Energy · Deep dive

AES Corporation

The 45-year-old global power company that nearly died with Enron, reinvented itself as the world's biggest seller of clean energy to hyperscalers — and is now being carried off the public market by BlackRock's GIP and EQT at $33.4B enterprise value, a 40% premium to a share price the equity market had left for dead.

well positioned

The equity market punished AES for its $22.7B debt stack and tax-credit dependence, but an 11.1 GW contracted backlog sold to the most creditworthy buyers on earth, two regulated utilities riding double-digit rate-base growth, and sponsors who just removed its cost-of-equity constraint make this a position that compounds — the sellers, not the asset, were the problem.

Snapshot

AES is a global power company with roughly $12.3B of 2024 revenue, ~$2.87B of FY2025 adjusted EBITDA, two regulated US utilities (AES Indiana, AES Ohio), a Latin American and Asian generation fleet, and the thing that made it a takeover target: the largest corporate clean-energy sales franchise in the world, with an 11.1 GW backlog of signed power purchase agreements (Q3 2025) sold disproportionately to Amazon, Google, Microsoft and their peers. BloombergNEF ranked AES the top global seller of clean energy to corporations in 2021 and 2022. The public market never paid for any of it — the stock fell 57.5% from its December 2022 peak of $29.27 to $12.45 by October 2023 on rate, debt and tax-credit fears — so in March 2026 a consortium led by BlackRock’s Global Infrastructure Partners and EQT, with CalPERS and QIA co-underwriting, agreed to take AES private at $15.00 per share: $10.7B of equity, ~$33.4B of enterprise value including ~$22.7B of assumed net debt. Stockholders approved on June 26, 2026; close is expected late 2026 or early 2027. It is the largest utility take-private of the AI-power era and the clearest single bet that US electricity demand growth is real.

Founding story

AES was founded on January 28, 1981 as Applied Energy Services by Roger Sant and Dennis Bakke, two Federal Energy Administration alumni who saw that the 1978 PURPA law had cracked open the utility monopoly on generation. It began as a consulting shop, built its first independent power plants in the mid-1980s, went public in 1991, and rode the 1990s deregulation wave into a sprawling multinational — by 2000 it operated on five continents and traded near $70. Bakke, CEO from 1994, ran it on a famously radical “fun at work” decentralized philosophy. Then Enron collapsed, emerging-market currencies broke, and merchant power prices cratered: AES stock fell to roughly $1 in 2002, Bakke resigned mid-crisis, and successor Paul Hanrahan spent years selling assets and refinancing to keep the company alive. The near-death is load-bearing history — it explains both the company’s discipline about contracted (rather than merchant) revenue and the market’s long memory about AES leverage.

Andrés Gluski, a Venezuelan-American former IMF economist who joined in 2000 and ran Latin America, became CEO in September 2011 and never left. His pivot defines the modern company: coal, about 60% of capacity when he started at AES, was driven toward a full announced exit by end-2025; AES published the first TCFD-compliant climate report by a listed US energy company in 2018; and it co-founded Fluence, the grid-scale storage venture, with Siemens in 2017-18. On the other side of the table sits Bayo Ogunlesi, whose GIP — bought by BlackRock for ~$12.5B in 2024 — has made AES its signature post-acquisition move: the classic Ogunlesi playbook of buying irreplaceable infrastructure with contracted cash flows at a moment of public-market mispricing.

How it works

Mechanically, AES is three machines. The first is a development machine: originate a long-term power purchase agreement — typically 15-25 years, fixed or escalating price per MWh — with an investment-grade offtaker, most valuably a hyperscaler that needs new clean supply near its data centers. Only then does AES finance and build the wind, solar or storage project, using non-recourse project debt at the asset level plus tax equity or transferable tax credits. The PPA turns a construction project into a bond-like asset; the signed-but-unbuilt pipeline is the 11.1 GW backlog (Q3 2025), of which 5 GW was under construction. AES completed 3.2 GW of new projects in 2025 and targeted 4 GW of new PPA signings for the year. Microsoft alone has signed over 1 GW of PPAs with AES across the US and Latin America, including 475 MW of solar; Amazon and Google are repeat counterparties.

