GE Vernova's 20% Pullback: De-Rating Inside the Power Supercycle

Two months ago GE Vernova was the trade everyone seemed to own. The stock that became shorthand for the AI-power boom peaked at $1,195.94 on July 6, and by the September 11 close it had fallen to $957.27 — roughly a fifth off the high. Jefferies, one of its most persistent bulls, calls that pullback an entry point, raising its price target to $1,185 from $1,155 while keeping a Buy rating.

The odd part: nothing about the underlying business got worse. The stock fell because the quarterly report disappointed on profit, not on demand.

A cost miss, not a demand miss

GE Vernova's second-quarter adjusted profit came in at $2.47 a share, short of the $3.04 Wall Street had wanted — an 18% gap — even as revenue of $11.1 billion beat expectations and grew 22%. That is the kind of print that bruises a stock priced for near-perfect execution. The drag was identifiable rather than existential: its wind business bled a $275 million EBITDA loss for the quarter and the company now expects roughly $400 million of wind losses for the full year, while global tariffs add an estimated $100–200 million of cost in 2026.

The demand engine, though, accelerated in the same quarter. Orders jumped 88% to $24.2 billion with a book-to-bill ratio above two — meaning it took in more than twice as much new business as it shipped — and backlog hit a record $176 billion, up $13 billion in a single three-month stretch. Management raised full-year revenue guidance for a second straight quarter, to $45.5–46.5 billion, and nearly doubled free-cash-flow guidance to $11.5–12.5 billion. CEO Scott Strazik described the company as being in "the early stages of this electricity investment super cycle."

The slot math is the story

Seen from the macro level, the number that matters most is how far ahead the order book already reaches. GE Vernova builds gas turbines against a queue: it expects to end 2026 with 125 gigawatts of turbines under contract, and more than half of its 2031 production slots already booked. Data-center orders in the first half topped $5 billion, more than doubling the whole of 2025.

That is the exponential curve bending. Power is the choke point of the AI buildout — you cannot run the compute without the electrons — and GEV's backlog converts today's demand into years of visible revenue before the turbines are even built. A miss in one quarter does not unwind a queue that extends six years out.

The price paid is the risk

Here is the honest friction. A forward multiple above 30 leaves no room for a bad quarter, which is exactly what happened. And the near-term earnings path is not the tidy growth that multiple seems to imply: the street's own estimates actually show per-share earnings declining in 2027, in part the flip side of the buybacks and step-up in cash flow that flattered 2026, plus the wind drag.

So the entry-point debate is really a question about price paid versus what the backlog buys you. Jefferies' bet is that the queue of future revenue de-risks the multi-year story enough that a 20% de-rating overshoots. The order data support the demand side of that argument; they do not erase the valuation risk on the way down.

The optimistic read holds if orders keep running ahead of production capacity, if the backlog converts to profit at the margins management guides, and if data-center capital spending does not stumble. It would break if wind losses widen, if the 2027 earnings rollover proves the start of a slower phase rather than an accounting quirk, or if the market simply refuses to keep paying a premium for the stock.

A 20% drawdown in a consensus favorite is often a sentiment event more than a demand event — the moment when the evidence and the mood part company. Record orders and sold-out slots say the business is still compounding; the multiple says the stock was priced as if it could not wobble. Whether this is an entry point depends on which of those you trust more.