SoftBank Books $3.5 Billion Loss on Failed AI Energy Cornering Strategy

SoftBank Group has recorded a $7.1 billion quarterly loss, driven largely by a failed multi-billion dollar wager on the energy infrastructure required to power the artificial intelligence boom. According to TechCrunch, more than half of that loss stems from nonrefundable deposits on power-generating equipment that the firm now considers worthless. The write-down marks a significant setback for CEO Masayoshi Son, who attempted to use SoftBank's massive capital reserves to corner the market for natural gas turbines, betting that AI developers would be forced to buy power directly from his subsidiary, SB Energy.

The strategy involved a massive procurement play where SoftBank placed deposits on more than 13 gigawatts of turbines. This volume represented nearly 50% more capacity than the firm actually required for its own planned 9.2-gigawatt natural gas power plant. By over-ordering, SoftBank hoped to secure immediate delivery in an environment where manufacturing queues for turbines now stretch into the early 2030s. Peter Gardett, CEO of Noreva, told TechCrunch that SB Energy had little choice but to overbook if they wanted to secure any turbines in the immediate term, describing the loss as a form of expensive price discovery.

This aggressive capital deployment followed SoftBank's $30 billion investment in OpenAI in early 2025, signaling Son's intent to own both the models and the underlying energy supply. However, the $3.5 billion loss on these deposits suggests that the market for private natural gas power plants is developing more slowly than SoftBank anticipated. While natural gas was initially viewed as a security blanket for hyperscalers facing long grid connection delays, the financial reality of these projects is proving difficult to manage even for the world's most aggressive venture investors.

Industry analysts suggest that the AI sector's energy strategy may be shifting. Brent Nelson of Ascend Analytics noted on the Equity podcast that data center operators are increasingly finding that smaller grid connections supplemented by large-scale batteries are more cost-effective than independent gas plants. These battery systems allow operators to shave peak loads and manage utility curtailment requests without the risks associated with natural gas, such as supply dips during winter spikes. This shift in preference likely contributed to SoftBank's decision to walk away from its excess turbine orders.

While SoftBank's loss is unlikely to trigger a broader collapse in gas power plant construction due to the sheer scale of AI power demand, it establishes a clear ceiling for speculative energy plays in the venture space. The write-down serves as a stark reminder of the risks inherent in vendor-financed growth and infrastructure cornering. For now, the US venture market is watching to see if other major AI players will follow SoftBank's retreat from independent power generation in favor of more integrated grid solutions.