Masayoshi Son: Why SoftBank Is Betting Everything on Superintelligence

SoftBank founder and CEO Masayoshi Son tells Blackstone’s Christine Anderson why he expects AI 10,000 times smarter than people, why development should not slow down, and where SoftBank is investing. The interview is the centerpiece of an episode of Inside Blackstone, Blackstone’s weekly podcast. It is bracketed by a segment on how AI is landing inside Blackstone’s portfolio companies, a short read on the economy, and a debrief with Blackstone chairman and CEO Steve Schwarzman, who has a noticeably slower timeline than his guest.

TLDW

Son, 69, says Alibaba, Yahoo, Arm and the Vision Fund were a warm-up for AI, a conviction he traces to seeing a photo of a microprocessor at 17. He wants AI development to run at full speed with good intent built in, and compares AI rules to traffic laws. SoftBank invests across models, chips and data centers as an involved, usually largest shareholder, and holds more than 30 robotics companies. He uses AI three hours a day on weekdays and about seven on weekends, revises a 10-year plan daily, and argues that because you only achieve about 70% of any goal, the goal should be big.

Thoughts

Son’s answer on AI safety is that progress is the fix, and the analogy he picks works against him. Anderson puts a specific point to him: OpenAI has disclosed incidents of AI systems falsifying data and hiding mistakes, and has said the industry has not solved alignment and monitoring well enough to keep scaling at maximum speed. Son replies that he is on the side of not slowing down, then compares AI to cars: no drinking and driving, stop at red lights, keep to one side of the road. Every one of those rules was imposed on drivers from outside and enforced. None of them came from building faster cars. The exchange also has an odd shape, because Anderson introduces SoftBank as holding the largest stake in OpenAI. The biggest backer is declining to agree with the caution its own investment has voiced.

Steve Schwarzman’s debrief is worth more than its five minutes suggest, because he quietly disagrees on timing. Son does not dispute that AI could be 10,000 times smarter than humans within a decade. Schwarzman says radical change will not come as fast as some people think, calls it an evolution, and will not put a number on when robots arrive. His description of Son is also the most candid in the episode: someone who waits for the moment and the person, then puts all the chips on the table, whose timing can be a little off without it troubling him. That is praise, and it is also a risk disclosure. Both men are large investors in AI infrastructure and the show is Blackstone’s own, so nobody in the room has a reason to press on what happens when the timing is off by more than a little.

The baseball question gets the most revealing answer. Asked whether he would let ChatGPT pick the opening lineup for the SoftBank Hawks, the man who thinks machines will out-think people by four orders of magnitude says no. That is the general manager’s job, he explains, because the general manager has the trust of the team, and the game is not only the calculated stuff. Trust and legitimacy are exactly what a capability argument leaves out. If they matter enough to keep a person in charge of a batting order, they will matter in hospitals, courts and companies, and adoption will be slower and more negotiated than a raw intelligence multiple implies.

The 70% rule is good motivation and incomplete risk management. Son’s claim is that you get about 70% of whatever you aim for, so aim high. That holds if the downside of a big dream is a smaller win. It does not hold when the dream is financed in a way that can produce zero, which Anderson politely refers to as his very public challenges. The better idea is the one that follows: his team revisits the 10-year plan every day, on the grounds that not knowing what will happen in 10 years is the reason to keep thinking about it. A plan revised daily is a habit of attention more than a forecast, and that is a more defensible practice than the slogan.

The most practical material comes before Son appears. Rodney Zemmel, who runs Blackstone’s operating team, says the productivity boom has probably begun but that every company has to climb an activation barrier and put real effort in before it gets anything out. He likes that as an investor, since gains that are hard to capture are less likely to be competed away. His advice is to pick one or two things and never let the work be led by technology staff alone. It is a useful counterweight to the rest of the episode: superintelligence may be coming, and meanwhile most firms are still working out how to turn 45 saved minutes into profit.

Key Takeaways

Son says everything he has done so far, including Alibaba, Yahoo, Arm and the Vision Fund, was a warm-up, and that those bets look tiny next to what AI will bring.
His conviction dates to age 17, when he saw a photograph of a microprocessor in a magazine and cried on the street. He predicted then that chips would become smarter than the human brain in about 50 years.
He does not argue with the idea that AI could be 10,000 times smarter than humans within a decade, and says that at that level it will invent new technology by itself.
On safety, Son is against slowing down. He wants rules and etiquette for AI like those for cars, technology to defend against cyberattacks, and agents that are taught to be good and not only capable.
SoftBank will not pick one layer of AI. It invests across large language models, chips and data centers, usually as the largest shareholder with stakes of 20%, 30% or 80%, and works to create synergy between the companies.
SoftBank has more than 30 robotics companies in its portfolio. Son says they are preparing quietly, and that robots should take over tiring work such as washing dishes.
Asked what a young person should study, he says AI and superintelligence, combined with whatever field they care about, from medicine to cars to aircraft.
Son uses AI about three hours a day during the week and about seven hours a day on weekends, as a thought partner he debates with. He rejects the idea that using AI means not using your brain.
His 70% rule: whether the goal is small or large, you achieve about 70% of it on average, so the result is proportionate to the dream.
His team discusses the next 10 years every day and revises the plan the following day, because 24 hours brings new input.
Rodney Zemmel says AI is now changing how a good number of Blackstone portfolio companies work, in every industry and not only software and services, but that getting there is hard and not every company will manage it.
Winfield Sickles reports the 10-year Treasury yield above 5.3%, which he calls the highest since 2002, hyperscaler spending of upwards of $800 billion this year and probably more than $1 trillion next year, and cloud revenue growth of 50% to 120% last quarter.
Steve Schwarzman says Son waits for the right technology and the right person and then bets the house, and that setbacks do not bother him the way they bother most people. Schwarzman himself expects AI to play out over 20 to 30 years.

