — became chief executive in September 2023, succeeding Roy Harvey. Almost every commitment that has since come due was made and settled by that pair, which makes the said-versus-did test a clean one.

On the operational side, the team has delivered. Into the 2023 trough — a net loss, free cash flow of -$440 million, the Kwinana refinery curtailment — management set a roughly $645 million profitability-improvement program. A year later it reported it had "delivered and exceeded" the target "ahead of schedule," reaching $675 million [1]. The balance-sheet promise was kept too: management defined an adjusted net-debt target of $1.0 to $1.5 billion — a measure that layers pension and other retirement liabilities on top of reported borrowings — and reached that range by the end of 2025, having stood at $2.1 billion as recently as the first quarter [2]. On the simpler reported basis, net debt fell from $1.41 billion in FY2024 to $842 million in FY2025.

Source: net debt derived from reported financials (FY2021–FY2025 Forms 10-K); the $1.0–1.5B adjusted net-debt target (which adds pension and retirement liabilities) is management's own, stated on the Q1 FY2025 earnings call [3].

The portfolio commitments held on the same schedule. Management restarted the San Ciprián smelter in Spain and, in July 2025, monetized the non-core 25.1% stake in its Saudi Arabian joint venture, exchanging it with Ma'aden for consideration of roughly $1,350 million — cash plus marketable shares — which fed the deleveraging above [4]. Credibility runs both ways, and on operations and the balance sheet the arrow points up.

Sources: Q4 FY2024 [5] and Q1 FY2025 [6] earnings calls; FY2025 10-K Note C [7].

The Return That Stopped

The commitment that has not held is the one made to shareholders directly. At the 2022 cycle peak, then-CEO Roy Harvey returned $387 million year-to-date — $275 million of buybacks plus dividends — and announced a fresh $500 million repurchase authorization on top of $150 million still open [8]. Returning cash to shareholders was named, explicitly, as one of three priorities ranked after protecting the balance sheet [9].

What followed was silence on that channel. Buybacks went to zero in 2023, 2024 and 2025, and the July 2022 authorization sat entirely intact: as of December 31, 2025 the full $500 million remained available, unused for roughly three and a half years [10]. The only return that continued was the quarterly dividend, held flat at $0.10 per share every quarter since the program began in late 2021 [11].

Source: dividends paid per FY2025 10-K Statement of Consolidated Cash Flows [12]; repurchases per the FY2022 buyback disclosure [13] and company cash-flow data.

The growing dividend line is largely optical. Total dividend cash rose from $72 million in 2023 to $104 million in 2025, but the per-share rate never moved; the outlay grew only because there were more shares to pay [14]. On a per-share basis, the shareholder return has been flat for four years while analysts pressed repeatedly, across at least six recent calls, on when buybacks would resume — and were deferred each time.

One number captures how that period can be misread. Buybacks plus dividends as a share of free cash flow swung from -16.4% in FY2023 to 211.9% in FY2024, a move that looks like a policy reversal toward aggressive payout. It is almost entirely a denominator effect. The dividend numerator barely moved, from $72 million to $89 million; what changed was free cash flow, which climbed off a -$440 million trough to a razor-thin +$42 million, so the ratio against it careened. The swing measures the depth of the 2023 cash trough, not any decision to return more.

Source: free cash flow and shareholder-return figures derived from reported financials (FY2023–FY2024 Forms 10-K); dividends paid per FY2025 10-K Statement of Consolidated Cash Flows [15].

The Ledger the Cash Flow Hides

If the returns channel went quiet, the deployment channel did not — it simply ran through a line the cash-flow statement does not show. The single largest capital move of the period was the August 1, 2024 acquisition of Alumina Limited, which took Alcoa's stake in the AWAC bauxite-and-alumina joint venture from 60% to 100%. It was paid for entirely in stock: 78,772,422 common and 4,041,989 preferred shares, for aggregate consideration of approximately $2,700 million [16].

