The AES Corporation (NYSE: AES) — A Value-Destroying Global Power Sprawl the Market Discounted, Now a $15 Merger-Arb Bond With a CFIUS-Shaped Tail

Independent equity research. Report date: 2026-07-11. The analysis carries no recommendation and no price target — that discipline is deliberate. The single exception is the labeled “Author’s Take” block immediately below.

⚡ Author’s Take

This block is the author’s own subjective opinion and general information, not investment advice. The analysis in the numbered sections below takes no position and sets no price target.

Verdict: HOLD / merger-arb hold at $14.78 — a low-single-digit spread that is fair, not fat, for the regulatory tail it carries. Not a fundamental buy; not a short. Hard ceiling $15.00; standalone-if-broken floor ~$10–13.

AES has been rescued from itself. Strip out the deal and you are left with a four-decade-old global power conglomerate that earned a ~5% return on invested capital against a 7–8%+ cost of capital, took an impairment in most years of the last decade, funded a chronically FCF-negative renewables build with project debt, hybrids, tax credits and asset sales, and rewarded a decade of shareholders with a −7.4% five-year annualized return, a negative Sharpe, and a −79% lifetime drawdown. The public market correctly refused to pay up for that; it took a private-capital consortium — GIP (BlackRock), EQT, CalPERS and Qatar’s QIA — to put a floor under the equity at $15.00 in cash (signed 1-Mar-2026; shareholders approved 26-Jun-2026, ~97.8% for). Note the tell that the definitive $15.00 came in below the rumor-inflated ~$16.7 the stock had run to, and shares fell ~16% on the signing — $15.00 is a firm ceiling, not a floor, and a topping bid looks unlikely.

At $14.78 you are buying a ~1.5% gross spread plus a permitted ~$0.176/quarter dividend to close (guided late-2026/early-2027, outside date extendable to ~Dec-2027) — call it a ~5–7% annualized arb yield if it closes on schedule. The framing is special-situation / merger-arbitrage bond-proxy, and the factor tape confirms it: beta collapsing toward the deal, idiosyncratic vol suppressed, the +16.8% one-year return entirely the takeover pop on top of a genuinely ugly long record. The catch is a negatively-skewed payoff: ~1.5% up to $15 against ~15–30% down to a ~$10–13 standalone value if a regulator balks. The real tail is CFIUS — a large US critical-infrastructure operator being bought by a consortium that includes a Qatari sovereign fund and a European sponsor — plus a genuinely long list of foreign competition/FDI clearances. The buyer’s willingness to post a $588M reverse termination fee (~5.5%) tells you they, too, see real regulatory risk. Conviction: medium. Flip me bullish: clean CFIUS/FERC/EU/PUC sign-offs that walk the spread to zero into the close. Flip me bearish: a CFIUS second-request, a foreign-regulator objection, or a financing wobble — any of which blows the spread out and re-exposes the ugly standalone business underneath. Tag: “Dead money with a $15 seatbelt — and a CFIUS airbag that may or may not deploy.”

📈 Stock Price Action — Five-Year Event Map

Factual price history — no recommendation, no price target. Prices below are unadjusted (what the stock actually traded at). Source: AZI 5-year price CSV; events cross-referenced to earnings 8-Ks, the merger filings, and the news feed.

AES round-tripped from a clean-energy darling to a value-trap and out the other side into a buyout. The stock ran to a ~$29.9 high in December 2022, collapsed to a ~$9.46 five-year low in May 2025 (−68%) as rates, leverage and IRA-rollback fear repriced it, then was rescued by takeover speculation (July 2025) and a definitive $15.00 cash deal (March 2026). It trades at $14.78 (10-Jul-2026), a ~1.5% discount to the deal price, roughly −51% below its 2022 high and near the top of a 52-week range of roughly $11.8–$17.1.

Cycle narrative. (1–2) AES rode the 2021–22 clean-energy premium to ~$30, then gave it all back through 2023 as rising rates repriced a highly-levered, long-duration renewables developer and recurring impairments confirmed weak capital deployment. (3) A 2024 relief rally on rate-cut hopes and the emerging AI-power-demand story faded. (4) Through H2-2024 into May-2025 the stock cratered to a ~$9.46 five-year low — the market pricing IRA policy risk, ~$29B of consolidated debt, EM/FX exposure and doubts about dividend coverage into a sub-cost-of-capital utility. (5) On 9-Jul-2025 an FT report of private-capital interest lifted it ~20% off a ~$11.07 unaffected price. (6–7) Speculation carried it to ~$16.7 before the definitive $15.00 cash agreement (1-Mar-2026) — struck below the rumor price — knocked it back ~16%. (8) Since signing it has traded as a merger-arb instrument, pinned just under $15 through the 26-Jun-2026 shareholder approval, now awaiting a long slate of regulatory clearances.

1. Executive Summary

The AES Corporation is a diversified global power company — two US regulated utilities (AES Indiana, AES Ohio), a large US-and-emerging-market renewables developer, a shrinking-but-still-cash-generative fossil fleet, and a lossmaking new-technology venture arm (28% of Fluence, plus AI/robotics bets). It operates a 34,740 MW fleet across the US and roughly a dozen emerging markets, serving 2.7 million utility customers, on ~$12.2B of revenue and ~$3.4B of EBITDA.

The investment situation has changed character entirely. On 1-March-2026 AES agreed to be acquired for $15.00 per share in cash — total equity value ~$10.7B — by a consortium of Global Infrastructure Partners (part of BlackRock), EQT Infrastructure VI, CalPERS (via a GIP-managed vehicle), and the Qatar Investment Authority. Shareholders approved the merger on 26-June-2026 with ~97.8% of votes cast in favor. The stock, at $14.78, now trades as a near-closed merger-arbitrage position: a ~1.5% gross spread to the $15.00 consideration, plus a contractually-permitted regular quarterly dividend (~$0.176/share) collectible until closing. The deal is expected to close in late-2026/early-2027, with an outside date of 1-June-2027 that auto-extends to as late as ~December-2027 for regulatory reasons.

The underlying business, which defines the downside if the deal breaks, is unattractive. AES has earned a return on invested capital of roughly 5% against a 7–8%+ cost of capital — it destroys economic value on incremental investment — and has confirmed as much by taking impairments in most years of the last decade ($0.3–1.7B annually). Its “growth” is a rotation of the revenue base from fossil to renewables/utilities rather than genuine top-line expansion (revenue has been flat-to-down for three years), funded by a chronically FCF-negative capital program that leans on project debt, tax-equity, minority sell-downs (including ~30% of both US utilities sold to CDPQ) and federal tax credits ($1.5B of IRA tax-attribute monetization in 2025). GAAP earnings are near-meaningless noise; the relevant owner metric is Parent Free Cash Flow (~$1.15–1.25B) against ~$6.0B of recourse parent debt (the other ~$23.2B of debt is non-recourse project debt).

