Towards the end of last year, Masayoshi Son spent a lot of time crying. The investment funds run by his company SoftBank would report an annual loss of $32bn in the 12 months to March 2023 (imagine burning $1,000 every second, day and night, for a year). But – as he told investors in Tokyo at a meeting in June of this year – what really upset the 66-year-old CEO was the idea that his best years could be behind him, that he might never realise his lifelong dream of becoming one of history’s greatest inventors. “The tears didn’t stop for days,” he told shareholders.
The good news was that Son had found solace in a new friend: ChatGPT, the OpenAI chatbot of which Son declared himself a “heavy user”. He claims to have produced more than 600 ideas for inventions with the help of ChatGPT. The software, he told investors, praises his ideas as “feasible and wonderful”. A team of staff waits around the clock to receive their billionaire employer’s bursts of creativity, which often occur before dawn.
This is normal for Son, who set himself the task of coming up with a new invention each day as a student. As a young entrepreneur, he slept for three hours a night, and in 1978 Son missed his own wedding, twice, because he was absorbed in the development of a handheld portable translator. As a new parent he spent his evenings and weekends in the office, or travelling through Japan as he established SoftBank. In Atsuo Inoue’s biography of Son, Aiming High (translated into English in 2021), these privations are reported not as the tragedy of a man who sacrificed his emotional and social life to his career but as evidence of his unique dedication to a higher purpose. “I am a genius,” he once told a fellow businessman with whom he sought to make a deal (he succeeded).
Son is renowned for his quirky presentations, which include Powerpoint slides forecasting 200-year lifespans and telepathy, and are illustrated with pictures of winged unicorns or geese laying golden eggs. But his presentation in June also carried a serious message: Son believes he can cash in, in a spectacular fashion, on the current boom in AI investment.
He plans to do so using a British company: the semiconductor designer Arm, which emerged from another British company, Acorn Computers, in the 1990s and is still headquartered in Cambridge. Son’s company, SoftBank, bought Arm in 2016 and on 21 August, SoftBank filed to begin selling shares in the company – not in London, as our government had long hoped, but on New York’s Nasdaq stock exchange.
The sums involved are ambitious even by the standards of Masayoshi Son. The company’s public filings reveal that it values Arm at more than $64bn – more than twice what it paid for the company – and SoftBank reportedly hopes to raise around $10bn from its initial public offering of securities.
Such paydays are the basis of Son’s huge wealth. He is one of the 200 richest businesspeople in the world, and was briefly its richest, thanks to an early investment in Yahoo, before the dotcom crash awarded him another record: the most money lost by a single person (an estimated $70bn on paper). Fortunately for Son, before the crash he had made another bet: a relatively small ($20m) investment in a new ecommerce website, Alibaba, which had been founded a year earlier by the Chinese entrepreneur Jack Ma. When Alibaba went public in 2014, SoftBank’s stake had grown to $60bn.
This spectacular rebound helped Son establish new relationships around the world. He met with Donald Trump shortly after his election victory in 2016 and agreed to invest in the United States. In 2017 he established the world’s largest private equity vehicle, the $98bn Vision Fund, in which the biggest investor was the sovereign wealth fund of Saudi Arabia. This is the truth of the technology boom of the past decade: many of the nifty new apps and services – Uber, Slack, DoorDash, Gopuff, Gympass, Klarna, Revolut – were backed by the old-fashioned oil money, through the Vision Fund, of governments that flog and behead their citizens.
[See also: How Saudi Arabia is buying the world]
In this frothy period, the biggest bubble of all was WeWork. In the company’s eccentric CEO, Adam Neumann, Son encountered a personality that made him seem restrained. Neumann claimed that with enough investment he would deliver growth so spectacular it seemed impossible: he would become a trillionaire, build a WeWork on Mars and live forever. Neumann would walk the offices of WeWork barefoot, dispensing shots of tequila (Don Julio 1942, at just under £200 a bottle), and schedule meetings for 2am. Son told Neumann he was “not crazy enough”.
Perhaps Son found something personally compelling in WeWork. His grandparents emigrated from Korea to Japan, where they lived in a shack they built by hand next to a railway line. Despite being born in Japan, Son grew up a Zainichi Korean, an outsider in his own country – he found it difficult to obtain a Japanese passport despite being a naturalised citizen – and began spending long periods in the US from the age of 16. The idea of a place that could be called home, anywhere in the world, may have had special significance for him.
After a visit to the WeWork headquarters in 2016 that lasted just 12 minutes, Son agreed – in the car back to the airport – to invest $3bn in the company. SoftBank would eventually invest at least $16bn in WeWork, in the hope that when the company went public, markets would be similarly captivated by its mission “to elevate the world’s consciousness”. As a private company, to spout this kind of nonsense about renting desks to small businesses is fine, but the thing about public markets is that they require companies to make things public. In WeWork’s case, the messianic vision very quickly took a back seat to the financial information it was required to publish, which revealed that in its headlong pursuit of scale it had amassed tens of billions of dollars in lease obligations without ever making a profit. The plan for a stock-market bonanza degenerated into an acrimonious battle between Neumann and Softbank for the company; Neumann was ultimately ousted though he kept a large chunk of money – and made even more when the company went public in 2021. WeWork recently warned there is “substantial doubt” that it will stay in business.
Other, smaller SoftBank investments have included Greensill Capital, which employed David Cameron as a lobbyist until it collapsed in 2021, and FTX, the bankrupt cryptocurrency exchange whose founder, Sam Bankman-Fried, faces charges including securities fraud and money laundering. Even Uber, once the most promising of SoftBank’s unicorns, exposed its Vision Fund to significant losses.
Arm seems like a much better bet: you are almost certainly using Arm’s technology now, on at least one of the 250 billion devices produced so far that use either its chip designs or its software. Of these, the most interesting to investors are those produced by another chip manufacturer, Nvidia, the leading producer of the specialised silicon on which the current generation of AI runs. Nvidia’s market capitalisation has almost trebled since the launch of ChatGPT last year, as investors pile into what many believe will be a transformative company. Son’s first idea was to sell Arm – which makes some of the architecture used in these chips – directly to Nvidia, which SoftBank also had a stake in at the time, to create a semiconductor supergiant. This sale was rejected by competition regulators.
But the geopolitical significance of semiconductors – which are subject to export controls from the US and UK – is a risk for Arm, which makes about a quarter of its revenue in China through a joint venture (the unfortunately named Arm China). And while the market for AI chips has boomed, Arm’s sales fell slightly last year, which might suggest it isn’t as integral to the machine intelligence boom as some believe.
It may be that even Nvidia isn’t quite the world-changing force that investors are hoping for. Son professes to believe that AI will make us “superhuman”, but real-world applications are scarce. What we do have, so far, is plenty of evidence for AI’s capacity to cheapen and degrade the work that was once done by people. Perhaps the inventions streaming from Son’s late-night ChatGPT sessions are the architecture of humanity’s future, but perhaps they’re manifestations of an inferiority complex – the flimsy hopes of a would-be inventor, asking his computer to tell him what he wants to hear.
[See also: The human era is ending]