Son remakes SoftBank in his own image
The veteran investor has put himself at the centre of the global AI boom. Some think he now has too much control
Last month in central Tokyo, Masayoshi Son stood on stage in front of a screen showing a goose producing a series of golden eggs from a Fritz Lang-inspired factory in its stomach.
One slide read: “Eggs do not lay eggs”. Another: “What matters is not the eggs. It is the Goose itself”.
To dispel any ambiguity, the SoftBank founder and Japan’s richest man suggested he should be called not “Chairman Son” but “Goose Son”. “Which would you prefer? Cash just in front of you or future golden eggs?” he said, before railing against holdout investors for doubting his vision.
The audience, largely SoftBank shareholders, erupted into applause. They have reasons to be cheerful. On June 1, SoftBank, the sprawling tech investment vehicle Son founded 45 years ago, overtook Toyota to become Japan’s most valuable company for the first time since the dotcom bubble.
It marked another remarkable comeback for Son, whose career has been defined by dramatic swings in fortune and who is derided by critics as a gambler who relies on luck over skill. Some of his wagers, such as on Alibaba and UK chip designer Arm, have paid off spectacularly. Others, like the billions lost on WeWork, have not.
Now Son has bet the house on AI, pouring billions of dollars into what he calls “the four corners of the ASI [artificial superintelligence] board”: data centres, AI models, chips and robots.
“This is really the period that we have to show what SoftBank is,” Son tells the FT.
Through Arm, massive funding of OpenAI and a plethora of other tech companies, Son has put himself at the centre of the AI boom, leaving an increasingly systemically important SoftBank once again heavily reliant on its founder.
Over the past decade, Son, who owns a third of SoftBank, has reshaped the company into what he wants it to be. He has doubled the personnel in his office to close to 60, representing more than a fifth of the company’s staff. Prominent lieutenants, presumptive heirs and high-profile independent board members have all departed.
For Son’s allies, who view him as one of the greatest investors of his generation, such moves are welcomed. Many buy SoftBank shares because of Son’s ability to take long-range bets. “Complaining about a lack of a brake on Masa is missing the point,” says a SoftBank insider.
But as a series of high-stakes market listings approaches, as well as complicated operational manoeuvres, some investors question whether SoftBank is too firmly in the grip of a singular CEO who has often had a partiality for charismatic corporate leaders.
In the past this predilection has worked in SoftBank’s favour, as with Son’s early backing of Jerry Yang at Yahoo and Jack Ma at Alibaba. But his wholehearted belief in Adam Neumann’s WeWork tarnished his reputation.
With some $64.6bn committed by SoftBank, for an eventual stake of roughly 13 per cent, Son’s largest bet of all could now be on Sam Altman’s OpenAI. A central reason for SoftBank’s share price surge is investors trying to get exposure to what is expected to be a blockbuster initial public offering. Listings by other SoftBank-backed companies, including robotics company Roze and energy and data centre developer and operator SB Energy, are also planned in the near future, all in service of what Son suggests is a cohesive strategy.
But he does not want to give his master plan away. “It’s like going to a boxing match, and you tell the audience what is your strategy. What is the punch that you are preparing to hit,” he says. “That means your opponent will also prepare how to protect. So I don’t want to explain too much.” He adds that he held back for years from explaining why he bought Arm.
Son’s secrecy unnerves some investors, as does the scale of SoftBank’s exposure to the AI industry, which on some measures resembles earlier tech-driven market bubbles. Since early June the company’s share price has fallen again by close to 25 per cent, in part due to suggestions that OpenAI’s IPO might be delayed. OpenAI not only has to time the market with its share issue but deal with huge competition in the form of SpaceX, which has already gone public, and Anthropic, a key rival that is preparing to.
Some analysts, such as Atul Goyal at brokerage Jefferies, believe that OpenAI is overly dependent on SoftBank’s own funding, with the Japanese group benefiting in turn from the rising valuation of the ChatGPT developer. Saying there are “parallels to WeWork”, Goyal warned clients this year that “risks for . . . investors are rising”.
If the IPO goes right, then it could help reset Son’s reputation. If it goes wrong, then it will compound fears that his strategy is misguided and that he is often late to the party, investing after valuations have hit critical levels.
“The fear is that Masa has learnt from his mistakes on the assets side but not about his investing style or how he bets on people,” says one Asia-based investor.