The competitive verdict is blunt: no durable company-wide moat. The regulated utilities have a narrow regulatory-franchise moat on ~30% of EBITDA (itself ~30%-owned by CDPQ); the far larger global generation/renewables business is a capital-intensive price-taker with a genuine-but-replicable development capability, not structural pricing power, and its emerging-market footprint is a liability as often as an asset.

The variant question is purely probabilistic. Consensus (correctly) treats AES as an arb, not an investment. The dominant risk is a binary deal-break — most plausibly at CFIUS, given foreign (Qatari, European) sponsors buying a large US critical-infrastructure operator, plus a long list of foreign competition/FDI approvals. A break re-rates the stock toward its pre-deal ~$10–13 standalone value (~15–30% downside) against ~1.5% of upside — a negatively-skewed payoff. The buyer’s $588M reverse termination fee (~5.5%) signals that real regulatory risk was priced in. No BUY/SELL and no price target appear below; the discussion is framed as embedded expectations and scenarios.

2. Business Overview

What AES is (FACT). The AES Corporation, incorporated in 1981 and headquartered in Arlington, Virginia, is a globally diversified electric power company that owns and operates a 34,740 MW generation fleet and six regulated utilities serving 2.7 million customers, with 8,336 employees as of 31-December-2025 (FY2025 10-K, Item 1). It is a hybrid business: part US-regulated utility, part global independent power producer (IPP), part renewables developer. It earns money three ways — (i) a regulated rate of return on its US and El Salvador utility rate base; (ii) long-term contracted (PPA) generation cash flows, largely with capacity payments and fuel pass-throughs designed to hedge commodity and FX exposure; and (iii) merchant/short-term spot and ancillary-service sales exposed to marginal power prices. Management frames the strategy as “partnering with large corporations to deliver the electricity they need” — renewables to US hyperscale data-center customers and to large non-US mining companies.

Segment structure (FACT). AES reports four technology-organized Strategic Business Units (SBUs). FY2025 revenue and Adjusted EBITDA (a non-GAAP segment measure, before Corporate drag):

Consolidated revenue was $12,233M in FY2025 (vs. $12,278M in 2024 and $12,668M in 2023), split ~$8,195M non-regulated / ~$4,038M regulated (10-K, Consolidated Statements of Operations; Note 19).

The capital-intensity tension is visible in one comparison (INTERPRETATION). The Energy Infrastructure (fossil) SBU still generated the most Adjusted EBITDA — $1,130M off only $4,726M of long-lived assets — while Renewables generated $932M off $23,945M of long-lived assets (Note 19). AES has poured capital into renewables (which carry roughly five times the asset base) to produce less EBITDA than the shrinking fossil fleet it is replacing. Renewables assets grew from $17.3B (2023) → $19.2B (2024) → $23.9B (2025); fossil assets shrank from $5.8B → $4.7B. This is the first, blunt read on the return problem developed in Section 4 and Section 6.

Regulated utilities (FACT). The two US utilities are AES Indiana (Indianapolis Power & Light — a fully integrated generation-and-T&D utility) and AES Ohio (Dayton Power & Light — wires-only distribution). Both are sole distributors in their territories. Recent constructive rate outcomes include AES Indiana’s April-2024 base-rate order (+$71M/yr) and an AES Ohio distribution settlement (+$168M/yr). Management guides to double-digit rate-base growth through 2027. Critically, CDPQ (Caisse de dépôt) owns ~30% of both IPALCO (AES Indiana’s parent) and, since April-2025, ~30% of DPL/AES Ohio — the crown-jewel regulated assets are already partly sold down, so AES common holders own only ~70% of even this quality core. El Salvador adds four small distribution utilities.

Geography (FACT). AES is genuinely global and emerging-market-heavy: ~59% of 2025 revenue was earned outside the US. By country: United States ~$5,056M (incl. Puerto Rico ~$404M), Chile $1,516M, Dominican Republic $1,363M, El Salvador $1,086M, Mexico $760M, Bulgaria $687M, Panama $649M, Colombia $422M, Argentina $366M, Vietnam $321M (Note 19). The fleet is ~54% renewables + hydro by capacity, ~29% gas, ~15% coal, ~2% oil/pet-coke — AES is mid-transition away from coal.

New Energy Technologies (FACT). Holds a 28.19% economic interest in Fluence Energy (Nasdaq: FLNC), the AES/Siemens battery-storage JV; plus Maximo (an AI solar-install robot), the AES AI Fund, Uplight and 5B. This SBU loses money (–$35M Adjusted EBITDA in 2025) and is a venture/equity-method bucket, not an earnings engine.

Recurring vs. non-recurring (INTERPRETATION). Utility revenue (~33% of the total) and long-dated PPA cash flows are genuinely annuity-like. But GAAP earnings are heavily distorted: FY2025 consolidated net income was only $162M, yet net income attributable to AES was $910M — the ~$748M gap is losses allocated to noncontrolling/tax-equity investors under hypothetical-liquidation (HLBV) accounting on US renewables. Reported results are further swung by impairments, asset-sale gains/losses and FX. Headline GAAP diluted EPS of $1.26–1.33 in 2025 is a poor guide to run-rate economics; management steers on Adjusted EPS ($2.34 in 2025) and Adjusted EBITDA.

Verdict (Business Overview): A sprawling, four-decade-old global power conglomerate — two solid US regulated utilities (~30% already sold to CDPQ), a large US-plus-emerging-market renewables developer, a still-cash-generative but shrinking fossil fleet, and a lossmaking venture arm. It is complex, capital-hungry, FX- and sovereign-exposed, and accounting-opaque. The regulated utilities are the quality core; everything else is a contracted-but-commoditized generation business wrapped in emerging-market risk. This discounted-conglomerate character is precisely why a private-capital consortium is taking it out at $15.00 — the pieces are worth more owned separately by long-duration infrastructure capital than as a public holdco.

3. Industry Dynamics

Two industries under one ticker (INTERPRETATION). AES straddles (a) rate-regulated utility distribution/generation — a structurally attractive, low-but-stable-return, natural-monopoly business — and (b) merchant/contracted power generation and renewables development — a capital-intensive, competitive, commoditized business. The blended structural quality is a weighted average dragged down by the larger, riskier generation leg.

The demand tailwind is real (FACT). US power demand is inflecting upward for the first time in roughly two decades, driven by data-center/AI load, electrification and reshoring. AES is a direct beneficiary — it is consistently ranked by BloombergNEF among the top two largest sellers of renewable power to corporate customers globally. Its backlog of signed-but-not-operating projects reached 12.0 GW (5.7 GW under construction); it signed/was awarded 4.0 GW of new long-term PPAs in 2025 and completed 3.2 GW of construction. Its US arm, AES Clean Energy, carries a 7.6 GW signed backlog against a 46 GW development pipeline, with >$12 billion of construction budget on contracted projects. Named offtakers across the fleet read like a hyperscaler roster — Amazon, Microsoft, Google, Apple — plus utilities (Constellation, Dominion, PG&E) and mining majors (Codelco, Collahuasi, Los Pelambres). Utility service territories are also attracting incremental data-center load above existing rate-base plans.

