SoftBank’s Masayoshi Son Looks for Investing Redemption

In theory, it shouldn’t matter whether Son makes his investments from the Vision Fund or not, since the team behind it still has about $9 billion left to invest from its second fund, all of it SoftBank’s own capital. But the past several years haven’t been kind to the Vision Fund’s reputation in startup and investing circles.

For Son, the next phase of SoftBank’s tech investing could offer a shot at redemption. The 66-year-old billionaire—whose fortunes have swung wildly over decades of investments in the tech industry—lately seems to be in a melancholy mood as he reflects on what he has achieved over his career. At a June shareholder meeting, Son said he had been thinking “about the rest of my time as a businessperson,” “felt ashamed that I made many mistakes, bad investments,” and had been filing hundreds of patents himself.

“How many years do I have? How many years [are] there for me? What have I done?” he said through an interpreter, adding: “Sometimes I felt a bit sad.”

It’s easy to understand his lack of cheer. Over the last six years, SoftBank spent more than $150 billion in outside money and its own capital investing in startups through two Vision Funds, without a lot to show for it.

Through June, the first Vision Fund had reported just a 14% return since it started in May 2017, a gain from both private and public companies that pales in comparison to the more than 80% return ordinary investors could have earned over that same period by putting money into the S&P 500. That fund, which has been depleted, invested an average of $950 million in each of the 94 startups in its portfolio—a staggering sum, many times the average check size for other growth equity funds.

Several of those startups flamed out in spectacular fashion after raising billions of dollars each from SoftBank, including WeWork, lender Greensill Capital and construction firm Katerra. One of SoftBank’s few big winners was still bittersweet: It booked a $3 billion gain on its bet on Nvidia stock, but sold it in 2019 well before the chipmaker’s recent stock run-up.

The second Vision Fund, started with SoftBank’s own cash a couple years later, has been an even bigger black eye. It was down about 36% through June since it started in 2019. While that fund reduced the average amount SoftBank plowed into each firm—an average of $190 million across 274 startups—its timing was horrible. Among its bets were once buzzy pandemic-era darlings, like celebrity video app Cameo and fintech Klarna, whose businesses later slumped.

One of Son’s biggest home runs, a turn-of-the-century investment in Alibaba, helped fund many of those investments by allowing SoftBank to borrow money using its stake in the Chinese e-commerce giant as collateral. But a regulatory crackdown in China sent Alibaba’s stock into a tailspin, diminishing the firepower it brought to the second Vision Fund.

After Alibaba’s stock lost more than half of its value in 2021, SoftBank spent only about $4 billion on startups between April 2022 and June 2023, filings show. That compared to about $50 billion during the 15 months prior. “There was a lot of, ‘How busy are we going to be in the next two months? Look at Alibaba share price as a leading indicator,’” said Sukhvinder Gill, a former vice president of investing at the Vision Fund, who left in January.

In response to the downturn in tech stocks, Son has spent the past few years cleaning house. SoftBank had multiple rounds of layoffs. Virtually all the high-powered investors and tech executives he brought in to run the company have left or stepped back. Meanwhile, Son has made moves to cement his own power. In 2021, he appointed the firm’s top outside lawyer to its board of directors, angering shareholders, who felt that not enough board members would stand up to Son.

The investment professionals who remain at the Vision Fund have taken a much more cautious approach to investing. For instance, this spring they passed on investing in AI startups like Anthropic and Cohere because of valuation concerns, according to a person with direct knowledge of the matter. Vision Fund investors have sat out the burgeoning sector in part because the fund’s leadership hasn’t given them a clear directive, which may be because they’re still licking their wounds from past years when they invested in tech companies at sky-high prices, according to two people close to the Vision Fund.

Son, though, has become increasingly vocal in his AI boosterism, telling a Tokyo audience at SoftBank’s annual conference last week that they will become like confused goldfish if they don’t embrace the technology. In June, he declared that SoftBank would “shift to offense mode” in investing again.

He now could amass more ammo at his disposal after chip designer Arm went public last month. SoftBank still owns 90% of the firm, which has a $55 billion market capitalization, and could take out additional loans against the stock. Earlier this year, SoftBank also liquidated its entire stake in Alibaba, fetching $7.2 billion in the sale. David Gibson, a senior research analyst at MST Marquee who covers SoftBank, estimated the conglomerate could have $20 billion to $30 billion to spend over the next few years.

“What everyone will be concerned about is: Has Masa really learned his lessons from mistakes made in Vision Funds 1 and 2?” Gibson said, using Son’s nickname.

‘Strategic Masa’

Son has long cultivated relationships with powerful figures in tech at key moments in the industry’s history. A year before Apple launched the iPhone in 2007, Son’s broadband and telecommunications company elbowed its way into the wireless business by spending more than $15 billion buying the Japanese unit of Vodafone in 2006. SoftBank later got exclusive carrier rights for the iPhone in Japan when the device launched, which Son credited to an earlier meeting with then-Apple chief executive Steve Jobs.

That agreement sent SoftBank shares soaring, emboldening Son at a moment when the tech industry was undergoing radical change due to the mobile internet. A few years later, Son stood on stage in Tokyo and announced that SoftBank would become a broader strategic investor, “where we hold stakes in 20% to 40% of the companies.”

