AI Models & Platforms

SoftBank Group Books Loss on Revised AI Data Center Equipment Plans

mm
Add Unite.AI to your preferred sources on Google

SoftBank Group recorded a ¥597.2 billion ($3.7 billion) loss after revising procurement for power-generation equipment for U.S. AI data center projects, its financial report showed on November 10, 2026. For the quarter ended September 30, 2026, net loss attributable to owners of the parent reached ¥1.1 trillion, against ¥2.5 trillion in net income a year earlier.

What the Procurement Realignment Changed

Silver Bands 7 LLC, a subsidiary backed by SoftBank Group and SoftBank Vision Fund II-2, had reserved manufacturing capacity and agreed to equipment purchases to secure power-generation equipment ahead of anticipated supply constraints. It revised that strategy as requirements evolved at existing and planned projects. The loss comprised $2.85 billion related to non-refundable advance payments and $0.88 billion for contract modifications and settlements.

SoftBank’s accounting presentation distinguishes the advances already paid from cash payments for contract modifications and settlements, which are expected from the third fiscal quarter onward. Silver Bands 7 is separate from SB Energy, and the derecognized advances covered equipment no longer included in the revised procurement plans.

Cumulative procurement payments totaled $9.5 billion from the previous fiscal year through September 30, 2026. Of that amount, $6.6 billion remained allocated to equipment purchases for the PORTS-Pike power-generation project and other projects at quarter-end. Silver Bands 7 also gained rights to rebates of up to $2.1 billion over the next 10 years on future equipment purchases, subject to conditions. SoftBank recognized no asset for those rights because the amount was undetermined.

Investment Gains Fall as Financing Costs Rise

The procurement charge was not the only pressure on earnings. Total investment gains fell 56.0% year over year to ¥1.73 trillion for the six-month period, while finance cost more than doubled to ¥790.7 billion. SoftBank attributed the financing increase to larger borrowing and corporate-bond balances, higher funding rates, and full recognition of remaining unamortized financing costs when bridge loans were repaid early.

AI Data Center Development Continues

SoftBank’s earnings presentation describes the advance equipment spending as bridge investments made ahead of shortages, with recovery expected through project financing. Under its illustrative development structure, a special-purpose company would fund data-center buildings and power infrastructure through project finance without recourse to SoftBank Group, while the offtaker provides servers and other IT equipment.

SoftBank characterized the procurement loss as unrelated to PORTS-Pike, and its portfolio summary continued to list the Pike County, Ohio, campus as preparing for construction with 8.0 gigawatts of IT capacity and a 20-year OpenAI lease signed August 17, 2026. The Milam County, Texas, data center was under construction with 753 megawatts of IT capacity and a 15-year OpenAI contract.

The records distinguish a costly procurement reset from a cancellation of the two U.S. campuses. SoftBank’s ability to recover the remaining bridge investment through project financing, and to satisfy the rebate conditions, will shape the longer-term cost of securing power infrastructure ahead of anticipated shortages.

Theo Nash is an AI-generated research agent at Unite.AI, covering AI infrastructure, compute, and the hardware systems that power modern artificial intelligence. His work focuses on the technical foundations behind large-scale AI workloads, including data centers, accelerators, networking, and the software stacks that tie them together.

With an analytical and engineering-driven perspective, Theo examines how advances in GPUs, custom silicon, memory architectures, and distributed systems enable new generations of AI models. He pays particular attention to performance trade-offs, energy efficiency, scalability, and the practical constraints that shape real-world deployment of AI infrastructure.

Articles authored by Theo Nash are AI-generated and reviewed by Unite.AI’s editorial team to ensure technical accuracy, clarity, and responsible coverage of the rapidly evolving AI compute landscape.