But this demand is a rising tide for everyone (INTERPRETATION — Marathon capital-cycle lens). Visible demand growth and (until recently) attractive returns in renewables have attracted a flood of capital: NextEra, Brookfield Renewable, Vistra, Constellation, oil-major renewables arms and infrastructure funds all chase the same hyperscaler PPAs and interconnection-queue positions. This is a classic capital-cycle setup in which abundant supply competes returns down — PPAs are auctioned/negotiated, panels/turbines/batteries are globally-sourced commodities, and the marginal developer’s cost of capital sets the clearing price. Published analysis of NextEra has framed renewables development as “a capital-intensive, commodity-exposed merchant-development machine dressed up in long-term contracts… a superbly run developer in a fragmented, commoditizing, subsidy-dependent, capital-intensive business” (NextEra, a comparable large-cap renewables developer). If the best-capitalized, lowest-cost-of-capital developer in the world earns thin, subsidy-dependent economics here, AES — with a higher cost of capital and a sub-scale US position — earns worse.

Regulation is a two-headed risk (FACT). Domestically AES answers to FERC, the Indiana URC and the Public Utilities Commission of Ohio (PUCO), plus the NY PSC — constructive but slow rate regulation that is both the utilities’ moat and their ceiling. Internationally, AES carries emerging-market regulatory, FX and sovereign risk as a defining feature, not a footnote: Argentine peso devaluation and CAMMESA receivables; Chilean spot-price and stabilization-fund dynamics; Colombian hydrology/regulation; the Bulgaria Maritza PPA expiration (which triggered a $264M impairment in 2025); Puerto Rico’s PREPA restructuring; and Vietnam BOT contracts. The 10-K’s own risk factors run to expropriation, currency-repatriation restrictions and governments “unwilling… to honor their commitments.” AES carries the lowest investment-grade rating (BBB-/BBB-/Baa3) with only ~$1.4B unrestricted cash against a heavy consolidated debt load and a continuous external-funding need.

IRA/tax-credit dependence is the policy fault line (FACT + INTERPRETATION). US renewables economics lean heavily on Investment/Production Tax Credits. In 2025 AES recognized $1.5 billion of tax-attribute monetization (tax-equity plus transferability sales) on US renewables — the primary driver of the wedge between $162M consolidated net income and $910M AES-attributable income, and a flatterer of the effective tax rate. Management asserts its contracted and advanced-stage backlog is “resilient to recent changes in the IRA.” That is management’s hypothesis; the reality is that the growth engine’s returns are underwritten by a federal subsidy regime under active rollback pressure. A materially faster IRA phase-out would compress new-build returns and the tax-equity funding the model depends on — the single biggest structural swing factor for the value of the renewables SBU.

Verdict (Industry Dynamics): Structurally mixed, tilting unattractive for the majority of AES’s capital. The regulated-utility slice sits in a good industry (monopoly franchise, allowed returns). The far larger generation/renewables slice sits in a capital-hungry, commoditizing, subsidy-dependent industry in the demand-up phase of a capital cycle that reliably competes returns down — layered with emerging-market FX/sovereign risk on ~59% of revenue. The AI-demand tailwind is real but shared by every well-funded competitor and confers no pricing power. Good demand growth is not the same as a good industry.

4. Competitive Position

Where the moat is (limited) — the utilities. AES Indiana and AES Ohio have a genuine but narrow regulatory-franchise moat: exclusive service territories, allowed returns on a growing rate base, high barriers to entry (Greenwald’s government-conferred barrier). This is real and shows up in stable regulated returns. But it is (i) small — Utilities is ~30% of Adjusted EBITDA; (ii) already ~30% sold to CDPQ, so common holders own ~70% of even this crown jewel; and (iii) capped by regulation — you cannot earn super-normal returns on a rate base.

Where the moat is largely absent — global generation & renewables (INTERPRETATION). The generation and renewables businesses are fundamentally price-takers with weak durable differentiation. In Greenwald’s taxonomy, none of the three genuine advantage types holds durably:

• Supply/cost advantage: No. AES buys the same commodity panels, turbines and batteries as every competitor; its cost of capital is higher than NextEra’s or Brookfield’s, which is the binding input in a capital-intensive business. It is a mid-pack, sub-scale US developer (Chile ~6% share; no dominant US position).
• Demand/captivity (switching costs): Weak. Once a PPA is signed the counterparty is locked for the term, but winning the next PPA is a competitive auction. There is no customer captivity at the point of sale — hyperscalers multi-source and negotiate hard.
• Economies of scale + captivity: No. AES is not the scale leader in any product; renewables development is a fragmented, contestable market.

What AES does have is a development-platform capability — a four-decade track record of building customized projects on time and on budget, deep hyperscaler relationships, an interconnection-queue and powered-land position (46 GW pipeline), and develop-transfer-agreement optionality (e.g., the July-2025 ~$481M Texas data-center DTA). This is a real skill and a real backlog, but it is an execution/relationship edge, not a structural moat — it must be re-earned on every project, it is replicable by better-capitalized rivals, and it does not confer pricing power. By the Section 9 test — a “moat” that would erode the moment execution slipped is not a moat.

The emerging-market footprint is as much liability as asset (INTERPRETATION). The bull frames Chile/Colombia/Argentina/Brazil scale as a differentiator. In practice it is a source of FX translation losses (~$102M of incremental FX translation loss in 2025), sovereign non-payment risk, hydrology risk and stranded-PPA impairments (Maritza $264M) — precisely the exposures a US-focused competitor avoids. It raises AES’s cost of capital and caps its multiple; it is a reason the stock traded at a discount, not a premium.

The financial verdict on the moat — returns below cost of capital (FACT/INTERPRETATION). A moat must show up in ROIC above WACC. AES’s returns do not clear its cost of capital: ROIC.ai reports return-on-invested-capital of ~5.4% (2024), –11.1% (2023, impairment-driven) and ~5.6% (2020) — low-single-digit-to-negative — against a cost of capital comfortably in the 7–8%+ range for a BBB- issuer with emerging-market exposure. (GOTCHA/caution: ROIC.ai also reports a “191% ROE” and a “366% sustainable growth rate” for 2025 — meaningless thin-common-equity artifacts, because AES’s GAAP common equity is small relative to the vast NCI/tax-equity and debt in the capital structure; do not use them. The invested-capital return is the relevant read, and it is sub-WACC.) This is the acid test: a business earning below its cost of capital destroys economic value on incremental investment, regardless of how much EBITDA it books. The regulated utilities earn their allowed return; the global generation/renewables business, taken as a whole and after impairments and FX, does not.