Some people close to SoftBank see echoes of the past in Son’s recent moves to try to get close to OpenAI, the startup powerhouse whose development of large language models has stirred excitement among technologists about the possibility of a new computing paradigm. So far, SoftBank hasn’t put money into the firm, which has raised money from tech luminaries like Microsoft, Khosla Ventures, Sequoia Capital, Thrive Capital and Andreessen Horowitz.

But Son is still getting close to the action. As The Information first reported, he has participated in discussions with Altman and Ive, who played a key role in designing the iPhone and other smash Apple products, about a device that would feature AI at its core. The Financial Times later reported that OpenAI is in advanced talks to launch a new venture to build the device, with over $1 billion in SoftBank funding.

“It’s strategic Masa,” said a person close to SoftBank. “He’s not [just] trying to place a bet on Uber or WeWork, but trying to be involved in business creation.”

At the same time, Son is taking financial swings in other categories that depart from traditional venture capital deals.

In March, SoftBank paid $375 million in cash to buy Berkshire Grey, a decade-old firm that sells robots to e-commerce warehouse owners like Walmart and Target. The deal amounted to a bailout of sorts, as Berkshire Grey was losing so much money that auditors raised concerns about its ability to continue operating.

And in a complicated deal announced in July, the Japanese conglomerate took a 65% stake in GreenBox Systems, a new company focused on warehouse automation. The firm spun out of Symbotic, a logistics company SoftBank had already taken public through a special purpose acquisition company.

A couple months later, SoftBank became the sole backer of Stack AV, an autonomous trucking startup founded by longtime AI and automotive entrepreneurs. In press releases, SoftBank Group, rather than the Vision Fund, was named as the investor in the GreenBox and Stack AV deals.

Investors in SoftBank Group, which is listed on the Tokyo Stock Exchange, don’t appear to have been won over by these deals or by Son’s claims that SoftBank is returning to dealmaking. SoftBank’s share price is up 8% since the start of the year, compared to a 20% gain for Japan’s main stock index.

“Officially, they’re saying they’re taking a much more careful, much more measured approach,” Gibson said.

But, he added, some of its recent investments, like the one in Stack AV, are even earlier stage than previous Vision Fund deals in pre–initial public offering companies. “What they’re doing so far looks more risky than Vision Fund 2,” Gibson said.

Differing Visions

As Son increasingly focuses on investing through other parts of SoftBank, he has left the Vision Fund to its own devices, frustrating employees, according to four people close to the fund.

Over the past year, he hasn’t been involved in the fund’s day-to-day work, a sharp shift from the peak of the bull market in 2020 and 2021, when he held deal meetings with Vision Fund partners multiple times a week, people close to the fund said. The fund is officially led by two co-CEOs: Rajeev Misra, a former Deutsche Bank credit trader, and Alex Clavel, a former Morgan Stanley investment banker. But Misra largely stepped back from his duties last year, and recently raised $6.8 billion for his own separate firm, One Investment Management.

Son has entrusted longtime Vision Fund investor Vikas Parekh, who was heavily involved in its 2019 WeWork rescue financing, to work on its AI and logistics strategies. One person who worked closely with Parekh a few years ago described him as a level-headed member of what was then a chaotic and factitious firm.

The shift has also cut some red tape for Vision Fund investors. They can invest up to $300 million without asking permission from Son or other Tokyo-based executives, a person with direct knowledge of the matter said.

However, that has created a leadership vacuum. In recent months, Vision Fund investment partners have debated how aggressive to be in their investing. Few strong late-stage startups—the bread and butter of the Vision Funds—have been raising money. AI startups are raising at high prices. But some at the Vision Fund are frustrated with what they view as a missed opportunity: There are many startups raising money at a discount to their last round, but SoftBank hasn’t pounced on those deals.

In addition to passing on Anthropic and Cohere, the Vision Fund earlier this year strongly considered bidding on investment deals for Stripe and Databricks—which raised two of the biggest funding rounds among pre-IPO startups this year—but decided they also were too expensive.

The Vision Fund hasn’t made many bets on the generative AI boom, although that could soon change. It is currently weighing an investment in Lambda Labs, a startup that rents servers with Nvidia chips to other companies, according to a person familiar with the matter. Lambda is raising $300 million at a $1.5 billion valuation, including the investment. SoftBank also recently considered an investment in CoreWeave, a Lambda rival, which is in the process of raising $500 million in a sale of secondary shares that values the company at $7 billion.

The firm made at least one AI investment this year—in Tractable, which uses AI and computer vision to determine the condition of cars and homes. SoftBank led a $65 million investment in the company in July.

That deal isn’t part of the buzzy wave of generative AI companies most venture firms are chasing. SoftBank also is finalizing an investment in TravelPerk, a Barcelona-based company that develops software for managing corporate travel, that values the business at $1.3 billion, according to two people familiar with the deal.

“The Vision Fund now resembles the type of fund you would see elsewhere,” said Gill, the former SoftBank investor. "He said while underwriting assumptions were “too optimistic” before, “that’s gone out the window now. It’s a lot more selective.”

But, he added, it is clear that Son marches to his own beat. “Masa is an enigma—don’t get me wrong,” he said.