The AES Corporation AES

Key data

The business

AES is a global power company with four kinds of business under one roof. Two US regulated utilities, AES Ohio and AES Indiana, earn a set return on the poles, wires, and plants their regulators approve. A large and growing renewables arm builds solar, wind, and battery storage, then signs 15-to-20-year contracts to sell that power, increasingly to data centers and technology customers. A legacy fleet of gas and coal plants across Latin America and Asia throws off cash while the company sells or retires it. The moat is two-fold and real: a regulated monopoly you cannot bypass in Ohio and Indiana, and long-dated contracts that lock in the price of every renewable megawatt before it is built. Management has spent two years narrowing the map, fewer countries, larger projects, out of coal, to point the balance sheet at the parts that compound.

The numbers

The five quarters show a business swinging back to profit as its Latin American plants and new renewables come online.

The inflection was the September 2025 quarter, when net income jumped to $639 million. The June quarter carried $426 million against a $95 million loss a year earlier, revenue up 20%. On the adjusted basis AES guides to, it earned $0.44, a penny under the $0.45 estimate, ending three straight quarters of clearing its own bar handily. The GAAP figures in the table run higher than the adjusted number because of one-time gains and currency swings across a dozen countries.

The full years tell the harder story: revenue flat around $12 billion for five years, earnings that lurch from losses in 2021 and 2022 to a $2.36 peak in 2024 and back to $1.26 as one-time gains rolled off. The last table is the whole argument in one place, cash made versus cash spent.

Here is what you are buying. AES has spent more cash building than it collects from operations every year for five straight years. In 2023 and 2024 the gap ran over $4.5 billion, plugged by debt and asset sales. The build is now moderating as management narrows focus, and 2025 free cash flow, still negative, improved to about negative $1.6 billion. That matters because net debt is roughly $30 billion, about 6.7 times EBITDA, with interest covered only 1.6 times. The market prices AES as a levered laggard, and the single print that settles whether it is instead a self-funding data-center power supplier is net-debt-to-EBITDA falling quarter over quarter as capex drops and contracted plants switch on.

Management

Andrés Gluski runs AES and has spent the back half of his tenure remaking it, out of coal, out of a dozen small countries, into US regulated utilities and contracted renewables. Insiders neither bought nor sold a share in the open market over the past year, so there is no signal from their own money either way. The company pays a dividend and has not meaningfully repurchased stock in years, with every spare dollar and then some going into the capital plan. The record on guidance is decent, three clean quarters over the line before the June penny miss, and the company reaffirmed the multi-year earnings-growth target. What the pay rewards is the earnings-growth figure the build is supposed to deliver, the same figure the debt is financing, so the incentive and the risk point the same way.

How it fails or surprises you

The debt is the whole risk. Net debt is about $30 billion, 6.7 times EBITDA, and operating profit covers interest just 1.6 times. The equity is a thin $10.5 billion sliver sitting under that pile. If rates stay high or a rating agency cuts the outlook, the entire capital plan gets more expensive and the equity absorbs it first. Watch net-debt-to-EBITDA and coverage each quarter.

It has never funded its own growth. Free cash flow has been negative every year for five years, plugged by debt and asset sales. That is the fact the cheap-utility story explains least well: a six-times-earnings multiple assumes durable earnings, but if asset-sale proceeds dry up before capex falls, debt compounds faster than EBITDA and the multiple is cheap for a reason.

Data centers could re-rate it (right tail). If the AI power build accelerates and AES fills its renewables backlog at firmer contract prices, EBITDA compounds and the stock re-rates off book value near a twelve-year low. The market pays today for a levered Latin American laggard, not a data-center supplier. The first tell is signed contract volume and how fast the backlog converts to operating plants.

Closing thoughts

This is priced about right, and knowing that is the edge. At $14.79, a little over two times book and six times next year's earnings, AES is valued where a heavily levered utility with negative free cash flow belongs. The other side of the table is everyone who will not own 6.7-times-levered EBITDA, and they are not wrong to hesitate, so your read only beats theirs if the leverage is falling, not rising. A specific print settles it, net-debt-to-EBITDA trending down as capex moderates and contracted plants come online, and an ambiguous print, flat leverage against a flat backlog, means you wait and collect the dividend. The take-private talk that hung over this stock has not broken it to a distressed number, and it carries no visible deal premium at today's price either, so you own the standalone utility on its own merits regardless. The left tail, a refinancing squeeze if rates spike, is real and would be permanent. The right tail, a data-center re-rating, is worth more than the downside if the build lands, but the fatter near-term risk is the balance sheet.

The bet is still that AES builds solar, wind, and batteries to power data centers and sells electricity across Latin America while carrying thirty billion dollars of debt. What breaks it is the debt itself, not the business. The one pair of numbers that tells you first is net-debt-to-EBITDA and interest coverage. If those two move the wrong way for two straight quarters while the backlog stalls, the cheap multiple was correct pricing, not an opportunity.