The second machine is regulated: AES Indiana and AES Ohio earn a state-approved return on an expanding rate base — wires, grid hardening, generation — growing about 10% a year, with data-center interconnection requests now driving load growth in both territories. The third is the legacy international fleet — gas, LNG terminals, hydro and renewables across Chile, Colombia, Panama, the Dominican Republic, Vietnam and elsewhere — largely contracted, still throwing off cash, steadily pruned by asset sales. Around the edges sit the technology stakes: 28.19% of Fluence (as of December 31, 2025), plus Uplight, the Maximo solar-installation robot, and 5B.

Product and business overview

Renewables (the growth engine). Utility-scale solar, wind and battery storage development and operation, ~52% of proportional MW in operation (Q2 2025 fact sheet). Renewables segment EBITDA grew 46% year-over-year in 2025 quarters as new projects came online.

US Utilities. AES Indiana and AES Ohio — regulated transmission, distribution and generation serving well over a million customers combined, with ~10% annual rate-base growth and mounting data-center load.

Energy Infrastructure. The international contracted fleet — gas-fired generation and LNG infrastructure in Latin America and Asia, plus the dwindling coal tail AES pledged to exit by end-2025 (the Puerto Rico coal plant being the most contested remainder).

New Energy Technologies. The Fluence stake (co-founded with Siemens 2017-18, IPO’d late 2021), Uplight, Maximo and the AI Fund — optionality the market valued at roughly zero by 2025.

Business model and pricing

Revenue books three ways. Contracted generation revenue: energy and capacity payments under long-dated PPAs, priced per MWh at contract rates set at origination — corporate renewable PPAs signed in the 2023-2025 window generally cleared in the $40-70/MWh range depending on market and technology, and AES’s data-center deals increasingly bundle storage. Regulated utility revenue: tariff-based, set by Indiana and Ohio commissions on allowed ROEs. And monetization: asset sell-downs to partners, and — post-IRA — sales of transferable tax credits, which is why AES guides to “Adjusted EBITDA with Tax Attributes” ($3.95-4.35B for 2025) alongside plain adjusted EBITDA ($2.65-2.85B guided; ~$2.87B delivered). That gap — roughly $1.3-1.5B of annual value flowing from US tax policy — is the heart of the bear case. The structure is heavily non-recourse: most of the ~$26.5B of consolidated long-term debt (September 30, 2025) sits at project subsidiaries against contracted cash flows, but the market priced it as one undifferentiated pile. The parent has paid a rising dividend since 2012 ($0.69/share 2023, $0.70 2024, ~$0.176 quarterly into 2026), which yielded near 7% at the lows — a yield that screamed distress.

Traction over time

Market analysis

The structural story is the first sustained US electricity demand growth in two decades. US data centers consumed 183 TWh in 2024 and are projected by S&P Global to reach 426 TWh by 2030 — a 133% increase — with data centers approaching 9-12% of all US power consumption by 2030 depending on the forecast; 451 Research pegs US data-center demand at 75.8 GW in 2026 rising to 134.4 GW by 2030. Amazon, Meta, Google and Microsoft alone accounted for about half of the 55.9 GW of global corporate clean-power PPAs signed in 2025 (BNEF), and deal sizes have jumped from sub-500 MW to gigawatt scale. The countervailing force is policy: the One Big Beautiful Bill Act (July 2025) phases out wind and solar tax credits on an accelerated clock, raising the cost of new renewables and pulling forward a construction rush before safe-harbor deadlines lapse. That squeezes late-decade supply exactly as demand accelerates — bearish for tax-credit-dependent developers’ margins, but bullish for the value of anything already contracted, interconnected or under construction. AES’s backlog is precisely that asset. Forecast dispersion is real (2030 estimates run 200 to 1,050+ TWh), and a slowdown in AI capex is the single biggest swing factor for the whole thesis.

