Masayoshi Son Leadership Style: Conviction Betting at SoftBank

Key Facts: Masayoshi Son (born August 11, 1957, in Tosu, Saga Prefecture, Japan) founded SoftBank in 1981 as a software distributor after graduating from UC Berkeley in 1980, and has run it as founder, chairman, and CEO ever since. In 2000, he put roughly $20 million into a little-known Chinese startup, Alibaba, a stake Wikipedia records as worth $132 billion by 2018 before SoftBank began selling it down. That same year, the dot-com crash erased an estimated $58.6 billion of Son's own net worth, a Guinness World Record for the largest single-year personal wealth loss that stood until Elon Musk's 2022 decline surpassed it. Son launched the $100 billion SoftBank Vision Fund in 2017, drove it to a record 4.3 trillion yen ($32 billion) loss for the fiscal year ended March 2023, and wrote SoftBank's roughly $14.3 billion WeWork investment down to zero after WeWork's November 2023 bankruptcy. Arm Holdings, the chip designer SoftBank still majority-owns, listed on Nasdaq on September 14, 2023, in a $5.23 billion IPO that valued it at $54.5 billion. On January 21, 2025, Son stood at a White House event with President Trump, OpenAI CEO Sam Altman, and Oracle chairman Larry Ellison to announce Project Stargate, a venture targeting up to $500 billion in AI data center investment by 2029; SoftBank put in $19 billion of the initial capital and Son chairs the venture. SoftBank separately committed $40.0 billion to OpenAI on March 31, 2025, funded in two closings through SoftBank Vision Fund 2 ($7.5 billion in April 2025 and $22.5 billion on December 26, 2025) plus $11.0 billion from co-investors, for $41.0 billion funded and roughly 11% ownership of OpenAI, financed in part by selling SoftBank's entire Nvidia stake for $5.83 billion and $9.17 billion of T-Mobile shares in 2025.

The Conviction Bet Doctrine

Son doesn't diversify his way to safety. He picks a small number of ideas he believes will define the next few decades, puts an amount of capital behind each one that would terrify most boards, and holds through drawdowns that would get most CEOs fired. The doctrine has a name inside SoftBank: the 300-Year Vision, Son's stated plan for SoftBank to keep making these bets for three centuries, treating the next thirty years as merely the opening move. It's the same instinct that put $20 million into Alibaba when nobody outside Hangzhou had heard of it, and the same instinct now putting tens of billions into OpenAI. Whether that's a repeatable method or a survivorship story, one bet in a hundred paying for the other ninety-nine, is the question this profile keeps circling back to.

Son was born in 1957 in Tosu, on the southern Japanese island of Kyushu, into a Zainichi Korean family. He has said he faced discrimination for his Korean heritage as a child, and at sixteen he left for California, eventually enrolling at UC Berkeley, where he studied economics and computer science. As a 19-year-old undergraduate he designed a multilingual translator device with help from his professors, patented it, and sold it to Sharp Corporation for close to $1 million, an early proof that he could turn conviction into capital before he'd built a company at all. He graduated in 1980 and returned to Japan the following year to found SoftBank in Fukuoka, initially a distributor of packaged software to Japanese retailers.

Through the 1990s, Son built SoftBank into a publishing and internet holding company, buying trade shows, magazines, and stakes in early web properties, including an early investment in Yahoo. By 1999, at the height of the dot-com bubble, his paper net worth briefly made him the richest person in the world for a few days, before the crash of 2000 wiped out most of it within months. SoftBank's own shares fell more than 90% from their peak. Son has said the loss taught him to size his conviction bets against his own survival, not against the market's mood.

It was in that same collapsing year that Son made the decision this profile can't avoid. In January 2000 he sat down with a young English-teacher-turned-entrepreneur named Jack Ma, and after a meeting Son himself describes as lasting about six minutes, decided to put roughly $20 million into a company called Alibaba that had, by Son's own account, no business plan and no revenue. He's said he backed the "look in his eye," not a spreadsheet, a founder-first instinct that shows up again and again in how he underwrites bets. That stake reached a peak paper value of more than $200 billion in 2020, but SoftBank has since sold down most of the position rather than holding it to today's valuation, so the amount Son and SoftBank actually realized is smaller than the peak figure most retellings quote.