Versus peers (FACT/INTERPRETATION). NextEra (NEE) is superior — larger scale, lower cost of capital, a genuine crown-jewel regulated monopoly (FPL); AES is a smaller, higher-cost, EM-exposed version of NEE’s weaker leg without FPL. Vistra / NRG / Constellation run a different model — US merchant generators levered to tightening PJM/ERCOT power and capacity prices, owning scarce nuclear/thermal baseload; AES’s fleet is commodity solar/wind/gas plus EM assets. Brookfield Renewable is a comparable global-developer model but with permanent infrastructure capital and a lower cost of funding.

Verdict (Competitive Position): No durable, company-wide moat. A narrow regulatory-franchise moat on ~30% of EBITDA (itself ~30%-owned by CDPQ), and a genuine-but-replicable development capability on the rest — not structural pricing power. The global generation/renewables business is a capital-intensive price-taker whose returns on invested capital run below its cost of capital, and its emerging-market footprint is a liability as often as an asset. This is a good-execution business in a bad-economics structure. The buyout makes sense not because of a hidden moat but because patient infrastructure capital can finance the assets more cheaply and monetize the pieces at private-market values the public market never awarded.

5. Growth History and Forward Opportunities

Revenue growth has been essentially nil (FACT). Consolidated revenue ran roughly $9.7B (2020) → $12.7B (2023) → $12.3B (2024) → $12.2B (2025) — a modest early-decade lift then flat-to-slightly-down for three straight years. The 2024 sale of AES Brasil removed ~$616M of revenue and reshaped the base. Segment mix shifted meaningfully: Renewables revenue grew $2,416M → $2,913M (2023→2025, +21%) and Utilities grew $3,495M → $4,122M (+18%, on rate cases), while Energy Infrastructure shrank $6,805M → $5,402M (–21%) as coal wound down and Chile’s coal-indexed contracts expired. The “growth” is a rotation of the revenue base from fossil to renewables/utilities, not top-line expansion.

The growth story is a capacity/backlog story, not a revenue story (FACT). Renewables assets-in-operation reached ~17.8 GW in 2025 (+3.2 GW built), with a 12.0 GW contracted backlog (5.7 GW under construction) and a 46 GW US pipeline. Management’s historical algorithm has been ~7–9% annual Adjusted EPS growth plus rate-base-driven double-digit utility growth through 2027. The forward opportunity set is credible in volume terms: hyperscaler PPAs, US utility rate-base build-out, data-center DTAs and powered-land monetization.

But the quality of this growth is poor (INTERPRETATION — the core skeptical point). Three tells:

1. It is extraordinarily capital-hungry and self-funding-negative. Capital expenditures were $7,733M (2023), $7,519M (2024) and $5,982M (2025) — roughly 2–2.7x the ~$2.6–2.9B of annual Adjusted EBITDA, with Renewables alone consuming $4.6–6.0B/yr. AES cannot fund its growth from operating cash flow; it relies on project debt, tax-equity, minority sell-downs and asset sales to plug the gap.
2. Growth is funded by selling the best assets and by tax subsidies. AES hit its $400–500M 2025 asset-sale target (e.g., a $450M AGIC minority sale), sold ~30% of AES Ohio to CDPQ, sold AES Brasil (2024) and sold down Dominican Republic renewables — cannibalizing ownership of existing cash flows to fund new build. And the reported returns on US renewables are inseparable from the $1.5B of IRA tax-attribute monetization in 2025 — strip the subsidy and the growth’s economics deteriorate sharply.
3. The returns on the growth are sub-WACC (see Section 4). Deploying $6–8B/yr of capital at a ~5% ROIC into a business with a ~7–8%+ cost of capital is, by definition, value-destructive growth — the more it “grows,” the more economic value it consumes, even as Adjusted EBITDA and Adjusted EPS tick higher.

Organic vs. acquired. The renewables build is largely organic development (AES Clean Energy Development is the greenfield engine); the portfolio has simultaneously been shrunk through disposals. Net, AES has been a capital recycler — building renewables organically while selling mature/fossil/EM assets — rather than a net grower of the asset base’s earning power to common holders.

Verdict (Growth): Low-quality growth. The volume/backlog pipeline is real and the demand is real, but the growth is capital-intensive to the point of being self-funding-negative, financed by asset sales, minority sell-downs, project leverage and federal tax credits, and — most damningly — deployed at returns below the cost of capital. Revenue has been flat for three years; the “7–9% Adjusted EPS growth” leans on non-GAAP add-backs and tax-attribute monetization rather than rising cash returns on capital. This is growth that expands the balance sheet faster than it expands per-share economic value — exactly the profile that makes a discounted, patient-capital take-private the rational endgame.

6. Financial Quality

The headline problem: GAAP earnings are noise, and the balance sheet is the business. AES is a $12.2B-revenue global power company whose reported profit is dominated by non-cash items with almost nothing to do with the cash the parent actually collects. GAAP net income to common has whipsawed from –$546M (2022) to +$1,679M (2024) to +$910M (2025) on essentially flat revenue (FACT — ROIC.ai; 10-K). The 2025 reconciliation makes the point cleanly: GAAP diluted EPS from continuing operations was $1.33, but stripping out $0.52 of impairment losses, $0.34 of disposition losses, $0.17 of unrealized derivative/equity losses, $0.12 of restructuring and $0.04 of FX (net of a $0.22 tax benefit) lifts Adjusted EPS to $2.34 — up from $2.14 in 2024 (FACT — 10-K, “Reconciliation of Adjusted EPS”). Interpretation: the impairment and disposition add-backs (coal-plant retirements, emerging-market write-downs, asset sales funding the renewables build) are recurring features of this business model, not one-time noise — AES has taken impairments in most years of the last decade. An investor who accepts the $2.34 figure at face value is choosing to ignore that management has serially destroyed carrying value; the “adjusted” number flatters a company that keeps writing off yesterday’s capital.

Margins are utility-like but not improving with scale. EBITDA of ~$3.4B on $12.2B revenue is a ~28% margin, with EBIT ~$1.97B (~16%) (FACT — ROIC.ai). Revenue has gone sideways — $12.6B (2022) → $12.7B (2023) → $12.3B (2024) → $12.2B (2025) — despite years of heavy growth capex, so the Greenwald test (do unit economics improve as the asset base grows?) fails: invested capital has grown far faster than operating profit. AES is not a scale-economics compounder; it is a capital-intensive asset assembler whose incremental returns hover around its cost of capital.

Cash flow: strong at the consolidated line, thin at the level that matters. Consolidated cash from operations jumped to $4,306M in 2025 from $2,752M in 2024 (FACT — 10-K), a genuine improvement on working-capital normalization and higher subsidiary output. But this number is misleading for an equity holder because it is generated across dozens of partly-owned subsidiaries and is largely spoken for: investing outflows ran roughly –$6.2B as AES funds one of the largest US renewables construction backlogs. The metric AES itself guides to — Parent Free Cash Flow (subsidiary distributions less parent interest, opex and dividends), guided to roughly $1.15–1.25B — is the true owner-economics figure (management non-GAAP measure; reconcile to the earnings release). The gap between $4.3B consolidated OCF and ~$1.2B parent FCF is the story of this company: most of the cash belongs to non-controlling interests and project lenders, not AES common.

