Masayoshi Son, chairman and chief govt officer of SoftBank Group Corp., speaks on the SoftBank World occasion in Tokyo, Japan, on Wednesday, July 16, 2025.
Kiyoshi Ota | Bloomberg | Getty Pictures
Masayoshi Son is making his largest guess but: that his brainchild SoftBank would be the heart of a revolution pushed by synthetic intelligence.
Son says synthetic superintelligence (ASI) — AI that’s 10,000 instances smarter than people — might be right here in 10 years. It is a daring name — however maybe not shocking. He is made a profession out of massive performs; notably, one was a $20 million funding into Chinese language e-commerce firm Alibaba in 2000 that has made billions for SoftBank.
Now, the billionaire is hoping to copy that success with a sequence of investments and acquisitions in AI companies that may put SoftBank on the heart of a basic technological shift.
Whereas Son has been outspoken about his imaginative and prescient during the last yr, his pondering precedes a lot of his latest bullishness, in line with two former executives at SoftBank.
“I vividly bear in mind the primary time he invited me to his dwelling for dinner and sitting on his porch over a glass of wine, he began speaking to me about singularity – the purpose at which machine intelligence overtakes human intelligence,” Alok Sama, a former finance chief at SoftBank till 2016 and and president till 2019, informed CNBC.
SoftBank’s huge AI performs
For Son, AI appears private.
“SoftBank was based for what function? For what function was Masa Son born? It might sound unusual, however I believe I used to be born to appreciate ASI,” Son mentioned final yr.
That will go some approach to clarify what has been an aggressive drive over the previous few years — however particularly the final two — to place SoftBank on the heart of the AI story.
In 2016, SoftBank acquired chip designer Arm in a deal price about $32 billion on the time. As we speak, Arm is valued at greater than $145 billion. Whereas Arm blueprints type the idea of the designs for practically all of the world’s smartphones, today, the corporate is trying to position itself as a key player in AI infrastructure. Arm-based chips are a part of Nvidia’s programs that go into knowledge facilities.
In March, SoftBank additionally introduced plans to amass one other chip designer, Ampere Computing, for $6.5 billion.
ChatGPT maker OpenAI is one other marquee funding for SoftBank, with the Japanese big saying not too long ago that deliberate investments within the firm will attain about 4.8 trillion Japanese yen ($32.7 billion).
SoftBank has additionally invested in quite a few different firms associated to AI throughout its portfolio.
“SoftBank’s AI technique is complete, spanning the complete AI stack from foundational semiconductors, software program, infrastructure, and robotics to cutting-edge cloud companies and finish purposes throughout important verticals akin to enterprise, schooling, well being, and autonomous programs,” Neil Shah, co-founder at Counterpoint Analysis, informed CNBC.
“Mr. Son’s imaginative and prescient is to cohesively join and deeply combine these parts, thereby establishing a robust AI ecosystem designed to maximise long-term worth for our shareholders.”
SoftBank’s inventory efficiency since 2017, the yr that its first Imaginative and prescient Fund was based.
There’s a widespread theme behind SoftBank’s investments in AI firms that comes immediately from Son — particularly, that these companies must be utilizing superior intelligence to be extra aggressive, profitable, to make their product higher and their prospects completely satisfied, an individual accustomed to the corporate informed CNBC. They may solely remark anonymously due to the sensitivity of the matter.
It began with and mind computer systems and robots
As SoftBank launched “SoftBank’s Subsequent 30-12 months Imaginative and prescient” in 2010, Son spoke about “mind computer systems” throughout a presentation. He described these computer systems as programs that might be taught and program themselves ultimately.
After which got here robots. Main tech figures like Nvidia CEO Jensen Huang and Tesla boss Elon Musk are actually speaking about robotics as a key software of AI — however Son was pondering about this greater than a decade in the past.
In 2012, SoftBank took a majority stake in a French firm referred to as Aldebaran. Two years later, the 2 firms launched a humanoid robotic referred to as Pepper, which they billed as “the world’s first private robotic that may learn feelings.”
Later, Son said: “In 30 years, I hope robots will turn into one of many core companies in producing earnings for the SoftBank group.”
SoftBank’s guess on Pepper in the end flopped for the corporate. SoftBank slashed jobs at its robotics unit and stopped producing Pepper in 2020. In 2022, German agency United Robotics Group agreed to amass Aldebaran from SoftBank.
However Son’s very early curiosity in robots underscored his curiosity for AI purposes of the longer term.