Competitive intel

The named set is in the sidebar; the shape is this. NextEra outguns AES on cost of capital and pipeline depth. Constellation and Vistra sell what AES mostly cannot — firm nuclear and dispatchable power for loads that cannot flicker — and captured the AI-trade multiple AES never got. Brookfield Renewable runs the same hyperscaler-PPA model with Microsoft’s 10.5 GW framework (May 2024) as proof, and had patient sponsor capital first. Invenergy, Blackstone-backed at ~31 GW, shows the private version of the model working. AES’s edge: origination relationships (top BNEF corporate seller 2021-2022), speed to contract, storage integration via the Fluence lineage, US utility platforms with data-center load, and Latin American assets none of the US pure-plays contest. Its structural disadvantage — a public-equity cost of capital inflated by leverage fears — is exactly what the take-private erases. Notably, all five names above are collaborating with AES and NVIDIA on flexible “AI factory” grid integration, a reminder that in this market the constraint is supply, not share.

History and evolution

• January 28, 1981 — Founded as Applied Energy Services by Roger Sant and Dennis Bakke.
• 1991 — IPO.
• 1994 — Bakke becomes CEO; radical decentralization era; global expansion.
• 2001-2002 — Post-Enron collapse: stock from ~$70 to ~$1; Bakke resigns; Hanrahan restructures.
• September 2011 — Gluski becomes CEO; DPL (AES Ohio) acquired for ~$3.5B.
• 2017-2018 — Fluence JV with Siemens; sPower bought with AIMCo; TCFD climate report; carbon-reduction strategy set.
• Late 2021 — Fluence IPO at a multi-billion valuation.
• May 2022 — Third investment-grade rating attained; stock peaks $29.27 that December.
• 2023 — Coal exit by end-2025 announced (April); stock collapses to $12.45 (October) on rates, leverage and renewables selloff.
• May-July 2025 — House bill then OBBBA (signed July 4) guts renewable tax credits; AES falls ~22% in a week in May.
• July 8, 2025 — Last trading day before first media reports of takeover interest (the reference date for the 40.3% premium).
• October 1, 2025 — FT reports GIP in talks, ~$38B including debt; stock jumps 16.8% to $15.37.
• March 2, 2026 — Definitive agreement: $15.00/share, $10.7B equity, ~$33.4B EV, ~$22.7B net debt assumed; GIP/EQT lead, CalPERS and QIA co-underwrite.
• June 22-26, 2026 — HSR waiting period expires; stockholders approve with ~97.9% of votes cast (~67% of shares outstanding).
• Late 2026/early 2027 — Expected close, pending FERC and state regulatory approvals.

What people say

The case for. The sponsors are the loudest bulls: GIP framed the deal as backing a leader in competitive generation at a moment demanding massive investment in US capacity, and the consortium committed to fund growth “beyond 2027” that the public market would not finance. Sell-side coverage through 2025 kept flagging the disconnect — Seeking Alpha contributors called AES “oversold for a utility with dividend and growth,” and Q3/Q4 2025 results beat expectations (Q4 adjusted EPS $0.81 vs. $0.61 consensus) with renewables EBITDA up 46%. BNEF’s 2021-2022 rankings and the Microsoft/Amazon/Google repeat business validate the origination franchise. Glassdoor reviews are strikingly good for a utility — 4.2/5 across ~950+ reviews, 85% recommend — citing mission and flexibility.

The complaints. They are structural, and they were winning until the bid arrived. Balance-sheet critics pointed to consolidated leverage (long-term debt ~$26.5B at September 30, 2025), a debt-to-equity ratio near 8.8, one commentator’s Altman Z-score of 0.48, and a dividend not covered by free cash flow — feeding persistent dividend-cut speculation. Tax-credit dependence is the sharpest knife: with “EBITDA with tax attributes” roughly $1.3-1.5B above plain EBITDA, OBBBA’s credit phase-out threatens the economics of everything not yet safe-harbored, and skeptics called post-2028 growth an open question. Some shareholders called $15.00 a lowball — the 40.3% premium was measured against a depressed pre-leak price, the stock had briefly traded above $15 on the October rumor, and Seeking Alpha ran pieces litigating whether the takeover was “robbery” of the AI-power upside. Environmental critics have decades of material: Earthjustice documents 20+ years of coal-ash contamination claims around AES Puerto Rico, an EPA settlement over coal-ash compliance, and a $6M Dominican Republic cleanup settlement — a legacy that complicates the clean-energy narrative. And internally, some Glassdoor reviewers describe hierarchy, reorganization fatigue and layoff cycles despite the high average rating.

