SoftBank launches SB Neo to provide AI cloud services and computing power across America

SoftBank Enters the Neocloud Business with SB Neo

SoftBank is launching a major push into America's AI infrastructure market by establishing SB Neo, a new company dedicated to providing AI cloud services to hyperscalers and enterprises seeking powerful AI computing resources. The Japanese technology conglomerate announced that SB Neo will begin operations in fiscal 2027, which ends March 31, 2028, marking what SoftBank Corp. head Junichi Miyakawa calls "a second founding for our company" 3

3

. The ownership structure splits control between SoftBank Corp., holding 51 percent, and SoftBank Group Corp., retaining 49 percent, with SB Neo operating as a consolidated subsidiary 1

1

.

The neocloud business represents a specialized segment of cloud computing infrastructure focused on renting access to graphics processing units and AI accelerators. These platforms emerged to address the massive demand for compute power needed for AI model training and inference. SB Neo enters a competitive landscape already populated by dedicated providers like CoreWeave and Nebius Group, as well as established giants including AWS, Microsoft Azure, and Google Cloud 3

3

.

Ambitious Infrastructure Plans and Revenue Targets

The venture plans to scale rapidly, targeting 10 gigawatts of AI datacenter capacity by around 2030 to support large-scale AI training infrastructure needs 3

3

. This aggressive expansion could transform SoftBank Corp.'s financial profile dramatically. According to people familiar with the company's plans, supplying neocloud operations in the US could triple or quadruple the mobile carrier operator's annual operating income to between $18.5 billion and $25 billion 2

2

. Miyakawa emphasized that the new business has potential to generate profit "on a different order of magnitude" 3

3

.

SB Neo will draw on SoftBank Group's existing 10-gigawatt-scale energy and AI infrastructure already under development in the US, including a $500 billion datacenter-focused project in Ohio 3

3

. The company also disclosed plans to construct gigawatt-scale AI datacenters in Japan once preparations are complete 1

1

. At home, SoftBank's telecom unit is already building datacenter campuses in Hokkaido and Sakai in Osaka Prefecture 3

3

.

Leveraging Infrinia AI Cloud OS and Japanese Experience

SoftBank Corp. has been operating a beta version of its GPU cloud service in Japan since May, powered by Infrinia AI Cloud OS, a proprietary software stack developed by the company's Infrinia team 1

1

. This platform supports Kubernetes-as-a-Service in multi-tenant environments and Inference-as-a-Service to provide large language model inference capabilities via APIs 1

1

. The company intends to apply the expertise and insights gained from this Japanese operation to drive its US rent-a-GPU business 1

1

.

SoftBank chief Masayoshi Son declared that "The SoftBank Group will work together to deploy world-class AI infrastructure and drive the AI revolution" 1

1

. His commitment to AI remains unwavering despite market concerns, recently telling shareholders that calling AI a bubble is "an insult to AI" and "blasphemy" 1

1

.

OpenAI Connection and Market Challenges

SB Neo may have a ready-made customer in OpenAI, in which SoftBank has committed a staggering $65 billion investment by October 2

2

. This massive stake likely ensures significant influence over the ChatGPT maker's infrastructure decisions. To finance its ambitious plans, SoftBank has reengaged with lenders to secure a $10 billion loan backed by its OpenAI stake, now offering to guarantee repayment to address lender concerns about collateral value 2

2

.

However, questions linger about the financial viability of the neocloud business model. A McKinsey & Company report warned that neoclouds face fragile economics due to their inherently commoditized nature, with limited differentiation possible when renting specific hardware 1

1

. Miyakawa countered that SoftBank has a competitive edge through its ability to secure power sources, mainly from gas-fired plants 3

3

. As hyperscalers and AI companies continue seeking reliable AI training infrastructure, SB Neo's success will depend on whether energy access and scale can overcome the commoditization challenges that have troubled other neocloud operators.