TOKYO -- Bill Huang likes to think in hundred-year cycles. A former telecoms engineer, Huang founded CloudMinds in 2015, driven by the belief that the unstoppable rise of artificial intelligence and the widespread adoption of robotics will outpace the capabilities of computer chips. His solution: to build a "cloud brain" that connects millions of robots in homes, offices and factories with the vast computing power of remote servers.
CloudMinds' pitch, with its huge ideas, long-term vision and its worldview centered on the rapid evolution of artificial intelligence, was almost guaranteed to appeal to Masayoshi Son, founder and CEO of SoftBank Group. Overseer of the Vision Fund, the world's largest technology investment fund, Son has famously talked about his desire to invest based on a 300-year business plan, and of his belief in the imminence of "the Singularity": the point at which artificial intelligence outstrips human intelligence.
The Vision Fund has become CloudMinds' largest investor, with a 34.6% stake. Huang's company joins a $70 billion portfolio that spans automated pizza delivery, an "Uber for dog walkers," ride-hailing giant Uber Technologies itself and five of the world's 10 most highly valued unicorns (private companies worth $1 billion or more). In just three years, the Vision Fund has become one of the most powerful players in global technology, making headlines with a "winner takes all" approach to investing: It throws huge sums of money at its portfolio companies to allow them to outgun their competition, then keeps them private until they can be listed in outsize initial public offerings.
However, for all of Son's talk of century-long horizons and the evolution of intelligence, the Vision Fund's future hinges on time scales measured in weeks and months. With markets already nervous of an impending global slowdown, SoftBank has to get several major listings away, including that of CloudMinds, before the end of the year -- because its giant flagship fund needs to make money, fast.
To fill up the coffers of a new, even larger fund, which is due to start investing in a matter of months, SoftBank needs to realize billions of dollars in profits from its first round of investments. From then on, the fund must run on a treadmill of relentless exits and IPOs -- Son expects five or six in total this fiscal year, then 10 or more per fiscal year thereafter -- in order to meet the commitments it has made to its investors. It will test the limits of "Masa" Son's legend as an investor, as, for the first time, the fund's success is measured not by the size of the investments it puts in, but by how much it is able to take out.
"Deploying capital is the easy part," said Claudia Zeisberger, a professor at Insead. "The true skill is to help the companies grow or even scale and ultimately exit again. Until we see exits, the verdict is still out whether [the Vision Fund's] model is working."
Prodigal Son
Son's entrepreneurial drive started early. As a college student at the University of California, Berkeley, in the late 1970s, he sold everything from translation machines to arcade games, hiring his peers to help out.
Hong Liang Lu, a contemporary who graduated from Berkeley's civil engineering department, became a key lieutenant of Son's. "I was reluctant at first, because [selling arcade games] was not a reputable business in the U.S.," he recalled. "I was mistaken for a Chinese mafioso." But he stuck with Son, lured by his drive and ambition. "He started a business in college at a time when there were no venture capital or startups. He had no role model to follow."
When the two went to Las Vegas in the early 1980s, Lu recalled that Son refused to gamble, because the chances of winning were less than 50%. "He does not make a losing bet," Lu said.
The relationship paid off. Lu launched his own telecoms business in 1991. Four years later, he pitched it to Son, who invested straight away.
"He started a business in college at a time when there were no venture capital or startups. He had no role model to follow"
After graduation, Son returned to Japan, where in 1981 he founded SoftBank as a distributor of PC software. The company listed in 1994 and rode the dot-com boom up and back down again. In 2000, the year that the bubble burst, Son made a deal that turned into one of the most successful bets in modern investing: a $20 million investment in an up-and-coming Chinese online retailer, Alibaba Group Holding. SoftBank's 26% stake in the company, now one of the world's largest tech companies, is worth more than $110 billion.
Son's investment in Alibaba has become a foundational element of his legend: He got in early, took a large stake, and then held onto it until the valuation was enormous. SoftBank waited 16 years before it sold a single share in Alibaba.