Outlook: well positioned or at risk?

Well-positioned — because every input that made AES look broken as a public stock makes it valuable as private infrastructure. The market’s three complaints were leverage, tax-credit dependence and a dividend it couldn’t really afford. Under GIP/EQT/CalPERS/QIA, the dividend obligation disappears, the debt is re-underwritten by owners who price project-level non-recourse structures properly, and the equity capital that AES could not raise at 6x-forward-earnings prices arrives from the deepest pools in the world. What remains is the asset: an 11.1 GW backlog (Q3 2025) contracted to hyperscalers whose credit is better than most sovereigns, two regulated utilities compounding rate base at ~10% a year into genuine data-center load growth, and an origination team that outsold every rival on earth in 2021-2022. Demand is the tide — US data-center consumption headed from 183 TWh (2024) toward ~426 TWh (2030, S&P Global) — and OBBBA, perversely, helps incumbents by strangling future supply while grandfathering the safe-harbored pipeline.

The honest risks: an AI-capex pause would hit the growth PPA machine first, and forecast dispersion on data-center demand is enormous. Post-2028 project economics without tax credits are unproven — renewables must compete on raw LCOE plus storage against gas and nuclear restarts, and Constellation-style firm power is winning the premium AI contracts today. Latin American currency, hydrology and political risk never went away. And LBO history says $22.7B of assumed net debt plus new acquisition financing leaves little room for execution error if rates stay high. But “at-risk” requires a plausible path to dislocation, and there isn’t one: the contracts are signed, the buyers are locked, the load growth is physical, and the company’s binding constraint — the public market’s refusal to fund it — is being surgically removed. AES spent four years as the wrong asset in the wrong wrapper. The wrapper is what changed.

Sources and further reading

• Consortium Led by Global Infrastructure Partners and EQT Agrees to Acquire AES — PR Newswire, March 2, 2026. Deal terms: $15.00/share, $10.7B equity, ~$33.4B EV, consortium composition.
• AES Stockholders Approve Acquisition by Global Infrastructure Partners and EQT-Led Consortium — PR Newswire, June 26, 2026. Vote result and closing timeline.
• EQT, GIP Move to Take AES Private in $33B Bet on Data Center Power Demand — POWER Magazine, March 2026. 40.3% premium, ~$22.7B net debt, deal rationale.
• AES Reports Third Quarter 2025 Results — PR Newswire, November 2025. 11.1 GW backlog, EBITDA, guidance.
• AES Outlines Long-Term Strategy for Growth in Renewables & US Utilities, While Executing on Intent to Exit Coal by Year-End 2025 — PR Newswire, April 2023. Coal exit, 25-30 GW renewables plan, utility rate-base growth.
• Microsoft signs solar PPAs totaling 475MW with AES — Data Center Dynamics. Hyperscaler PPA detail; >1 GW cumulative with Microsoft.
• Data center grid-power demand to rise 22% in 2025, nearly triple by 2030 — S&P Global, October 2025. 183 TWh (2024) to 426 TWh (2030) demand forecast.
• Renewable Energy Stocks Crash as U.S. Advances Bill That Could Decimate the Industry — The Motley Fool, May 23, 2025. AES’s ~22% one-week drop on the tax-credit bill.
• Why The AES Corp. Takeover Isn’t The ‘Robbery’ You Think It Is — Seeking Alpha, 2026. The shareholder debate over the $15 price.
• AES Puerto Rico: 20 years of coal ash contamination — Earthjustice. The coal-ash record and EPA settlement context.
• History of AES Corporation — FundingUniverse. 1981 founding, Sant and Bakke, 2002 crisis.