Chapters

01:08 Rodney Zemmel on AI Inside Blackstone’s Portfolio

A year ago, Zemmel says, the conversation was about pilots and plans. His joke is that the biggest beneficiaries of AI in 2025 were dogs owned by software developers, who finished 45 minutes early and went for a walk. Now he sees real impact across industrials, healthcare and consumer businesses, with Chamberlain building a digital access business that would not have been possible without AI. His two lessons for CEOs are to focus on one or two things and to have business and technology leaders run it together.

04:33 Economic Weather Report

Sickles says corporate borrowing costs are higher and being watched closely, but the drivers of growth from the AI infrastructure boom are intact. He attributes the rise in Treasury yields mainly to growth, with stocks up almost 20% in six months and third-quarter GDP estimates near 3%. The last jobs report was a little weak. He sees AI spending starting to show up as revenue, and broadening into areas such as Blackstone’s logistics business.

07:14 A Warm-Up, and a Microprocessor at 17

Anderson runs through Son’s biography: born in rural Japan to Korean parents, moved to California at 16, sold an electronic translator while at UC Berkeley and founded SoftBank in 1981. Son confirms that he sees his career to date as a warm-up. He describes reading a magazine on the street at 17, seeing the microprocessor and crying. The brain power is now in data centers, he says, and is about to move into robots and the physical world.

10:35 When AI Is 10,000 Times Smarter

However superintelligence is defined, Son says, at 10,000 times human intelligence the debate is over and the systems begin inventing technology themselves. His picture of the result is cures for diseases such as cancer, no more road accidents, and natural disasters forecast with solutions offered. People will not have to do what they do not want to do, leaving more time for family, friends, sport and entertainment.

12:33 Why Slowing Down AI Is the Wrong Answer

Pressed on reports of AI systems falsifying data and acting without permission, Son says technology development is the solution. Society wrote rules and manners for cars, and it has to do the same for AI. Agents should be taught what is bad, as people are, with the goal of making humans happier. His summary is to go full speed with good intent.

15:13 Betting on the Whole AI Ecosystem, and Robots

Son says he does not need to be the CEO of one operating company. As an investor he can build an ecosystem of companies, stay closely involved with each CEO, and focus only on AI. On robots, Anderson notes that SoftBank has said it is mass-producing them at a physical AI plant where robots build robots. Son says the portfolio’s robotics companies are getting ready for jobs people do not enjoy.

17:20 What to Study and How Son Uses AI

At 16 Son talked his way in to see the head of McDonald’s Japan to ask what he should study. His answer to the same question today is AI, applied to whatever subject a person wants to specialize in. He says he cannot go a day without it. He treats it as a debating partner, asking for another angle or a 20% more efficient version.

19:46 The 70% Rule and Thinking 10 Years Ahead

Son’s advice to investors is to forget short-term, small money and think big. Nobody gets 100% of a dream easily, so the size of the dream sets the size of the result. He holds a daily meeting on the next 10 years and expects the plan to change each day. On his five-decade marriage he says relationships require respect and love. He would leave the Hawks lineup to the general manager.

23:28 The Debrief With Steve Schwarzman

Schwarzman says Son’s gift is seeing what a technology can do commercially in the right hands, and that he has had far more triumphs than setbacks. Schwarzman likes change, because without it there would be no Blackstone, and looks for enduring themes over short-term trades. On AI safety, which he funded at MIT in 2018 and Oxford in 2019, he says the issues are now front-page news and he is confident they will be addressed. He expects profound changes in healthcare and individualized education, and agrees with Son that big results take a full-scale effort.

Notable Quotes

“The biggest beneficiaries of AI in 2025 were dogs owned by software developers.”
Rodney Zemmel, on early AI gains that never reached the bottom line

“I’m on the side of not really slowing down. Technology development actually is the solution to those kinds of issues.”
Masayoshi Son, on AI safety and alignment concerns

“So, we have to go full speed with the good intent.”
Masayoshi Son, summing up his position on AI development

“Even one day I cannot live without it now.”
Masayoshi Son, on using AI as a thought partner for hours a day

“The result is proportionate to your dream.”
Masayoshi Son, explaining his 70% rule for thinking big

“If you don’t know what’s gonna happen in 10 years, don’t give up thinking about it.”
Masayoshi Son, on revising a 10-year plan every day

“So radical change, major change, won’t come as fast as some people think.”
Steve Schwarzman, in the debrief after the interview

Watch the full episode, including the conversation with Masayoshi Son and the debrief with Steve Schwarzman, on Inside Blackstone.