The defining balance-sheet feature: recourse vs. non-recourse debt. Total consolidated debt of ~$29.2B looks alarming against ~$4.06B of common equity, but the 10-K splits it explicitly: ~$6.0B is recourse debt of the Parent Company and ~$23.2B is non-recourse project/subsidiary debt repayable “solely from the project’s revenues” (FACT — 10-K). The non-recourse tranche materially overstates the risk to the parent — if a given project defaults, AES’s downside is capped at its equity in that project, not the whole loan. But the structure is not a free pass: (i) parent leverage of ~$6.0B against ~$1.2B of parent FCF is still ~5x, high for an equity that sits behind project lenders and minorities in every waterfall; (ii) consolidated interest expense of $1,407M in 2025 consumes a large share of EBIT and rises with rates and incremental project debt; and (iii) at the consolidated level, current ratio 0.77, debt/cap ~84%, cash $1.38B and ~$20M of subsidiary debt already in technical default describe a perennially tight, highly-levered structure with thin tangible common equity (~$6.18 book value/share, much of it goodwill and regulated rate base). (ROIC.ai reports total debt ~$29.9B vs. the 10-K’s $6.0B + $23.2B ≈ $29.2B; the ~$0.7B difference is supplier-financing/lease obligations — immaterial to the thesis.)

Returns on capital sit below the cost of capital. Reported ROIC has run ~5.4–5.6% (2020 5.6%, 2024 5.4%) with negative readings in impairment years, and ROA ~1.8% (FACT — ROIC.ai). Against a utility WACC comfortably above that, AES has been a capital destroyer on a through-cycle basis (Greenwald/Marathon lens): high asset growth funded with cheap non-recourse leverage, returns that don’t clear the hurdle, and periodic write-offs confirming the capital was mis-deployed. Dilution/SBC is modest (regular dividend ~$0.70/share, $501M paid in 2025); the value leakage is through sub-hurdle reinvestment and impairment, not share count.

Verdict (Financial Quality): POOR-to-MEDIOCRE, and largely moot at the deal price. A low-return, highly-levered, emerging-market-exposed utility whose GAAP earnings are uninformative, whose true owner cash flow (~$1.2B parent FCF) is a fraction of headline OCF, and whose ROIC has not cleared WACC. The non-recourse structure limits catastrophic downside but does not make the economics good. The one genuine positive — a large, contracted renewables backlog and rising subsidiary distributions — is exactly what attracted an infrastructure-fund buyer willing to own the asset base off the public market. For a public equity holder today, financial quality is a backstop-value question, not a compounding question: the relevant number is not ROIC but the $15.00 in cash and whether it arrives.

7. Capital Allocation

AES’s capital-allocation record is the record of a serial growth-spender that has consistently deployed capital faster than it generates it, at returns at or below its cost of capital — the precise pattern Marathon’s Capital Returns framework flags as value-destructive, softened only by the regulated-utility share of the portfolio.

The growth engine and the funding gap. Under CEO Andrés Gluski (CEO since September 2011), AES pivoted from a global merchant-power conglomerate to a “renewables-plus-US-utilities” growth story, targeting a multi-gigawatt renewables build and rate-base growth at AES Ohio and AES Indiana. The spend is enormous and chronically exceeds internal cash generation: operating cash flow ran $2.7–4.3B annually over 2020–2025 (FY25 $4.3B), while investing outflows — overwhelmingly growth capex — ran $5.8–8.2B per year (FY24 –$7.7B; FY23 –$8.2B) (FACT). The result is persistently negative free cash flow after growth capex, structurally plugged by three levers: (1) non-recourse project debt at the subsidiary level; (2) an asset-recycling program (a stated ~$400–500M/year asset-sale target, plus larger stake sales — CDPQ ~30% of both US utilities, AES Brasil, AGIC $450M); and (3) hybrid/equity issuance — most visibly the March-2021 Equity Units (~$1.0B mandatorily-convertible preferred) and ongoing equity financing.

Returns on the spend — the red flag. The growth capital has not earned attractive returns. ROIC was ~5.4% (2024), negative ~(11)% (2023) and ~5.6% (2020) (FACT — ROIC.ai). GAAP net income is volatile and frequently negative — –$951M (2021), –$505M (2022), –$182M (2023), +$802M (2024), +$162M (2025) — because AES books large asset impairments essentially every year: $1,575M (2021), $1,715M (2022), $1,079M (2023), $374M (2024) and $337M (2025), the latest including the January-2026 write-down of the Maritza plant in Bulgaria (8-K, Item 2.06). These recurring impairments are the accounting confirmation that a material slice of prior capital deployment — legacy coal and international merchant assets — destroyed value, and they distort any run-rate read of earnings. (Reported ROE — a “191%” 2025 return-on-common-equity from ROIC.ai — is a meaningless thin-common-equity artifact; use ROIC, not ROE, for AES.)

Dividend. AES has grown the dividend steadily — cash dividends paid rose from $381M (2020) to $501M (2025), ~$0.70/share (the $0.1759 quarterly rate), historically at ~4–5%/year (FACT). But the dividend is not covered by post-growth-capex free cash flow; it is funded from the same asset-sale-and-leverage stack that funds growth. For a company simultaneously financing a capital-hungry build-out, a growing payout on negative post-capex FCF is an aggressive posture leaning on parent leverage and recycling proceeds.

Balance sheet. The critical distinction is recourse (Parent) vs. non-recourse (subsidiary) debt: at 31-Dec-2025, ~$6.0B was recourse Parent debt and ~$23.2B was non-recourse project/subsidiary debt (~$29.2B total). Much of the non-recourse debt is ring-fenced, but the parent-level ~$6B — plus the hybrid units — is what constrains equity value and drove the standalone plan’s reliance on continued asset sales and equity.

Management & incentives. Gluski (68) remained CEO through the latest filings; concurrent with the merger signing, the Board elevated Ricardo Falú (former EVP/COO) to President effective 2-March-2026 — a succession step. Notably, Gluski took no base-salary increase since 2021, and his 2025 long-term-comp target was set ~30% below the prior year — modest optics that do not offset the underlying return shortfall.

Verdict (Capital Allocation): WEAK-to-MEDIOCRE. Management executed a genuine strategic transformation and grew the dividend, but funded a chronically FCF-negative growth program with debt, hybrids and asset sales while earning ROIC at or below cost of capital and taking recurring impairments — value creation was not evident in returns. That the company is being taken private at $15.00 — a price the board’s own DCFs bracket but do not clear on a midpoint basis — is itself a comment: the public market never rewarded the growth spend, and infrastructure sponsors, not public shareholders, will now harvest whatever returns the asset base ultimately delivers.

