GE Vernova Selloff Puts AI Infrastructure Financing Under Scrutiny

GE Vernova lost 10.6% on October 22 even though the company released no material operating update that day. The trigger was broader: Meta's results revived investor concern about how hyperscalers will finance the enormous power and data-center buildout behind artificial intelligence.

Barron's linked the selloff to those financing fears, while other power suppliers also declined. That does not prove Meta caused every move, but it explains why GE Vernova's order book was suddenly viewed through a credit lens. Strong equipment demand matters only if customers can fund projects and convert reservations into firm orders.

GE Vernova entered the debate with substantial operating evidence on its side. In the second quarter of 2026, orders rose 88% organically to $24.2 billion and revenue increased 22% to $11.1 billion. Cash from operating activities reached $5.5 billion, and free cash flow was $5.1 billion. The company ended the quarter with a $176 billion backlog and $13.1 billion of cash.

Gas Power equipment backlog and slot reservation agreements rose from 100 gigawatts to 116 gigawatts during the quarter. Management expected at least 125 gigawatts under contract by year-end. Data-center orders in the Electrification segment exceeded $5 billion in the first half, more than double the total for all of 2025.

Those figures show real demand, but they do not remove the financing risk. A reservation is not the same as delivered equipment or recognized revenue. Large power projects require permits, transmission access, turbines, construction capacity and customers capable of paying across a multiyear schedule. If hyperscalers slow capital spending or rely on more expensive financing, suppliers can face delayed orders even when long-term electricity demand remains strong.

The balance sheet gives GE Vernova room to absorb timing shifts. The company generated more free cash flow in the second quarter than it produced in all of 2025, while returning $3.9 billion to shareholders in the first half. Management is also expanding annual gas-turbine output from 20 gigawatts in 2026 toward 24 gigawatts in 2028 and 30 gigawatts in 2030. That investment will pay off only if reservations mature into profitable deliveries without creating excess capacity.

The selloff therefore exposed a change in the market's question. Investors are no longer asking only whether AI needs more power. They are asking who will finance that power, how binding customer commitments are and whether suppliers are being paid early enough to protect cash flow.

GE Vernova's next report on October 28 should provide better evidence. Orders, the conversion of slot reservations, customer deposits and the timing of data-center projects will matter more than another long-range valuation target. If the backlog continues to expand while cash stays ahead of earnings, the October decline will look more like a financing scare than a deterioration in the business. If commitments weaken, the market will have its first company-specific evidence that AI infrastructure funding is becoming a constraint.