Son was still riding on this reputation when he unveiled the Vision Fund in late 2016. The fund launched with an extraordinary $93 billion in funding, including a $45 billion equity investment from Saudi Arabia's Public Investment Fund, and a further $15 billion from Abu Dhabi's sovereign wealth fund Mubadala Investment. SoftBank itself put in $28 billion, including a 25% stake in Arm Holdings valued at $8.2 billion.
Those funds would be invested in a very different strategy from the big, directional bet that Son made on Alibaba. Rather than capturing an early stake and holding on, the Vision Fund generally invests at least $100 million in late-stage, private tech companies that are close to going public.
The Vision Fund is often treated as a giant, buccaneering venture capital fund, whose investors are trading big risks for an opportunity to get in on the ground floor of the next Alibaba. But around $40 billion of the capital raised by SoftBank was in the form of "preferred equity," which gives a bond-like fixed return at a rate of 7% per year.
A typical venture capital fund with a 10-year investment period can go five years without returning any money to investors. But the Vision Fund paid out nearly $1 billion in such distributions for the year ended in March 2019, and more than $600 million in the April-June quarter. Sources say the unconventional structure was designed in response to investor demands to reduce downside risk, suggesting that there was some hesitancy among Son's backers to fully commit to his vision.
So far, the fund has had mixed success in exiting its investments. In May, Son announced that the fund had achieved a 29% blended internal rate of return -- a key measure of a fund's performance. The figure would be considered successful by industry standards, and rises further to 45% if the 7% fixed returns are stripped out. However, this was mostly thanks to unrealized valuation gains. Two exits -- the sale of U.S. chipmaker Nvidia and Indian e-commerce company Flipkart -- also contributed, but both are considered outliers to the fund's core strategy of taking private companies public. Nvidia was already publicly trading when SoftBank invested, and Flipkart was sold to Walmart.
And when the fund reaped a handsome return from Nvidia, despite a sharp fall in share price in late 2018, part of the credit went to Son's adeptness in financial engineering. The fund used margin loans -- borrowing with Nvidia shares as collateral -- to leverage its investment, and capped the downside using derivatives, in a technique rarely used in venture capital.
The Vision Fund has yet to cash out of five other companies that have gone public, and their performance has been varied. Some have performed well. Guardant Health, a U.S. health care company that aims to detect cancer using AI, has surged nearly nine times the price that the Vision Fund is estimated to have paid. The valuation of Slack, the workplace messaging app, has also risen substantially.
Others have been less-than-stellar performers. Shares of Hong Kong-listed ZhongAn Online P&C Insurance trade at less than a third of the Vision Fund's estimated purchase price before ZhongAn's IPO in 2017. Uber, which went public in May, is trading roughly equal to the price at which the Vision Fund bought in.
The Vision Fund needs to continue to attract high valuations for its companies ahead of their listings. However, the market is not giving SoftBank an easy ride.
CloudMinds, which plans to raise $500 million in an IPO in New York, issued its prospectus in July and intended to list in August. That date was pushed back, as investors struggled to understand the business model, and as a jittery global stock market created an uncertain environment for an IPO, according to sources with knowledge of the company's plans.
Analysts have also questioned the extent to which CloudMinds' business truly reflects the Vision Fund's stated ambitions. A large portion of CloudMinds' revenue, for example, came from a manufacturing contract for smartphones and optic detectors -- a far cry from the "end-to-end cloud AI operating system" on which it sells its long-term value. CloudMinds declined to comment.
Another of the Vision Fund's portfolio companies, The We Company -- commonly known as WeWork -- is also moving toward an IPO in the fall. But it faces questions over its bloated $47 billion valuation, its unorthodox corporate structure, and the fact that it has never come close to making a profit.
The skepticism over WeWork's worth has also exposed another danger of the Vision Fund's reliance on buying and listing late-stage enterprises. Analysts have been able to draw parallels between the fund's portfolio companies and comparable listed rivals. In WeWork's case, critics have pointed to the fact that IWG, a listed company with a broadly similar business model, is valued at one-tenth of WeWork's asking price.
