There is a moment when an investment thesis stops being an estimate and becomes a number in a report. In our pieces on power scarcity and the data-center grid bottleneck we wrote that electricity had turned from a transparent given into a scarce asset, and that whoever manufactures the equipment sits on rare pricing power. This morning GE Vernova (NYSE: GEV) reported its second quarter - and this is exactly the thesis, written in the language of the balance sheet.
But a good report is an opportunity to read closely, not to celebrate. So we will do two things here: break down the business engine - what this company actually sells and how it earns - and then go through the numbers from both sides. Because alongside the impressive headlines, this report also holds a bleeding division, an accounting gain that is easy to be confused by, and cash flow that needs to be understood for where it really came from.
Who Is GE Vernova, Anyway
In April 2024, General Electric - one of the companies most identified with 20th-century America - finally broke into three. The energy division was spun off into a standalone company called GE Vernova, and GE shareholders received one share of it for every four shares they held. What they received, in practice, is an extraordinary infrastructure asset: per company data, about 25% of the world's electricity is generated today on equipment it makes - an installed base of about 7,000 gas turbines and about 59,000 wind turbines.
The business is built of three divisions, and it is important to know the difference between them because this quarter demonstrates it brutally:
- Power - gas, nuclear and hydro turbines. The core, and the engine of the quarter.
- Electrification - transformers, switching, substations and high-voltage transmission. Everything that moves electricity from where it is created to where it is consumed.
- Wind - onshore and offshore wind turbines. The company's problem.
The Business Model: The Machine Is Sold Once, the Service Forever
Here is the heart of it, and it is a model worth knowing because it recurs across the entire heavy-equipment industry: selling the machine is only the entry ticket. A gas turbine works for decades, and throughout all those years it needs spare parts, upgrades, inspections and maintenance - from the manufacturer, which is usually the only supplier who can provide them. The result: every dollar of equipment creates a long, stable and far more profitable service revenue stream. Per the company, more than 55% of the backlog is services - meaning most of what is on the books is not machine sales but an annuity on an existing installed base.
And this quarter, a second and even more interesting mechanism enters the picture, one that explains the whole story: slot reservation agreements. When demand exceeds capacity, a customer can no longer simply order a turbine - it has to pay to reserve itself a place in the production queue, years in advance, even before a full order is signed. This quarter, 18 new gigawatts were reserved, and in parallel 10 gigawatts of prior reservations were converted into firm orders. This is the cleanest evidence of real scarcity: when the customer pays for the queue, the bottleneck is not a theory.
The Numbers: What Was Good, and There Was a Lot
The operational picture in the quarter is strong across all the main fronts:
Orders of $24.2 billion, an organic jump of 88%. Of that, the Power division alone recorded $16.7 billion, 2.3x the comparable quarter - including 52 heavy units (among them 15 HA-class turbines, its most advanced) and 61 aeroderivative turbines. The total backlog climbed to $176 billion.
The gas-turbine backlog and reservations rose from 100 to 116 gigawatts. In the power-scarcity piece we cited management's estimate that by year-end no production slot would remain free until 2030. This quarter not only confirmed that - it tightened it: the target was raised to at least 125 gigawatts by the end of 2026, and the expansion plan speaks of 20 gigawatts of annual output as early as the third quarter, 24 in 2028 and 30 in 2030.
Electrification is the story no one told. Orders of $6.3 billion, an organic jump of 66%, and a book-to-bill ratio of about 1.7 - meaning for every dollar sold, $1.7 of orders came in. The division's equipment backlog jumped to $40.6 billion, up 69% in a year. And the figure that connects directly to the AI world: data-center orders passed $5 billion year-to-date - more than double all of 2025 combined.
Profitability expanded, and this is the part that is harder to fake. The adjusted EBITDA margin rose to 11.3%, an organic improvement of 340 basis points. In Power the margin reached 18.8%, and in Electrification 18.4% - an organic jump of 700 basis points. When both volume and price rise at the same time, that is the classic signature of a supplier in a scarcity market.
And cash flow. Free cash flow of $5.1 billion in a single quarter - more than all of 2025 combined. The cash balance stands at $13.1 billion. Management raised its annual cash-flow guidance from $6.5-7.5 billion to $11.5-12.5 billion - nearly a doubling mid-year, a rare move at a company of this size. In parallel, $2.3 billion of shares were repurchased in the quarter, a quarterly dividend of 50 cents was paid, and a voluntary contribution of about half a billion dollars was made to the pension fund.