AI way or the highway
Shaping every decision Son makes is his unwavering belief in AI. Any suggestion of a bubble is “an insult”, he says, since the revolution is just beginning.
Markets will fall, wars will break out and crises are inevitable, he says. But downturns remain opportunities to buy, not excuses to retreat. His regret during the dotcom crash was not having the capital to keep buying. “At the moment of crash, I was saying internally, ‘This is the best moment that we should really invest,’” he says.
Son gives short shrift to fears of a dystopian AI future. He believes that models will become more “moral” as they grow more sophisticated and that the market is capable of a form of self-regulation. “These hyperscalers who have more power, they try to behave with the morals that they are supposed to . . . Otherwise, they get punished by regulators or by the citizens,” he says, adding that he believed there was more risk in open-source models.
The market is sceptical of Son’s strategy. SoftBank trades at a roughly 50 per cent discount to its net asset value, far from the premium enjoyed by Berkshire Hathaway, another holding company built around a totemic investor, but one that eschewed the use of debt finance and earned the markets’ trust.
The clearest example of the discount is Arm, of which SoftBank owns roughly 90 per cent. The share price of the UK company, bought in 2016 and relisted in 2023, has climbed by about 175 per cent this year. SoftBank’s stake is now worth more than the Japanese group’s own market capitalisation.
SoftBank’s price-to-organic-growth ratio — a measure which takes into account expected growth rates — based only on Arm and listed telecoms subsidiary SoftBank Corp, is an impressive 1.5 times, according to Richard Kaye, an investor at Comgest.
In effect, says Kaye, investors are getting the rest of the portfolio cheaply.
But there are further considerations.
Some are wary of the group’s complex financing, involving the rapid leveraging of assets and margin loans against its stakes in companies, including Arm. The group’s current leverage — as seen through its own stated 17 per cent loan-to-value ratio at the end of March — has been helped by rising asset prices but that could quickly change if markets fall. Having put in place $40bn in bridge loans for its latest investment in OpenAI, SoftBank is pushing up against borrowing limits. More listed equity to borrow against would be one way to allow Son to access cheaper financing.
Others worry about SoftBank’s status as both investor and customer of Arm and OpenAI. SoftBank paid Arm roughly $704mn in the fiscal year to March, according to filings, and agreed a $3bn a year contract with OpenAI in 2025 to develop a joint enterprise product called Cristal Intelligence.
Shareholders also grumble about governance and Son’s personal co-investment arrangements — around SB Northstar, an internal asset management arm, and Vision Fund 2, which holds the OpenAI investment — which they say could create conflicts of interest. The company notes that board members have defended the concept.
The lack of people inside SoftBank capable of restraining Son only exacerbates fears. The group’s influential chief financial officer, Yoshimitsu Goto, who presents quarterly earnings, has limited capacity to rein him in, say people familiar with company operations, despite internal concerns about the scale of investment in OpenAI and the faith placed in Altman.
But it is the traits SoftBank presents as virtues that cause investors most pause — its long-term asymmetrical bets and willingness to move before the market understands the plan. “IPO is not the goal, IPO is just the beginning,” Son says.
He sees a future of power plants feeding his data centres, built by his robots and using Arm-produced chips, that will provide computational power to LLM companies such as OpenAI. It represents a shift from a more scattergun investment approach and a return by Son to something closer to the telecoms-led operating model he put together close to 20 years ago.
“SoftBank is becoming more and more integrated compared to the past decade, where it invested in various companies,” says David Dai at brokerage Bernstein Research. “It has never been as synergetic as it is today.”
“At times, investing in SoftBank has been close to making a leap of faith,” says Kaye at Comgest, who had been forced to defend the logic of his investment since first buying into the group in the mid-2000s. Now, he says, that decision “is based on increasingly large amounts of evidence”.
Secrecy still surrounds Son’s plans and the companies he is trying to bring to market. Roze, SoftBank’s robotics bet, is targeting a valuation as high as $100bn, but Son says he does not want to go into detail about what the company is preparing.
When the robots are introduced, he says, “people will say: wow, that’s what you wanted to create.” He says he will be ready to explain more “step by step” and “sometime next year”, giving the impression that IPO plans might be pushed back.
To persuade investors to put more value on SoftBank’s strategy, Son might have to slow down and more clearly explain it, say analysts and investors. That would require him to overturn decades of fast-paced risk-taking based on his own long-term visions.