8. Changes and Headwinds — Last Two Years

The transformational change: the take-private. The dominant event of the period is the 1-March-2026 definitive merger agreement to be acquired for $15.00/share in cash by a consortium of Global Infrastructure Partners (part of BlackRock), EQT Infrastructure VI, CalPERS (via the GIP-managed “Tower Bridge” vehicle) and the Qatar Investment Authority, through the buyer entity Horizon Parent, LLC (FACT — DEFM14A filed 15-May-2026; merger 8-K 2-March-2026). Total equity value ~$10.69B, all-equity financed with no financing condition. The premium was ~35.5% over the unaffected 8-July-2025 close (the last trading day before an FT deal report) and ~40.3% over the 30-day VWAP to that date. Shareholders approved the deal on 26-June-2026 (For 479,072,642 / Against 10,131,991 / Abstain 506,143 — ~97.8% of votes cast). The deal now awaits a broad slate of regulatory clearances (Section 9, Section 10).

Portfolio reshaping. Over the period AES: sold AES Brasil (2024), removing ~$616M of revenue; sold ~30% of DPL/AES Ohio to CDPQ (April-2025) and holds a comparable ~30% CDPQ stake in IPALCO/AES Indiana; monetized renewables-development minority stakes (e.g., AGIC ~$450M) and Dominican Republic renewables; and continued retiring coal capacity. The Bulgaria Maritza PPA expiration drove a $264M impairment (recognized 2025; a related charge flagged in the January-2026 8-K).

Financial/operational developments. Constructive US rate outcomes (AES Indiana +$71M/yr, AES Ohio distribution +$168M/yr); ~3.2 GW of renewables construction completed in 2025 and 4.0 GW of new PPAs signed; a June-2026 debt issuance ($600M 5.200% notes due 2029 + $400M 5.750% notes due 2033) refinancing/pre-funding ahead of the change-of-control. Interest expense rose to ~$1.4B as rates and project debt climbed.

Leadership. CEO succession is in motion — Ricardo Falú appointed President (2-March-2026) alongside long-tenured CEO Andrés Gluski.

Headwinds. (i) IRA tax-credit rollback risk to US renewables economics; (ii) emerging-market FX/sovereign stress (Argentina, Chile, Colombia, Bulgaria); (iii) high leverage into a higher-rate refinancing environment; (iv) the political backlash against private-equity ownership of utilities and against “AI power” cost pass-through to consumers, visible in the news feed (e.g., a June-2026 Senator Warren critique of PE utility ownership) — a mild political overhang on the regulatory approvals the deal needs.

Verdict (Changes/Headwinds): The period’s changes converted AES from a struggling public compounder into a take-private target — a rational outcome given the sub-WACC returns and persistent discount. For the current equity holder these developments are dominated by the single fact of the pending $15.00 deal; the operating headwinds now matter chiefly as inputs to deal-completion risk and to the standalone downside if the deal breaks.

9. Risk Analysis

With shareholders having approved the merger on 26-June-2026, AES has ceased to trade on fundamentals and now trades as a merger-arbitrage instrument. The risk profile therefore inverts the usual utility analysis: idiosyncratic operating risks matter chiefly insofar as they threaten deal completion, and the dominant risk is binary deal-break. The pre-deal fundamental risks are retained below because they define the standalone downside that re-emerges if the deal fails.

Verdict (Risk): AES’s rich set of fundamental risks — EM sovereign/FX, leverage, policy, execution — are real but subordinated to the single binary question of deal completion. The $588M reverse termination fee and a well-capitalized infrastructure consortium argue for a high close probability; CFIUS and the breadth of foreign approvals are the genuine tail. A catastrophic permanent loss is unlikely (a break returns the stock to ~$10–13, not to zero), but the asymmetry at $14.78 is unfavorable in isolation: ~1.5% of upside to $15.00 against ~15–30% of standalone downside on a break — the classic negatively-skewed arb payoff, partially compensated by the dividend carry to close.

10. Valuation Discussion (Embedded Expectations)

The deal price is the valuation. With a signed, shareholder-approved cash merger at $15.00/share, AES’s equity value is anchored to deal-completion probability and timing, not to intrinsic-value estimation. At $14.78 (10-July-2026) the market embeds a ~1.5% gross spread ($0.22/share) to the consideration, plus the value of the contractually-permitted ~$0.176/quarter dividend collectible until close and a prorated “stub” dividend at closing. On a guided late-2026/early-2027 close (roughly two to three quarters out), a holder would collect the spread plus ~two-to-three dividends (~$0.35–$0.53), for a total absolute return of roughly ~3.9–5.0%, annualizing to a ~5–7% arb yield — the compensation the market is offering for bearing the regulatory tail. If close slips toward the ~December-2027 outside date, more dividends accrue but the annualized return compresses.

What must be true for $14.78 (embedded expectations). The market is underwriting a high (but not certain) probability of close near $15.00 within roughly 6–18 months, with the dividend uninterrupted. The residual ~1.5% discount is the market’s price for CFIUS/foreign-approval/financing tail risk and time value. This is a coherent, arb-consistent expectation — the debate is purely whether the ~5–7% annualized return adequately compensates for a ~15–30% left tail.

The board’s own fundamental valuation (standalone anchor). The two fairness opinions bracket the $15.00 price and reveal how the advisors valued the standalone business — the relevant range if the deal breaks:

• J.P. Morgan: sum-of-the-parts DCF (by-country WACC) $10.50–$20.25; comparable-company/EV-EBITDA analysis (2025 FV/EBITDA 9.3x–10.8x) $11.25–$17.75.
• Wells Fargo: DCF $11.14–$17.27; selected-public-companies $11.31–$16.39. The $15.00 consideration sits inside all four ranges — above the midpoint of the comps ranges, below the DCF midpoints. The wide DCFs reflect AES’s country-by-country risk and leverage; the comps cluster in the low-to-mid teens, consistent with the pre-rumor trading price.

Standalone/relative context (for the break scenario). Pre-rumor, AES traded around $10–13 through H1-2025 (a ~$9.46 low in May-2025). On the deal EV of ~$33.4B (per the company’s deal press release; the proxy does not restate a single EV figure — the FT rumor cited ~$38B) against ~$3.4B EBITDA, the take-out is ~9–10x EV/EBITDA — a full-but-not-rich infrastructure multiple, in line with the fairness comps and above where the public market was willing to pay (the stock’s own AZI valuation percentiles sit mid-range on its 10-year history: P/E ~19th, P/B ~34th, P/S ~57th, composite ~37th — i.e., the deal rescued a de-rated, not an expensive, stock). A deal break would likely return the equity to the low-to-mid-teens-or-below, toward the bottom of the fairness ranges and the pre-rumor tape.

