SoftBank placed very large orders for natural-gas turbines to secure capacity for rising AI-driven data-center power needs. Its subsidiary SB Energy announced in February that it would spend $33 billion on a 9.2-gigawatt natural gas power plant, and SoftBank itself made a $30 billion investment in OpenAI in early 2025.
Procurement strategy and consequences
According to reports, starting about two years ago SoftBank paid nonrefundable deposits on more than 13 gigawatts of turbines—roughly 50% more than it actually required. The company’s logic was to lock in equipment amid long lead times and long waiting lists for turbines, in order to accelerate its projects.
Peter Gardett, CEO of Noreva, told TechCrunch that "SB Energy did not have a choice except to overbook turbines in the future if they wanted some turbines right away." Gardett also described the outcome as a costly episode of price discovery: SoftBank overpaid and over-ordered, then had to write down part of those commitments.
Financial impact
SoftBank reported a $7.1 billion quarterly loss on Tuesday, with over half of that loss related to power-generating equipment. More than $3 billion of the writedowns are attributed to the misjudged turbine commitments and related impairments.
Why the market moved this way
The energy sector was taken aback by the scale and speed of AI’s rise in power demand: turbine order backlogs now stretch into the early 2030s. Because grid connections can take years to secure, many hyperscalers and developers treated natural gas as a reliable, deployable energy source to meet near-term needs.
But operators are learning more about actual operational requirements, and there are risks to relying solely on gas—among them seasonal curtailments when supply tightens and demand spikes, as Ascend Analytics senior managing director of markets and strategy Brent Nelson noted. Nelson suggested that a grid connection is often the best foundation for most data centers, albeit possibly at a smaller capacity than initially requested, supplemented by large batteries to shave peaks or ride out curtailment requests from utilities.
If more operators reach the same conclusion, demand for new gas capacity could grow more slowly than SoftBank expected.
Implications and outlook
SoftBank’s gas-related losses are not expected to trigger mass cancellations of gas power plants—the scale of AI’s electricity needs still supports significant gas demand. However, the company’s multibillion-dollar writedown illustrates the risk of large, front-loaded bets in a market where delivery times, grid constraints and evolving energy strategies can change the economics quickly.
In short: oversized turbine orders helped SoftBank secure equipment quickly, but the strategy also produced expensive price discovery that materially contributed to the company’s quarterly loss.
