AI dominates SoftBank's portfolio as OpenAI gains drive profits
AI-related holdings accounted for about 18% of the SoftBank group’s net asset value roughly three years ago, but now represent more than 60%.
SoftBank has hitched itself to OpenAI's trajectory, transforming the ChatGPT maker's soaring valuation into substantial paper wealth while preparing to deepen its commitment by another $30 billion.
SoftBank Group Corp. is intensifying its push into artificial intelligence, committing tens of billions of dollars across chips, infrastructure, robotics and foundational AI models, as the Japanese investment conglomerate positions itself for what it describes as the next phase of technological evolution.
AI-related holdings accounted for about 18% of the group’s net asset value roughly three years ago, but now represent more than 60%.
SoftBank’s most visible bet is its deepening relationship with OpenAI. The group has invested $34.6 billion into the AI developer, giving it an ownership stake of roughly 11% following a $22.5 billion follow-on investment completed in December 2025.
The investment has already produced large unrealised gains as OpenAI’s valuation climbed from about $150 billion in late 2024 to approximately $500 billion by October 2025. Now SoftBank is reportedly gearing up to invest another $30 billion in the ChatGPT maker, at a $850 billion valuation.
The group recorded a roughly $14.6 billion gain in OpenAI investment fair value last quarter, and a $4.2 billion gain in fair value in the latest.
SoftBank’s AI push extends beyond large-language models into hardware, robotics and digital infrastructure, reflecting a vertically integrated investment philosophy.
The company completed a $6.5 billion acquisition of chip designer Ampere Computing, while also announcing plans to acquire ABB’s robotics business for about $5.4 billion and digital infrastructure investor DigitalBridge for $3.1 billion.
Executives characterised these investments as foundational components required to scale AI. The company sees itself as constructing the enabling infrastructure for this transformation rather than betting on isolated startups.
SoftBank’s renewed AI optimism comes after years of volatility tied to aggressive venture investments, particularly during the early Vision Fund era. The group previously faced steep valuation losses across several portfolio companies, leading to investor concerns about its high-risk, high-concentration investment style.
The rapid increase in AI company valuations has revived comparisons to past technology bubbles, particularly given the scale of capital flowing into AI startups and infrastructure providers.
Critics warn that returns depend heavily on sustained demand growth and commercialisation of AI applications, both of which remain to some extent uncertain.
SoftBank is also expanding aggressively into artificial intelligence infrastructure through projects such as Stargate and a series of data centre and digital infrastructure investments that underpin its broader AI strategy.
The company recently partnered with OpenAI and its subsidiary SB Energy to advance the Stargate initiative, which includes large-scale AI data centre development. The project includes a 1.2 gigawatt data centre in Texas and long-term infrastructure agreements designed to support expanding AI computing workloads.
SoftBank has also moved to strengthen its presence in digital infrastructure through its $3.1 billion agreement to acquire DigitalBridge, a firmowns and invests in data centres, fibre networks and telecommunications infrastructure. The investment gives SoftBank indirect exposure to data centre operator Switch and other hyperscale infrastructure assets, reinforcing its position across the AI supply chain.
The company views such infrastructure spending as critical to enabling AI growth. During the earnings call, executives pushed back against concerns that rapid data centre construction reflects an overheating market.
“There is media coverage or discussion like the data centre business is a bubble or part of the AI bubble… Hyper scalers are in great need of data centres… the data centre capacity at the moment is not enough to support growing AI computing.” Yoshimitsu Goto, CFO of SoftBank, said.
“Technical innovation can be a risk… maybe some technology may come to provide capacity without using data centres and also electricity… investment in physical assets has such kind of risk,” Goto added.
For now, however, SoftBank remains confident that expanding AI workloads will require sustained infrastructure growth.
“In order for AI to grow, further computing power is needed… and that kind of computing is needed immediately,” Goto concluded.
Edited by Jyoti Narayan