Verdict (Valuation): The equity is priced as a merger-arb bond: ~1.5% spread + dividend carry to a $15.00 cash close, against a standalone fundamental value (per the board’s fairness opinions and pre-rumor trading) in the low-to-mid teens that would re-emerge on a break. There is no independent “intrinsic value” call to make here — the price is the deal, and the analysis is a probability-weighting of close-at-$15 vs. break-to-~$11. No price target; no BUY/SELL (those live only in the Author’s Take).

11. Variant Perception

Consensus. The market treats AES as a near-closed merger-arbitrage position. At $14.78 versus $15.00 cash, the stock is pinned to a ~1.5% gross spread plus continued ~$0.176/quarter dividends until close (expressly permitted by the merger agreement). Shareholders approved on 26-June-2026; the debate is no longer about AES’s business but about when the cash arrives and whether any regulator blocks it. The tape confirms it: dead money in a tight band just below the deal price.

The bull case (deal closes). Approvals clear across all jurisdictions (including CFIUS), financing funds (all-equity, no financing condition), and the deal closes in the guided window. The arbitrageur collects the ~1.5% spread plus one-to-three regular dividends for a low-single-digit absolute return that annualizes into a respectable, low-correlation ~5–7% arb yield. Supports are concrete: a well-capitalized consortium (GIP/BlackRock, EQT, CalPERS, QIA), a $588M reverse termination fee aligning the buyer to complete, and shareholder approval already secured.

The bear case (deal breaks). A regulator — most plausibly CFIUS, given Qatari (QIA) and European (EQT) sponsors acquiring a large US critical-infrastructure operator — imposes an unacceptable condition or blocks the deal, or a financing/MAC dispute derails it. AES re-rates to standalone value. It traded around $10–13 through 2025 before takeover speculation (a ~$9.46 low in May-2025), and the underlying fundamentals are unattractive: ~5% ROIC below WACC, ~$29B consolidated debt, EM/FX exposure, $1.4B annual interest. A break likely returns the stock to ~$10–13, a ~15–30% drawdown — the negatively-skewed arb payoff.

The 3–5 assumptions that matter: (1) probability of close — the whole position; reverse fee + shareholder approval imply high odds, CFIUS the residual tail; (2) timeline — every quarter of slippage cuts the annualized return; (3) dividend continuation — the arb carry, contractually permitted; (4) standalone downside on a break — the size of the left tail (~$10–13); (5) no topping bid / no price cut — the definitive $15.00 came below the rumor-inflated ~$16.7 (the stock fell ~16% on signing), so a higher competing bid is unlikely and $15.00 is a ceiling.

Factor overlay as evidence. The factor read reinforces that consensus is not offsides — it is simply frozen. Pre-deal, AES was a chronic value-destroyer: a five-year annualized return of −7.4%, a five-year Sharpe of −0.25, and a −79% lifetime max drawdown (FactorsToday). The +16.8% one-year return is entirely the deal pop. The stock has been surgically removed from factor space and converted into a fixed-income-like arb instrument (beta collapsing toward the deal, idiosyncratic vol suppressed). There is no contrarian “abandoned value” mispricing to exploit and no momentum to ride — only spread capture against binary regulatory risk.

What would falsify each side. Bull falsified: a CFIUS second request/mitigation demand, a foreign-regulator objection, or a financing wobble — any would blow the spread out well beyond 1.5% and signal rising break odds. Bear falsified: clean, on-schedule approvals (CFIUS clearance, FERC/PUC/EU/foreign sign-offs) narrowing the spread toward zero into the guided close, dividend uninterrupted.

Verdict (Variant Perception): Consensus is correct in substance — this is an arb, not an investment. The variant question is purely probabilistic: is ~1.5% + dividend carry adequate compensation for CFIUS/foreign-approval tail risk against ~15–30% of standalone downside? That risk/reward judgment (and any position stance) belongs to the Author’s Take.

12. Fact vs. Interpretation

13. Open Questions

1. Exact Parent Free Cash Flow figure — used management’s guided ~$1.15–1.25B; reconcile to the FY25 earnings release/MD&A for the precise reported number.
2. CFIUS pathway and timing — is a formal CFIUS filing already accepted; is a mitigation agreement (proxy board, US-person governance) contemplated for the QIA/EQT interests? Not disclosed in detail.
3. Per-sponsor equity ownership split — GIP/EQT/CalPERS/QIA percentages are not disclosed.
4. Precise enterprise value — the company press release cited ~$33.4B; the proxy does not restate a single EV figure (FT rumor cited ~$38B). The gap reflects non-recourse/minority treatment.
5. Standalone plan if the deal breaks — would management resume asset sales/equity to fund the build, and is the dividend safe absent the deal? Not addressed post-signing.
6. Foreign approvals still outstanding — which specific country FDI/competition clearances remain the long pole (EU FSR, Chile, Brazil, Colombia)?

14. What Must Be True

Bull (the arb pays):

• The consortium clears CFIUS and every named regulatory approval (FERC Section 203, HSR, Ohio PUCO, NY PSC, CA/NC, EU EUMR + FSR, and the full list of foreign FDI/competition regimes) without an unacceptable “Burdensome Condition,” and closes near the guided late-2026/early-2027 window.
• AES continues paying its ~$0.176/quarter dividend through close, delivering the carry on top of the ~1.5% spread.
• Falsification test: a CFIUS second request or mitigation demand, a foreign-regulator objection, or a financing/MAC dispute — any of which pushes the spread materially wider than ~1.5% and signals rising break odds. If the spread blows out past ~5%, the bull thesis is breaking in real time.

Bear (the deal breaks / value re-emerges):

• A regulator blocks or fatally conditions the deal, or the buyers walk (paying the $588M reverse fee), and AES reverts to a standalone, sub-WACC, highly-levered, EM-exposed utility worth ~$10–13.
• The fundamentals reassert: ~5% ROIC below cost of capital, recurring impairments, negative post-capex FCF, and renewed reliance on asset sales/equity to fund the build and the dividend.
• Falsification test: clean, on-schedule approvals (visible CFIUS clearance and staged FERC/PUC/EU sign-offs) that narrow the spread toward zero into a confirmed close date. If the deal closes at $15.00, the bear standalone case never gets tested.

15. Source Appendix

See the separately-stitched Appendix B — Source Appendix for the full source list with URLs and access dates (SEC filings, ROIC.ai/AZI/FactorsToday data pulls, and news items). Primary sources — AES FY2025 Form 10-K (2-Mar-2026), the DEFM14A merger proxy (15-May-2026), the merger and vote 8-Ks, and the Form 4 corpus — govern; third-party aggregated data (ROIC.ai, AZI, FactorsToday) is used for cross-checks and reconciled to filings.