“He was in very early and he has been excited about this obsessively for a very long time,” Sama, who’s writer of “The Cash Lure,” mentioned.
Within the background, Son was cooking up one thing larger: a tech fund that might make waves within the investing world. He founded the Vision Fund in 2017 with a large $100 billion in deployable capital.
SoftBank aggressively invested in firms the world over with a few of the largest bets on trip hailing gamers like Uber and Chinese language agency Didi.
However investments in Chinese language expertise firms and a few bad bets on firms like WeWork soured sentiment for the Imaginative and prescient Fund because it racked up billions of dollars of losses by 2023.
Imaginative and prescient however dangerous timing
The market questioned a few of Son’s investments in firms like Uber and Didi, which have been burning by money on the time and had unclear unit economics.
However even these investments spoke to Son’s AI view, in line with the previous associate on the SoftBank Imaginative and prescient Fund.
“His thought again then was the primary creation of AI can be self-driving vehicles,” the supply informed CNBC.
Once more this may very well be seen as a case of being too early. Uber created a driverless automotive unit solely to promote it off. As a substitute, the corporate has targeted on different self-driving automotive firms to convey them onto the Uber platform. Even now, driverless vehicles should not widespread on roads, although industrial companies like these of Waymo can be found.
SoftBank nonetheless has investments in driverless automotive firms, akin to British startup Wayve.
Timing clearly wasn’t on Son’s aspect. After file losses on the Imaginative and prescient Fund in 2022, Son declared SoftBank would go into “defense” mode, considerably lowering investments and being extra prudent. It was at the moment that firms like OpenAI have been starting to achieve steam, however nonetheless earlier than the launch of ChatGPT that might put the corporate on the map.
“When these firms got here to move in 2021, 2022, Masa would have been in an ideal place however he had used all his ammunition on different firms,” the previous Imaginative and prescient Fund exec mentioned.
“After they got here to age in 21, 22, the Imaginative and prescient Fund had invested in 5 or 6 hundred totally different firms and he was not able to spend money on AI and he missed that.”
Son himself mentioned this yr that SoftBank wished to spend money on OpenAI as early as 2019, nevertheless it was Microsoft that ended up changing into the important thing investor. Quick ahead to 2025, the Imaginative and prescient Fund — of which there are actually two — has a portfolio stacked filled with AI targeted firms.
However that interval was powerful for buyers throughout the board. The Covid-19 pandemic, booming inflation and rising charges hit private and non-private markets throughout the board after years of unfastened financial coverage and a tech bull run.
SoftBank did not see that point as a missed alternative to spend money on AI, an individual accustomed to the corporate mentioned.
As a substitute, the the corporate is of the view that it’s nonetheless very early within the AI investing cycle, the supply added.
Danger and reward
AI expertise is fast-moving, from the chips that run the software program to the fashions that underpin fashionable purposes.
Tech giants within the U.S. and China are battling it out to provide ever-advancing AI fashions with the goal of reaching synthetic basic intelligence (AGI) — a time period with totally different definitions relying on who you converse to, however one which broadly refers to AI that’s smarter than people. With billions of {dollars} of funding going into the expertise, the chance is excessive, and the rewards may very well be even larger.
However disruption can come out of no the place.
This yr, Chinese language agency DeepSeek made waves after releasing a so-called reasoning mannequin that appeared to be developed more cheaply than its U.S. rivals. The truth that a Chinese language firm managed the feat, regardless of all of the export restrictions for superior tech in place, rocked global financial markets that have been betting the U.S. had an unassailable AI lead.
Whereas markets have since recovered, the potential of shock advances in expertise at such an early stage in AI stays an enormous threat for the likes of SoftBank.
“As with most expertise investments the important thing problem is to spend money on the successful applied sciences. Lots of the investments SoftBank has made are within the present leaders however AI remains to be in its relative infancy so different challengers might nonetheless rear up from nowhere,” Dan Baker, senior fairness analyst at Morningstar, informed CNBC.
Nonetheless, Son has made it clear he desires to set SoftBank up with DNA that may see it survive and thrive for 300 years, in line with the corporate’s web site.
That will go some approach to clarify the large dangers that Son takes, and his conviction in terms of explicit themes and corporations — and the valuations he is prepared to pay.
“He (Son) made some errors, however directionally he’s going in the identical driection, which is — he desires to make sure that he’s an actual participant in AI and he’s making it occur,” the previous Imaginative and prescient Fund exec mentioned.