Because Alcoa was buying out a 40% noncontrolling interest ($1,472 million) rather than acquiring a new business, the deal was booked as an equity transaction — net assets and costs added to additional capital, with no goodwill recognized. The mechanical result is that the cash-acquisitions line reads $0, technically correct and quietly misleading: the biggest deal in the period is invisible where a reader would look for it [17].

The true cost shows up in the share count. Diluted shares went from 178 million (FY2023) to 214 million (FY2024) to 261 million (FY2025). Employee stock compensation cannot explain it — that expense ran just $41 million, $36 million and $35 million across 2025, 2024 and 2023, or roughly one to one-and-a-half million shares a year [18]. The dilution is acquisition currency, not pay. For an investor, the ledger of how growth was financed is the share count, not the cash-flow acquisitions line — and on that ledger holders own about 46% less of the company per share than they did two years earlier.

Source: diluted share counts per Forms 10-K (FY2021–FY2025); the FY2024–FY2025 step reflects the all-stock Alumina Limited issuance per FY2024 10-K Note C [19].

From Net-Debt Target to Mega-Deal

The sequence is what makes the capital story cohere. Management deferred buybacks to reach a net-debt target, reached it — and then, rather than turning the returns channel back on, committed the restored balance sheet to the largest acquisition in Alcoa's history.

The pivot was fast. On the Q4 2025 call, in January 2026, management said plainly it had "no greenfield expansion plans for aluminum" and none for refining or bauxite, because energy prices do not support the returns; growth would be limited to low-capital brownfield work [20]. Five months later, on June 30, 2026, it agreed to buy South32's bauxite, alumina and aluminum assets — the "AliGroup" package — for $3,100 million of cash plus roughly 17 million Alcoa shares valued at about $1,000 million, a 5% annual ticking fee on the cash portion until closing, and up to $750 million of contingent payments tied to future alumina and aluminum prices [21]. The $3.1 billion cash leg is backed by a committed 364-day bridge loan of the same size — the re-leveraging, in one line [22].

Management framed it as the largest transaction in company history, carrying roughly $900 million of net-present-value synergies and lifting capacity by about 53% in alumina and 37% in aluminum, with cash consideration sized to keep pro forma leverage within 2.0x [23]. Whatever its strategic merits, its effect on the capital ledger is unambiguous: the deleveraging that shareholder returns were postponed to achieve has become the funding capacity for a deal, again paid partly in stock, rather than for the buybacks the 2022 framework promised.

Sources: FY2022 buyback authorization [24]; Alumina Limited terms, FY2024 10-K Note C [25]; dividend, FY2025 10-K Note N [26]; AliGroup terms, Q2 FY2026 10-Q Note C [27].

Governance

The setting around these decisions is a conventional US large-cap. Ten of eleven directors are independent, with a separate independent chairman distinct from the chief executive; only Oplinger sits on the board as a non-independent member [28]. Pay is heavily equity-weighted and largely at risk: FY2024 chief-executive compensation was $13.5 million with no cash bonus and roughly two-thirds in stock, a 129-to-1 pay ratio [29]. There is no control block and no dual-class structure. The one feature worth carrying forward is that the annual incentive keys off adjusted-EBITDA and free-cash-flow measures whose adjustments management itself sets — an alignment question the record above should be read against, not a governance defect on its face.

Where the Record Stands

The honest state of the ledger is two-sided and specific. This management does what it says operationally: the profitability program was beaten early, the net-debt target was reached, San Ciprián restarted, the Saudi stake sold. It has not done what it said on returns: the 2022 buyback authorization has sat unused for three and a half years, the per-share dividend has not risen since 2021, and growth was financed by issuing shares — about 46% more of them in two years — with the restored balance sheet now committed to the largest deal in the company's history rather than to the returns that were deferred to build it.

That leaves an open commitment a reader can hold management to on dated, checkable terms: the AliGroup acquisition, expected to close in the first half of 2027, and the synergy and accretion claims attached to it, are the next entry the said-versus-did ledger will settle. What the market now pays for a company whose recent record profit came largely from one-time gains — and what its forward multiple quietly assumes — is the subject of What the Price Assumes.