APPENDIX A — Standard Diligence Questionnaire

The AES Corporation (NYSE: AES) — as of 2026-07-11. Supplemental to the memo. Fact/Interpretation/Assumption labels used where material. AES is subject to a signed, shareholder-approved $15.00/share cash take-private (GIP/BlackRock + EQT + CalPERS + QIA); answers reflect that the equity now trades as a merger-arbitrage instrument.

General

What thoughtful questions have other investors asked? (1) Will CFIUS clear a consortium that includes a Qatari sovereign fund and a European sponsor buying a large US critical-infrastructure operator? (2) What is the realistic close timeline against the 1-Jun-2027 (→ ~Dec-2027) outside date? (3) Is the ~$0.176/quarter dividend safe through close (yes, contractually permitted)? (4) What is the standalone downside if the deal breaks (~$10–13)? (5) Could a topping bid emerge (unlikely — the definitive $15.00 came below the rumor-inflated ~$16.7)? (6) Are the fairness-opinion DCFs credible given AES’s country-by-country risk and leverage?

Cyclicality & Earnings Nature

Cyclical high or low? GAAP earnings are dominated by impairments/FX/derivative marks and are near-meaningless as a cycle read; Adjusted EPS ($2.34 in 2025) is near a normalized level, not an obvious peak or trough (Interpretation). External vs. internal drivers? Both: regulated-utility earnings are internally-driven (rate cases), while the larger generation/renewables leg is exposed to external power/commodity prices, hydrology, FX and IRA policy. Revenue stability? Utility (~33%) and long-dated PPA revenue is annuity-like; merchant/EM revenue is volatile. Consolidated revenue has been flat-to-down for three years (~$12.2–12.7B). Outlook / market size? US power demand is inflecting up on data-center/AI load — a real multi-decade tailwind — but shared by all developers; internationally, emerging-market growth carries FX/sovereign risk. Large and growing market, weak pricing power.

Business Quality & Competitive Moat

Industry more or less competitive? More — renewables development is a capital-cycle magnet drawing NextEra, Brookfield, oil-majors and infrastructure funds; returns are competing down (Interpretation). How profitable (ROIC/ROE)? ROIC ~5–5.6% (negative in impairment years), below a ~7–8%+ cost of capital — value-destructive. Reported ROE (~191% in 2025) is a thin-common-equity artifact — do not use it; use ROIC (GOTCHA). Industry profitability / barriers? Utilities: high barriers (franchise monopoly), capped returns. Generation/renewables: low barriers, commodity inputs, price-taker. Easily understood? No — a sprawling, multi-country, multi-segment holdco with opaque HLBV/tax-equity accounting. Undermined by low-cost foreign labor? N/A (capital-intensive infrastructure), though panels/turbines are globally sourced and tariff-exposed. Do brands matter? No. Switching costs? Contractual within a PPA term; none at the point of winning the next auction.

Financial Condition & Balance Sheet

Unrecognized assets? The 46 GW development pipeline and powered-land/interconnection positions carry option value not on the balance sheet; the 28% Fluence stake is equity-method. Off-balance-sheet liabilities? PPA/purchase commitments, project guarantees and EM contingencies; most project debt is on-balance-sheet but non-recourse. How conservative is accounting? Mixed — heavy use of non-GAAP Adjusted EPS/EBITDA and HLBV tax-equity accounting; recurring impairments suggest carrying values were previously optimistic (Interpretation). CapEx-hungry? Extremely — capex $6–7.7B/yr, ~2–2.7x Adjusted EBITDA; the defining feature of the business.

Capital Allocation & Management

FCF generation and use? Consolidated OCF $4.3B (2025) but Parent FCF only ~$1.15–1.25B; nearly all is reinvested in the renewables build, with the gap plugged by project debt, tax-equity, minority sell-downs and asset sales. Philosophy? Growth-first capital recycling — build renewables organically, sell mature/fossil/EM assets and utility minority stakes (CDPQ ~30% of both US utilities) to fund it. Significant acquisitions? Net a seller over the period (AES Brasil 2024, AGIC, DR renewables). Buybacks? No meaningful buyback; capital goes to growth and the dividend. Issuing shares to insiders? Modest equity comp; 2021 ~$1.0B Equity Units (hybrid). Compensation policy? CEO Gluski took no base raise since 2021 and a ~30%-lower 2025 LTI target — restrained optics. Management motivations? Long-tenured CEO (Gluski since 2011) executing a transition; Falú elevated to President (Mar-2026) for succession. Insiders bought near the 2023 lows (Gluski 50k @ $16.38) but made no open-market purchases in 2025–26.

Valuation & Market Data

ADR/MLP/K-1? No — ordinary US C-corp common stock (NYSE: AES), 1099 dividends. Dividend policy? ~$0.70/share (~$0.176/quarter), ~4–5%/yr growth historically, ~4.8% yield at $14.78; contractually permitted to continue through the merger close. Profitability? Low-return (ROIC < WACC). Net income vs. cash from operations diverging? Yes and structurally so — consolidated OCF (~$4.3B) vastly exceeds GAAP net income to common ($910M) and, more importantly, dwarfs Parent FCF (~$1.2B); the divergence reflects NCI/tax-equity and project-level cash that does not accrue to AES common.

Risks & Downside

What would cause the stock to decline? A deal break — CFIUS/foreign-approval failure, financing/MAC dispute — re-rating the stock to a ~$10–13 standalone value (~15–30% downside); secondarily, any pre-close news lengthening the timeline (spread-widening). Catastrophic loss risk? Low — the equity is backed by $15.00 cash and, on a break, by hard infrastructure assets worth low-to-mid-teens; not a zero. Total loss? Very unlikely.

Recent News & Events

Environment changed recently? Transformationally — the 1-Mar-2026 $15.00 cash merger agreement and 26-Jun-2026 shareholder approval converted AES from a public utility into a pending take-private. Acquisitions? AES is the target, not the acquirer. Accounting-policy changes? None material flagged. Recent changes — markets/facilities/management? Falú appointed President (Mar-2026); a June-2026 $1.0B senior-notes issuance; ongoing coal retirements and renewables construction (3.2 GW built in 2025); CDPQ’s ~30% AES Ohio purchase (Apr-2025). A mild political overhang exists around PE/foreign ownership of US utilities and “AI power” cost pass-through (news feed, mid-2026).

APPENDIX B — Source Appendix

The AES Corporation (NYSE: AES) — sources for the 2026-07-11 research memo. Primary sources (SEC filings) govern; third-party aggregated data (ROIC.ai, AZI, FactorsToday) is used for cross-checks and reconciled to filings. Access date 2026-07-10/11 unless noted.

Primary — SEC Filings (EDGAR CIK 0000874761)

Third-party quantitative (cross-check; reconciled to filings)

News / secondary

Analytical frameworks

investment-research-frameworks skill — Greenwald & Kahn Competition Demystified (moat taxonomy, ROIC vs. WACC test) and Marathon Capital Returns (capital-cycle / asset-growth-anomaly lens) applied to Section 4, Section 5, Section 6, Section 7.