SoftBank Books $7.1 Billion Quarterly Loss on Failed Turbine Grab

SoftBank has reported a $7.1 billion quarterly loss, with more than half of it tied to power-generating equipment, after its attempt to corner the natural gas turbine market for AI data centres fell short.

The numbers behind the bet are striking. Starting roughly two years ago, the group placed deposits on more than 13 gigawatts of turbines — almost 50% more than it actually needed, according to reports. CEO Masayoshi Son's thesis was straightforward: natural gas was emerging as a key power source for the AI boom, turbine order queues had stretched into the early 2030s, and whoever controlled supply would force AI developers to come to them.

That thesis has now collapsed. The company considers much of its nonrefundable deposits worthless, which is what drove the bulk of the quarterly charge.

Why did SoftBank overbook turbines in the first place?

Industry watchers say the strategy wasn't irrational, just expensive to test. Peter Gardett, CEO of Noreva, said SoftBank's SB Energy had little choice but to overbook if it wanted turbines quickly, describing the charge as costly price discovery finally taking place.

In other words, Son paid a premium to learn where the market's real constraints sit. The tuition came to billions.

What does this change for AI data centre builders?

For practitioners, the lesson lands on power strategy. Brent Nelson of Ascend Analytics said natural gas carries risks such as winter curtailments, and that most data centres would be better served by a smaller grid connection supplemented with large batteries to manage peak loads.

That's a meaningful reframe for anyone planning capacity right now. Instead of chasing scarce gas turbines and betting on fuel availability through peak-demand seasons, Nelson's argument points toward a hybrid model: lean on the grid for baseline power, and use storage to absorb the spikes that gas plants were meant to cover.

It also suggests the scarcity premium SoftBank tried to arbitrage may matter less than expected. If batteries plus grid connections can do the job, the urgency to lock up turbine slots weakens — and with it, the leverage Son was trying to build.

Does the loss change the gas build-out?

Not much, at least for now. Despite SoftBank's losses, AI's power demand remains too large to trigger widespread cancellations of gas plants, per the reporting. The demand side of the equation hasn't moved; only one aggressive buyer's positioning has.

For gas developers and turbine makers, that means the order books stay intact even as one of the market's most aggressive depositors writes down its position. For SoftBank, the question is whether the remaining turbine commitments still fit a strategy that just cost $7.1 billion to stress-test — or whether the group pivots toward the battery-and-grid approach its analysts now recommend.