Son, however, has shrugged off the doubters, and is now doubling down with an even larger successor to the Vision Fund -- one that could put SoftBank in a position of unprecedented power to shape the future of the global tech industry. As Son himself has said, the combined firepower of the two funds is "comparable to the total investment by Silicon Valley venture capital since 1995."
Cooling interest
Whether investors will buy into the vision a second time around remains to be seen. "A lot of big pension funds and institutional investors are still sitting on the sidelines," said one investment banker.
On July 26, SoftBank announced that it had secured contributions of at least $108 billion, with a list of investors that included Apple, Microsoft, Hon Hai Precision Industry -- trading as Foxconn Technology Group -- and the National Investment Corp. of the National Bank of Kazakhstan, the central Asian country's sovereign wealth fund. But two notable names were missing: the sovereign wealth funds of Saudi Arabia and Abu Dhabi, the largest investors in the original vehicle. Sources say negotiations with both sovereign wealth funds are ongoing.
Details of the second fund's structure are scarce, but Son has hinted that there may be different classes of shares for existing Fund 1 investors and first-time participants.
There are also questions over why some companies have invested. For example, several banks, including Standard Chartered and several large Japanese institutions, have committed money; something some analysts have interpreted as a cynical move to gain access to the inevitable pipeline of IPOs. SoftBank already has a reputation as something of a whale in the investment banking world, paying $894 million in fees in 2018, according to the Financial Times.
"A $5 billion or $10 billion investment is meaningful," said Chris Lane, an analyst at Sanford C. Bernstein. "But it is possible that [an investor] is putting in a nominal amount just to enhance its chances of becoming a preferred banker in the IPOs."
SoftBank has also pledged $38 billion of its own capital -- an additional commitment larger than all of the cash that SoftBank holds on its balance sheet. In the past, the company has pulled off huge acquisitions by borrowing heavily, but Son has set a target of keeping its loan-to-value ratio -- net debt divided by the total value of stakes it owns -- below 25%. The figure stood at 19% in June, when SoftBank reported more than $130 billion in debt. However, the company has argued that some $75 billion in debt held by Sprint and SoftBank Corp. are "non-recourse," which means SoftBank is not liable when the subsidiaries default. It says net debt at the parent level is about $46 billion.
SMBC Nikko Securities estimates that SoftBank needs to sell about $28 billion in assets over the next four to five years to maintain the 25% threshold. That time frame may shorten if the second fund invests as aggressively as the first.
Successfully selling the investments at a profit will bring huge windfalls for SoftBank. The company does not own any preferred shares in Fund 1, meaning that it reaps about half of the equity gains. Like other venture capital funds, SoftBank additionally keeps 20% of returns above a certain threshold as carried interest.
Son has been clear that it is the Vision Fund, not the legacy telecoms business, that is, ultimately, the future of SoftBank. He plans to institutionalize the structure of the funds so that they can operate after he hands over to a successor. That includes expanding the current 450-strong team of investment professionals at the fund to 2,000 within 10 years. Son, 62, has said that he will step down at some point in his 60s.
Son has also said that Vision Fund 2 is not the end of his ambition, and that he intends to launch a new fund every few years. If each fund is as big as the last, Son and SoftBank will need to prove their model at an unprecedented scale over and over again -- and they could find themselves victims of their own success.
"The importance of a $100-plus billion fund, in an industry [AI] that only received $35 billion a year in funding a few years ago, is immense," said Arturo Bris, professor of finance at IMD business school. "Promising startups, which could not receive funding because of a constrained supply of funds, will now see the light."
However, as more money is drawn into emerging technologies in the wake of the giant Vision Funds, and valuations swell, the chances of a bubble are rising -- something Son, who would have lost big in the last crash without his windfall from Alibaba, must be cognizant of.
"This resembles the pre-2000 internet bubble, where money was looking for anything with the suffix 'dot-com,'" Bris said. "Investors poured money into flashy applications of technology that lacked a business model. Ultimately, there may be more money than ideas."
Nikkei staff writer Kentaro Iwamoto in Singapore contributed to this report.