Two decades later, the same instinct is funding a different kind of bet: not a single startup, but the physical infrastructure underneath the entire AI industry. Son now chairs Project Stargate, sits on OpenAI's cap table at roughly 11% ownership, and has sold off Nvidia, T-Mobile, and other holdings to fund it. It's the largest concentrated bet of his career, sized against a technology wave he's convinced won't wait for him to be cautious.

Leadership Style Breakdown

The 50/25/25 split matters because concentration without founder judgment is just recklessness, and founder judgment without a long horizon collapses the first time a bet goes against you publicly. Son's size of conviction only works because he's willing to back a person he's met for minutes over a business plan he hasn't read, and because he's built a public narrative, the 300-Year Vision, that gives him room to hold a losing position through years of criticism. Take away any one leg and the other two don't survive contact with a bad year.

Key Leadership Traits

Son pairs a long horizon and unusual risk tolerance with founder judgment, but the same concentration makes governance discipline essential.

The 3 Decisions That Defined Masayoshi Son as a Leader

Son's major decisions show how conviction changes both the size of a commitment and the assets he is willing to sell to fund it.

1. Backing Jack Ma on a Six-Minute Read, Not a Business Plan (2000)

Son met Jack Ma in January 2000, in the middle of the dot-com crash that would go on to erase most of his own net worth that same year. Ma's company had no revenue model, no reference customers Son could call, and nothing resembling the diligence package SoftBank's own investment team would normally require. Son decided to invest anyway, in a meeting he's said lasted about six minutes, putting in roughly $20 million against Ma's ask for closer to $40 million.

The lesson isn't "trust your gut and skip diligence." It's that Son had already built, over two decades of founding and running his own company, a specific pattern he was testing for: founder conviction that survives contact with total uncertainty. He wasn't guessing at Alibaba's market size. He was pattern-matching Ma's certainty against his own experience of building something from nothing. An operator applying this today should ask what pattern they've actually earned the right to bet on through hard-won experience, not treat "go with your gut" as a substitute for having a pattern at all.

2. Raising a $100 Billion Fund to Change What a Bet-Sized Decision Even Means (2017)

By 2017, Son wasn't just making individual conviction bets anymore. He raised the SoftBank Vision Fund at roughly $100 billion, an amount larger than the entire venture capital industry had deployed in most prior years combined, with backing from Saudi Arabia's Public Investment Fund and Abu Dhabi's Mubadala. The fund's structure let Son write checks so large that a single SoftBank investment could set the valuation for an entire sector, from ride-hailing to co-working space.

That scale is also what tested SoftBank's strategic fit as an organization, whether a company built to run telecom and internet businesses could actually govern a $100 billion venture portfolio, and made the fund's failures so visible. When WeWork and a run of other bets went wrong, the losses weren't contained to a normal-sized venture fund's mistakes, they became a multi-billion-dollar drag on SoftBank's public earnings, discussed in the sections below. The lesson for an operator scaling any bet-sizing decision, capital allocation, hiring, market entry, is that the size of your commitment changes the category of mistake you're capable of making. A bet that fails at $20 million is a write-off. The same bet failing at $10 billion is a governance crisis.

3. Chairing Project Stargate and Funding a $40 Billion OpenAI Bet by Selling the Rest of the Portfolio (2025)

In January 2025, Son stood at a White House event alongside President Trump, Sam Altman, and Larry Ellison to announce Project Stargate, an up to $500 billion AI infrastructure venture, with SoftBank holding financial responsibility and Son chairing it. Weeks later, SoftBank committed $40.0 billion directly to OpenAI, funded through two closings and, according to Son, financed in part by selling holdings he's said he didn't want to part with, including SoftBank's entire Nvidia stake, a position that had been one of SoftBank's best-performing bets of the AI boom.

Son has said publicly that he was "crying to sell Nvidia shares" but did it anyway, because the OpenAI bet needed the cash more than the Nvidia position needed to stay on the balance sheet. The lesson is a hard one for most operators to actually apply: conviction sometimes means selling your best-performing asset to fund your next one, rather than holding a winner indefinitely because it's comfortable. Portfolio discipline, not just bet selection, is part of what conviction-based leadership requires.

What Masayoshi Son Would Do in Your Role

If you're a CEO or founder, the Son move is to identify the two or three bets that could actually redefine your company's next decade, and refuse to spread your attention or capital evenly across everything on your roadmap instead. Most leaders dilute conviction by treating every initiative as equally important. Son's entire career argues the opposite: decide what actually matters, then commit resources at a scale that makes the bet real, even when it looks reckless from outside.

