SoftBank Launches SB Neo to Rent AI Computing Capacity in the US

SoftBank Group and its telecom unit SoftBank Corp. announced plans to rent AI computing resources to US companies through a new neocloud venture called SB Neo. The move signals a major push into the US AI cloud market, targeting enterprises and hyperscalers with dedicated AI computing capacity.

What happened

SoftBank Group and SoftBank Corp. said they will start renting AI computing resources to US companies starting next fiscal year, which begins in April 2027. The capacity will come through SB Neo, Inc., a newly formed neocloud venture. SoftBank plans to scale its US data center capacity to 10 gigawatts by around 2030 to meet growing demand for AI computing power.

The company also plans to build gigawatt-scale AI data centers in Japan. The US initiative is backed by SoftBank's broader infrastructure plans, including a $100 billion investment commitment and its recent $6.5 billion acquisition of Ampere Computing, a chip design firm.

Why AI builders should care

For AI builders, founders, and product teams, SoftBank's entry into the US AI cloud market could mean more compute supply and potentially different pricing or leasing models. The company is targeting both US enterprises and hyperscalers, which could increase competition with existing providers like CoreWeave and the major cloud platforms.

SoftBank's scale ambitions are notable. A 10-gigawatt target by 2030 would make SB Neo one of the largest dedicated AI compute providers in the US, if realized. For teams building AI products that require large-scale training or inference capacity, having another well-capitalized supplier could improve availability and pricing over time.

Practical implications

If SB Neo proceeds as planned, US enterprises may gain another source of AI compute capacity at significant scale. The venture is positioned to supply AI chips and cloud services to large firms, including hyperscalers. SoftBank's existing relationships with Arm (which it owns) and Ampere (which it is acquiring) could give it unique leverage in chip supply and data center design.

The company is also building AI data center capacity globally. SoftBank recently announced plans to invest 45 billion euros ($53 billion) over five years to build AI infrastructure in France, targeting up to 3.1 gigawatts of capacity. This global buildout could eventually support US customers with multi-region deployment options.

Caveats

These plans are based on briefings and secondary reports. Exact timelines, capacity targets, and partnership details may evolve as coverage continues. The 10-gigawatt target is an ambition for around 2030, not a near-term commitment. Pricing models, specific chip types, and service-level agreements have not been disclosed. AI builders should watch for concrete service announcements and pricing details before making procurement decisions.

FAQs

What is SB Neo and how will SoftBank rent AI computing capacity in the US?

SB Neo is SoftBank's newly formed neocloud venture intended to supply AI computing resources in the US market. It will rent AI computing capacity to US enterprises and hyperscalers, starting next fiscal year. The venture is a joint effort between SoftBank Group and its telecom unit SoftBank Corp.

When will SoftBank start offering AI cloud services in the United States?

SoftBank plans to start renting AI computing resources to US companies next fiscal year, which begins in April 2027. The company will establish SB Neo in July 2026 to begin operations. Timelines may be subject to change as the venture develops.

How many gigawatts of AI data center capacity does SoftBank plan to build?

SoftBank plans to scale its US data center capacity to 10 gigawatts by around 2030 through SB Neo. The company also plans to build gigawatt-scale AI data centers in Japan and has announced a 45 billion euro investment in French AI infrastructure targeting up to 3.1 gigawatts.

How does SoftBank's AI cloud initiative compare to other AI cloud providers?

SoftBank's SB Neo targets a similar market as existing neocloud providers like CoreWeave, but with the backing of a major global conglomerate. The company's 10-gigawatt capacity target by 2030 would make it one of the largest dedicated AI compute providers in the US. SoftBank also brings unique assets including ownership of Arm and its recent acquisition of chip designer Ampere Computing, which could give it advantages in chip supply and data center design.