SoftBank Creates SB Neo for US AI Cloud Market

Japanese SoftBank Group and its telecom operator subsidiary, SoftBank, are creating a new company that will operate a neocloud business in the US. The two companies are seeking to capitalise on the strong demand for AI computing resources among major enterprises and hyperscalers.

The new entity, SB Neo Inc., will be established in Delaware in July 2026 with SoftBank holding a 51% stake and SoftBank Group owning the remaining 49%. It will function as a consolidated subsidiary of SoftBank

SB Neo will use the SoftBank Group's 10GW energy and AI infrastructure currently under development to deliver computing resources for large-scale AI model training and inference. The company plans to launch neocloud services commercially in the 2027 fiscal year ending March 31, 2028.

The new subsidiary will expand its service capacity in phases to meet strong demand in the US market, with plans to eventually deploy AI infrastructure at a 10GW scale.

“The SoftBank Group will work together to deploy world-class AI infrastructure and drive the AI revolution,” says Masayoshi Son, Chairman & CEO of SoftBank Group.

Building on Japanese GPU cloud experience

SoftBank has been operating a beta version of its graphics processing unit (GPU) cloud service in Japan since May 2026, based on Infrinia AI Cloud OS, a software stack designed specifically for AI data centres. The company will apply the expertise and insights gained through this Japanese initiative to its US operations.

SoftBank Group is one of OpenAI’s largest shareholders. The group announced back in February 2026 that its cumulative investment in the AI frontier lab will rise to US$64.6bn. The two companies have also invested in SB Energy, a SoftBank Group company which is working to build and operate AI data centres under the Stargate Project in the US.

Group resources to support US expansion

SoftBank Group Corp. is already advancing large-scale AI infrastructure projects in the US and will support SB Neo's business by providing access to its infrastructure, personnel and other resources.

Junichi Miyakawa, President and CEO of SoftBank says: “With the strong demand for AI data centres in the US and the SoftBank Group making steady progress toward securing 10GW of power, we are partnering with SoftBank Group Corp. to develop our neocloud business in the US.

“We also plan to proceed with the construction of GW-scale AI data centres in Japan as soon as preparations are in place.”

Neocloud refers to cloud providers that focus on offering GPU-as-a-Service (GPUaaS). They supply frontier AI labs, enterprises and hyperscalers with the vast compute required to train large AI workloads and run inference operations. GPUs (graphical processing units) are the workhorses of AI, due to their ability to handle the parallel processing requirements of machine learning workloads. 

One step in a wider campaign

SoftBank is active in the AI infrastructure market across many fronts.

It announced plans to acquire US-based DigitalBridge, a digital infrastructure company focused on data centres, fibre networks, cell towers and edge facilities, for US$4bn in January 2026. DigitalBridge’s shareholders approved the transaction in April 2026 and the company expects the deal to close in the second half of 2026, subject to regulatory approvals. 

Back in June 2026, SoftBank committed to €75bn (US$87bn) for 5GW of AI data centre capacity in France, with delivery at the first sites targeted for 2031.

Closer to home, SoftBank is building up its ability to support AI data centres. In Japan, it has launched a battery business to develop energy infrastructure to support electricity demand from AI workloads. 

In addition, at a site in Sakai City, Osaka Prefecture, it plans to build two facilities: the AX Factory and the GX Factory. The former will focus on AI data centre operations and the manufacture of AI infrastructure hardware. The GX Factory will produce next-generation batteries, solar panels and related technologies.