Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
SoftBank’s founder Masayoshi Son is known for chasing the next big thing. Some of his big bets on tech companies, such as Alibaba and Arm, have paid off. There have also been expensive flops such as office-share provider WeWork. A bet on Son is a bet on an investment strategy driven by conviction as much as fundamental value.
Now, Son is making his most ambitious bet yet — on artificial intelligence. With high upfront investment demands, uncertain future returns and a reliance on backing the right horse, AI is right up Son’s street.
SoftBank’s latest earnings highlight the challenge. The company fell to a loss of ¥369.2bn ($2.4bn) in the quarter to December, compared with a gain of ¥950bn in the same quarter the year before, missing analysts’ forecasts for a net profit. A ¥352.7bn loss in its Vision Funds underscores the volatility of its tech-heavy portfolio and could raise concerns about the financial logic of its latest bet: Stargate.
That $500bn AI infrastructure initiative — essentially an enormous collection of data centres — is Son’s most high-profile attempt yet at establishing SoftBank as a contender in the AI revolution. If successful, it could open up lucrative, and much needed, stable revenue streams in cloud computing, AI model training and licensing.
But there are challenges. Entering the AI infrastructure sector is capital-intensive, with long development cycles and competition from Amazon and Google. That means rivals could secure a dominant market share before Stargate reaches scale. It also means expenses will come first, while earnings may not get a meaningful boost until the infrastructure is fully operational.
The good news is that Son is approaching the bet with a more measured approach to risk this time around. Unlike WeWork, where SoftBank took on much of the financial burden, Stargate will be structured through project financing, with external investors shouldering a large chunk of the risk. SoftBank will contribute the equity portion — which is likely to be about 10 per cent to 20 per cent — limiting its direct exposure while maintaining a strategic foothold in the latest advancements in the AI sector.
Shares are up more than 50 per cent from an August low, reflecting renewed hopes for Son’s AI-driven strategy. On Wednesday, SoftBank said it would execute the first Stargate project within this fiscal year, potentially accelerating the timeline to generate AI infrastructure revenue. For now, the market is giving Son the benefit of the doubt — but in the high-stakes world of AI, SoftBank will need to deliver tangible earnings fast.
june.yoon@ft.com
