TOKYO, November 10, 2026 — Japanese technology investor SoftBank Group reported a quarterly net loss of 1.1 trillion yen, or about $7 billion, after a costly change to equipment purchases for U.S. artificial intelligence data-center projects and a sharp rise in financing expenses.
The loss for the July-to-September quarter compared with a 2.5 trillion yen profit in the same period a year earlier, when a rise in the value of SoftBank’s OpenAI investment produced a large accounting gain. The result was also far below analyst expectations. SoftBank disclosed the figures Tuesday as it continued to expand its investments in AI infrastructure, robotics and software.
A central factor was a $3.7 billion charge linked to advance payments made by a subsidiary for power-generation and grid equipment intended for U.S. data-center projects. SoftBank said project plans changed after the equipment had been ordered. The company may receive rebates of up to $2.1 billion on future purchases, but those rebates are conditional and should not be treated as guaranteed recovery of the full amount.
Data-center plans changed after equipment commitments
The charge highlights the difficulty of building large data centers in a market where demand for electricity, power equipment and grid connections is rising quickly. Developers are competing for turbines, transformers, electrical systems and other components, while suppliers face long lead times. Companies may place orders well in advance to secure capacity, but changes in project design or timing can leave them with equipment that no longer fits the revised plan.
SoftBank executives said intense competition for power supplies had encouraged early purchases before some project specifications were settled. The company’s U.S. data-center ambitions are part of founder Masayoshi Son’s broader strategy to make AI a central pillar of SoftBank’s business. The group has committed large sums to computing infrastructure and has invested heavily in OpenAI.
SoftBank has spent about $64.6 billion for a 13% stake in OpenAI, according to Reuters. That exposure means the group’s financial performance is increasingly tied to the valuation and commercial prospects of one of the most prominent AI developers. A public listing by OpenAI could eventually provide SoftBank with a route to realize some of the value of its holding, but the timing and terms of any offering remain uncertain.
Borrowing costs add to investor scrutiny
SoftBank’s financing costs more than doubled from a year earlier to 462 billion yen in the quarter, Reuters reported. The company raised a record $56.5 billion during the period to fund its investment program. Its borrowing includes large loans and bond issues, and it has expanded a margin loan secured by shares in Arm, the British chip designer in which SoftBank holds a major stake.
Higher financing expenses matter because many AI infrastructure projects require substantial spending before they generate revenue. Data centers need land, power connections, cooling systems and specialized hardware. Returns depend on whether customers sign long-term contracts and whether computing capacity can be used efficiently. If construction is delayed or demand develops more slowly than expected, companies can face higher carrying costs and lower returns.
The quarter’s loss also reflects the volatility of SoftBank’s investment portfolio. Gains or losses in listed holdings and private-company valuations can make results swing sharply from one reporting period to another. A single quarter therefore does not show whether the group’s AI strategy will ultimately succeed, but it does reveal the scale of capital and execution risks involved.
What it means for U.S. readers
SoftBank’s spending is closely linked to the U.S. data-center boom, which is increasing demand for electricity and specialized power equipment. Its procurement changes show that the challenge is not simply finding customers for AI computing; developers must also secure reliable electricity, grid access and equipment that can be installed on schedule.
For U.S. utilities, technology companies and communities hosting data centers, the results underscore the financial consequences of rushed infrastructure commitments. Equipment shortages and shifting project requirements can affect costs well beyond the company placing the order. The episode may also influence how investors assess AI companies that rely on borrowed money and long-term infrastructure spending.
SoftBank has not abandoned its AI strategy. The company remains committed to major investments, while the potential rebates could reduce part of the equipment-related charge. But the latest results show that rapid expansion carries substantial financial risk, particularly when plans change after large commitments have been made.
Sources and reporting credit
This original report was checked against Reuters, November 10, 2026, The Financial Times’ report on SoftBank’s data-center equipment charge, and SoftBank Group’s official second-quarter FY2026 earnings materials.





