SoftBank Group reported a net loss of approximately 1.12 trillion yen ($7 billion) for the fiscal second quarter ended Sept. 30, 2026, as a $3.7 billion charge related to US artificial intelligence data center equipment and rising financing expenses weighed on the Japanese investment group’s results.
The loss disclosed reversed a profit of approximately 2.5 trillion yen in the comparable quarter a year earlier. It highlights the financial demands of SoftBank’s expansion into AI infrastructure, where large capital commitments and long development schedules are creating additional exposure for one of the world’s most prominent technology investors.
SoftBank’s fiscal 2026 second-quarter earnings presentation detailed the equipment-related loss, which followed changes to procurement plans for power-generation assets supporting US data center projects.
US Data Center Equipment Triggers $3.7 Billion Charge
The equipment charge arose after a SoftBank-backed subsidiary revised procurement arrangements for power-generation infrastructure intended to support AI computing facilities.
SoftBank had sought to secure equipment in advance amid competition for electricity-generation capacity. Subsequent changes to project requirements and procurement plans resulted in losses on nonrefundable advance payments and contract modifications.
The company recorded approximately $2.9 billion in losses associated with advance payments and a further $880 million related to contract adjustments and settlements.
SoftBank also disclosed potential rebates of up to $2.1 billion on qualifying future equipment purchases. Those rebates are conditional and should not be treated as an immediate reversal of the reported charge.
The distinction is important for infrastructure investors because procurement commitments can create substantial financial exposure before facilities become operational or begin generating revenue.
On Nov. 12, SoftBank provided additional information about its remaining equipment-related advance payments.
According to its updated US AI data center equipment disclosure, the subsidiary had made $9.5 billion in advance payments, with a remaining balance of $6.6 billion after the previously recognized loss.
SoftBank identified $5.6 billion associated with recovery through the PORTS-Pike power-generation project, including $3.4 billion expected in the near term and $2.2 billion by the end of 2027. Another $1 billion was expected to be recovered over time.
These figures provide investors with a more detailed view of the capital still tied to equipment procurement, although anticipated recoveries remain subject to execution and timing risks.
Financing Expenses Rise as SoftBank Expands Borrowing
SoftBank’s financing costs more than doubled year over year to approximately 462 billion yen during the July-September quarter.
The increase coincided with an unusually large capital-raising effort. The group raised approximately $56.5 billion during the quarter, its highest quarterly total, to support investments across artificial intelligence, infrastructure and related technologies.
Funding arrangements included borrowing facilities, bond issuance and financing backed by investment holdings.
The rising cost of capital adds another consideration for investors assessing SoftBank’s strategy.
Large infrastructure developments typically require substantial upfront spending, while the cash flows supporting those investments may not materialize until construction, commissioning and customer deployment are completed.
Higher interest expenses can therefore reduce financial flexibility, particularly when investment valuations fluctuate or asset monetization takes longer than expected.
For lenders and institutional investors, SoftBank’s results illustrate the importance of evaluating financing structures alongside the projected economics of AI infrastructure assets.
OpenAI Investment Concentrates Technology Exposure
SoftBank’s investment in OpenAI remains central to its AI strategy.
The group has invested approximately $64.6 billion for a 13% ownership stake in the ChatGPT developer, according to its November 2026 disclosures.
OpenAI represented approximately 21% of the equity value of SoftBank’s investment holdings at the end of September.
That concentration makes changes in OpenAI’s valuation particularly relevant to SoftBank’s reported investment performance and net asset value.
SoftBank’s earnings can fluctuate substantially because investment gains and losses reflect changes in the estimated fair value of portfolio holdings, including privately held businesses.
Such valuation movements do not necessarily correspond to cash proceeds from asset sales.
The distinction matters to private-market investors comparing reported investment gains with realised distributions, liquidity and capital requirements.
SoftBank’s exposure also extends beyond generative AI software to semiconductor technology, robotics, autonomous systems and the infrastructure required to operate large computing facilities.
Vision Funds Record Gains Despite Group Loss
SoftBank’s Vision Fund operations provided a partial counterweight to the group’s broader quarterly losses.
For the six months ended September, SoftBank Vision Fund 1 reported approximately $2.4 billion in investment gains, supported partly by its exposure to ByteDance.
Vision Fund 2 recorded investment gains of approximately $60 million over the same period.
These figures relate to the six-month reporting period and should not be directly compared with SoftBank’s consolidated quarterly net loss.
The positive investment performance demonstrates that the group’s portfolio results are not uniform across strategies and holdings.
However, unrealised gains in venture portfolios remain sensitive to valuation assumptions, financing conditions and the availability of exit opportunities.
For limited partners evaluating venture capital performance, distributions and realised returns remain important measures alongside reported portfolio appreciation.
AI Infrastructure Investment Faces Greater Financial Scrutiny
SoftBank’s results draw attention to a financing challenge affecting the wider AI infrastructure market.
Developing data centers at scale requires more than acquiring computing hardware. Investors must also finance electricity generation, grid connections, cooling systems, land, construction and supporting infrastructure.
Power availability has become a particularly important consideration as AI developers seek large amounts of computing capacity.
Securing equipment early may reduce exposure to supply shortages, but it can also increase financial risk when construction schedules, technical requirements or procurement strategies change.
The latest equipment charge illustrates how those risks can affect investment groups before the underlying facilities generate operating income.
SoftBank’s quarterly loss does not, by itself, establish that demand for AI computing infrastructure is weakening. It does, however, demonstrate the financial consequences of committing capital ahead of project completion.
For private equity firms, infrastructure managers and institutional lenders, the relevant considerations include counterparty strength, contracted revenue, equipment procurement terms, financing costs and the timing of expected cash flows.
SoftBank’s ability to manage these exposures will be an important factor in assessing the financial sustainability of its expanding AI investment portfolio.