“Masa has been ‘Masa’ since at least the mid-90s,” says Alex Clavel, CEO of the Vision Funds.
Son on the global stage
The child of ethnically Korean parents, Son left home as a teenager to study in the US. He returned to Japan in the 1980s convinced that software would define the future.
He built SoftBank as a software distributor, rode the dotcom boom through early bets on Yahoo and Alibaba, and then lost almost everything on paper when the bubble burst.
He recovered by investing in Japanese telecoms, acquiring Vodafone Japan, breaking NTT’s grip and securing the country’s first iPhone deal from Steve Jobs — the sort of operational manoeuvring that gives allies hope he can pull off his current plans.
Alibaba’s listing restored his fortune and paved the way for the Vision Funds, tech investment vehicles that changed the nature of venture capital, before AI became his next all-in wager.
“He is one of the only people who has been around for internet one, internet two and now three, with AI,” says one long-serving SoftBank insider.
His dominance of SoftBank began to dilute in the years leading up to the creation of the first Vision Fund in 2017, when Son surrounded himself with lieutenants — including presumptive heir Nikesh Arora from Google and former Deutsche Bank hires Rajeev Misra and Akshay Naheta — to raise capital, strike deals and professionalise SoftBank’s investment machine.
That era produced some of SoftBank’s most aggressive moves and damaging internal fights. It was punctuated by controversy, including large bets on Greensill and Wirecard, two companies engulfed in scandal, and huge tech options trades that distorted the market through SB Northstar, which earned the group the nickname the “Nasdaq Whale”.
After taking on huge losses coming out of Covid, Son went on the defensive in 2022 and “lost interest in the Vision Fund”, say people close to him, folding much of its functions back into the group as he reverted to type.
“There used to be other voices inside SoftBank and with investors,” says Goyal of Jefferies. “That is no longer the case.”
SoftBank, however, says that healthy debates still define which investments are made.
The board, regularly criticised as weak by investors, does push back, including this year when it questioned when funds promised by Japan for a $33bn gas-fired power plant in Ohio would be delivered. The plant, agreed under a $550bn trade deal between Japan and the US, is being developed by SoftBank and will eventually power its mammoth data centre nearby.
“This is not Masa proclaiming down from his mountaintop. Yes, Masa is the boss and always the final decision maker, but he makes his decisions based on analysis and following debate,” says Clavel, Vision Funds’ CEO.
More so than ever, Son’s role as one of Japan’s wealthiest people gives him outsized political clout despite many treating him as a perennial outsider. Already a key player in Japan’s AI strategy, Son is growing in influence on the global stage.
After an intimate dinner with French President Emmanuel Macron in Tokyo in April, Son pledged an up to €75bn data centre rollout in France. He is key to Japan’s trade agreement with the US and is in regular contact with President Donald Trump, in meetings and on golf courses. He also has a large exposure to China through investments such as ByteDance.
The deeper SoftBank pushes into critical infrastructure, the more uncomfortable questions it raises for lenders, regulators and politicians.
Multiple bankers and government officials describe the company as approaching “too big to fail” status. Japanese megabanks are searching for ways to spread the risk they have taken on.
Some inside SoftBank believe that if Son ever does retire, the group as it currently exists will be finished.
It could be broken up, with SoftBank Corp, which was listed in 2018, a template for survival. Others believe rising star Rene Haas, Arm CEO and the newly promoted executive driving SoftBank’s AI chip strategy, is waiting in the wings.
There is also a chance, say people familiar with SoftBank’s strategy, that the group might have to make more big operational hires once again to help manage its growing businesses, raising the question of whether Son could tolerate a new generation of powerful lieutenants.
Last year he said he had a handful of successors in mind and would hand over the reins in a decade. Now he says he has a constellation of subsidiary and portfolio company bosses to choose from and has stretched his retirement timeline out again, a move surprising to some of his closest advisers.
On stage in Tokyo with his golden goose, a metaphor he has conjured in the past, Son promised to lift SoftBank’s net asset value, his preferred measure of the group’s worth, from ¥74tn to ¥1 quadrillion ($6.189tn), powered by artificial superintelligence.
He appeared determined to stick around. “I’d planned to hand over in my sixties. But now I’m 68, I’ve become greedy . . . I’m revising that plan,” he told the audience. “I’m going to keep going for another 10 to 15 years.”