If you're a CFO or finance leader, the harder Son lesson is about funding conviction honestly. He didn't fund the OpenAI bet by hiding the cost, he funded it by selling Nvidia and T-Mobile stakes and saying publicly what that cost him. If your organization's biggest bet needs funding, look at what you're actually willing to give up for it before the number gets forced on you by a crisis, and be transparent with your board about the trade-off instead of pretending it's free.

If you're a manager evaluating people or opportunities, Son's founder-first pattern matching is worth stealing in a scaled-down form: build your own track record of what genuine conviction looks like in someone before you bet on them, then trust that pattern when the spreadsheet alone isn't conclusive. The risk is applying this without ever having earned the pattern through your own failures first, which is closer to guessing than judgment.

Notable Quotes & Lessons Beyond the Boardroom

On why he backed Jack Ma without seeing a business plan, Son has described his reasoning as almost entirely instinctual: "It was the look in his eye, it was 'animal smell.'" It's a strange phrase to build an investment thesis on, and it's also a reminder that Son has always treated the founder as the actual asset, with the business plan as a secondary artifact.

On selling SoftBank's entire Nvidia stake to help fund the OpenAI commitment, Son told reporters plainly: "I don't want to sell a single share. I just had more need for money to invest in OpenAI and other projects... I was crying to sell Nvidia shares." That's not the language of a leader who enjoys risk for its own sake. It's a leader describing a cost he paid deliberately because he judged the next bet was worth more than the comfort of the current one.

Son's SoftBank talks about a 300-Year Vision for the company, an explicit mission and vision statement stretched to a scale almost no other public company attempts, a commitment to think in centuries rather than quarters. Critics have pointed out that a plan spanning three hundred years is functionally unfalsifiable, no single decade can prove or disprove it. That's also precisely why it works as an internal narrative: it gives Son and SoftBank's employees room to treat a bad year, or a bad decade, as a chapter rather than a verdict. Whether that's wisdom or a convenient excuse for holding losing positions too long is a fair question, and it's one this profile answers directly in the next section.

Where This Style Breaks

The same founder-instinct that found Jack Ma also found Adam Neumann, whose own founder-to-CEO transition at WeWork ended in his ouster rather than the kind of long-term stewardship Son bet on. SoftBank poured roughly $16 billion into WeWork starting in 2017, including a $9.5 billion rescue package after WeWork's botched 2019 IPO attempt, and after WeWork's November 2023 bankruptcy filing, SoftBank wrote the entire position down to zero, a loss it puts at roughly $14.3 billion. Son later told shareholders directly, "I fell in love with WeWork," adding that board members had warned him his faith in Adam Neumann was misguided and that he himself may bear more fault than Neumann for pushing the founder to be more aggressive. The same read-the-founder instinct that made the Alibaba bet look prescient made the WeWork bet look reckless, and there's no version of the doctrine that reliably tells you in advance which one you're in.

That pattern shows up at the portfolio level too. SoftBank's Vision Fund segment posted a record 4.3 trillion yen ($32 billion) loss for the fiscal year ended March 2023, its worst year since Son launched the fund in 2017, driven by falling valuations across a portfolio built on the same concentrated-conviction logic that produced Alibaba. Governance concerns compounded the problem: by early 2023, Son personally owed SoftBank about $5.1 billion on side-fund arrangements tied to his own compensation structure inside Vision Fund 2, an arrangement that concentrated both the upside and the personal financial risk of the fund's bets in Son himself.

None of this resolves the question this profile opened with: whether Son's method is repeatable or a survivorship story. The honest answer is that it's both. Concentrated conviction, backed by a founder's willingness to sell his best-performing asset and absorb public losses without changing course, produced Alibaba and is now funding the biggest AI infrastructure bet of his career. The same concentration, without a system to catch a bad read early, produced WeWork and a governance structure regulators and shareholders have openly questioned. Anyone borrowing this style should copy the discipline of sizing bets to conviction, and build in the guardrails, on personal financial entanglement, on independent diligence, on when to cut a losing position, that Son's version of the doctrine has repeatedly skipped.

Explore related profiles: Jack Ma, John Doerr, and Jensen Huang.