And Now to the Other Side - Three Points You Must Read
And here comes the part most headlines missed.
First, the wind division is bleeding, and the loss grew. Orders plunged 40% organically, revenue fell 10%, and the quarterly operating loss reached $275 million - more than double the $165 million in the comparable quarter. In the first half the cumulative loss stands at $657 million. The causes: weak demand for onshore turbines in North America, and in parallel higher project costs in offshore wind - the same chronic problem that haunts the whole segment. Management itself expects about $400 million of EBITDA loss in the division this year. In other words: a third of the company's portfolio is in a market that broke, and management is choosing to manage it by shrinking rather than growing in it.
Second - and this is the most important point - the half-year profit is misleading. Anyone who looks at the first half's bottom line will see net income of $5.4 billion and earnings per share of $19.96, and think the company doubled itself. It did not. The figure includes a one-time accounting gain of about $4 billion created when the company acquired the remaining shares in the transformer company Prolec GE and was required to revalue the stake it already held, plus about $330 million of gain on the sale of the Proficy operation. These are paper gains, not cash from operations. The real quarterly net income was $649 million. Anyone who read here about General Motors knows the rule: always check what went into the bottom line before getting excited about it.
Third, the impressive cash flow also requires understanding. The company itself explains that the jump came "mainly from positive working-capital effects" - meaning, to a large degree, from advance payments customers make on those giant orders. This is entirely real cash, and it attests to the strength of demand and to bargaining power - but it is not the same as profit. When the equipment is produced and delivered in the coming years, the working capital will reverse and this cash will be "paid" in work. Cash flow that leans on advances is an excellent sign for the backlog - and not necessarily a pace that repeats forever.
And alongside the three, two further caveats: management noted that inflation offset part of the improvement in Power margins; and the 68% jump in Electrification revenue looks dramatic mainly because of the Prolec consolidation - organic growth, the number that really matters, was 29%. Still excellent, but not 68%.
The Valuation - the Note You Cannot Skip
A correct thesis and an expensive stock can live in the very same security, and this is a point we have repeated in all our infrastructure pieces. The figure that tells the story comes from the report itself: this year the company repurchased 4.3 million shares at an average price of $854 per share. This is a stock trading after a historic rally since the spin-off, and the market already prices in a substantial part of the future backlog. The practical meaning: today's good news is already inside. From here on, what will move the stock is not another good quarter but execution - the ability to convert $176 billion of backlog into actual production, on time, at the promised margins. And that is a far harder industrial task than taking orders.
The debate in one line
The bulls say: this is the company selling the bottleneck itself - a $176 billion backlog, a queue full to 2030, customers paying in advance to hold a spot, and a services annuity that will live for decades. The bears say: a third of the portfolio is losing money, the cash flow leans on advances, the half-year profit is inflated by an accounting gain, and the stock is already priced for perfect execution. Both sides read the same report - and both are right on the facts.
What I Take From This
Three points, in the first person.
The first is that the thesis we wrote about works - but not where I expected. The real surprise this quarter is not the gas turbines, which everyone talks about, but the electrification division: 700 basis points of organic margin improvement, and an equipment backlog that jumped 69%. Transformers and switching are the least glamorous product in the world - and precisely because of them, data centers do not light up. When the most boring layer in the chain shows the sharpest improvement, it means the scarcity is real all along the line, not just in the headline.
The second is about reading reports. The gap between $5.4 billion of "profit" and $649 million of real quarterly profit is exactly where the amateur investor and the professional investor part ways. No one tried to hide anything - it is all written in the report in full transparency. You just have to read.
And the third: an excellent company is not automatically an excellent investment - it all depends on the price and the execution. GE Vernova sits today on the right side of one of the largest demand-supply gaps in the global economy. The question that remains open is not whether the demand is there; it is how much of it is already paid for in the share price, and whether the plants will keep the pace management promised. The first will be decided by the market. The second we will know only in the coming quarters - and that is exactly where it pays to look.
Sources: GE Vernova's official Q2 2026 results as filed with the SEC (Form 8-K, including the results exhibit and Non-GAAP reconciliations), the company's announcements on the spin-off from GE, and installed-base data, current as of the time of writing. The chart is shown in real time via TradingView.