When I started developing software for the management consulting firms I worked for in the mid-1990s, we still had to connect to the World Wide Web on slow and clunky lines to access our coding work. The laptops we carried weighed at least five kilogrammes and couldn’t be used for actual programming. One of our clients was a matchmaking company, which operated a number of neighbourhood storefronts where lonely hearts could flick through albums with the pictures and profiles of potential partners and leave messages in numbered pigeonholes. The firm had hired us to ‘automate’ both their core business and their back-of-house customer service functions. Our job was to set up electronic kiosks in the stores, so that punters could watch videos of and send electronic messages on the internal company network to people they fancied. We also designed a spartan database to keep track of customers, subscription payments and dating history. Our long-term plan was to move the business ‘online’ so that customers could dial up into the ‘business-to-consumer’ (B2C) interface and browse virtual ‘face books’ from the privacy of their own home.
This was the era of what came to be termed Web 1.0. In the aftermath of the Cold War, many of the tech stars who used to graduate from MIT and Carnegie Mellon and enter the US war machine were in California working in the private sector. Personal computers moved out of institutions and into homes and the software boom began. ‘Dot-coms’ began producing goods and services to sell to other businesses and to consumers. Business-to-business (B2B) products included everything from servers and routers to customer service, advertising and marketing databases.
To build the customer service software for our matchmaking clients, we needed a database program from Oracle and server management software from a company called Scopus Technology. We had to work in the Scopus building in the San Francisco Bay Area to adapt their software to the needs of our client. One day I was summoned to the offices of the two Iranian-American brothers who owned the company and offered a massive pay package and stock options if I joined Scopus. Technology companies were desperate for software developers. Since I was about to move to New York to go to graduate school, I turned them down. A couple of years later Scopus was acquired by Siebel, which was itself later acquired by Oracle. If I had stayed, I could have cashed out those stock options for a lot of money.
In the 1990s, huge amounts of money and debt were sloshing around in the software business. If a company ‘founder’ claimed to be doing something new (the homo economicus discourse of ‘disruption’ came later), investors were very likely to buy in. Venture capital firms hired cocaine-fuelled young MBAs to build up their operations. Many of the venture capital firms were leveraged up to their eyeballs because interest rates were at a historic low and because, in the US, the Taxpayer Relief Act (1997) had lowered the marginal rates on capital gains. Smaller companies were acquired at inflated prices by bigger companies. Many of the men (they were almost all men) perceived to be business and technology visionaries were lucky beneficiaries of the end of the Cold War and the public funding of technological innovation.
One of these men was Masayoshi Son, founder of SoftBank. Between 1945, when the US’s seven-year occupation of the country began, and 1973, when oil prices, until then controlled by US and British cartels, were suddenly opened to the market, Japan was the world’s fastest-growing economy. This growth depended on the Pax Americana’s guarantee of cheap oil, stable dollars and US military protection. Japan provided a base for the US in the Pacific, allowing it to fight its wars in Korea and Vietnam, and pacify insurgencies in the Philippines, Laos, Thailand and beyond. Some of the biggest US military bases are still in Okinawa, control of which didn’t revert to Japan until 1971. During the Korean War, spending by the US military doubled Japan’s industrial output. The Vietnam War opened up the US market to Japanese electronics and cars, transported on container ships which had delivered materiel to Vietnam and now needed a cargo to take back across the Pacific. By 1980, Japan was the biggest producer of cars in the world, and by 1991 of steel.
At around this time, Japan’s Ministry of International Trade and Industry began co-ordinating and rationalising industrial production, upgrading equipment and factories, and encouraging new products: transistors, computers, petrochemicals and synthetic fibres as well as cars. Plentiful cheap labour, corporatist employment laws and close relations between banks, suppliers and producers meant that economic policies had rapid effects. Centralised industrial planning encouraged the standardisation and automation of production as well as innovations in logistics. From the 1970s until the early 1990s, Japan’s share of the US trade deficit was a third. This was seen by chauvinists in the US as an attack on their God-fearing country by disciplined, inscrutable Asians.
As the first major wave of Japanese electronics and cars arrived in the US, so did 16-year-old Son, the precocious child of a Korean-Japanese pachinko parlour tycoon – pachinko is an arcade game. His six-year stay in California was funded by his father, who granted Son permission to make the journey on condition that he accepted financial support from home. Son spent his time coming up with ideas for businesses and making connections. According to Lionel Barber in Gambling Man, Son claims that one of his first inventions was a portable electronic translator intended to be used by international travellers, though a former business partner told Barber that Son didn’t have the technical knowledge to design such a device. ‘There were a lot of people talking about electronic translators at the time. The beauty of Son’s idea was to sell it in [airport] kiosks.’
Son returned to Japan in 1980, and a year later launched a software distribution company called SoftBank Japan to take advantage of the market for word processors, spreadsheets and other programs designed for personal computers. He set up a publishing company to bring out computer and technology magazines, and put on trade shows, in effect owning both the products and the vehicles for marketing them. But Son had larger ambitions, and in 1995 purchased Comdex, which operated the world’s biggest technology trade shows, for $842 million, leaving its previous owner, the Las Vegas godfather Sheldon Adelson, to expand his gambling empire. Adelson was so grateful for the above-market-price sale that decades later he introduced Son to Donald Trump, and lent him his yacht so that he could go snorkelling with Saudi royals.
The same year, Son added the computer publishing house Ziff to his portfolio, at a cost of $2.1 billion. Barber – unimpressed – points out that this was ‘not as much as Sony paid for Columbia Pictures ($3.4 billion)’. With Ziff, Son had bought a stake in Yahoo! and in 1996 he launched Yahoo! Japan. He bought Vodafone Japan for $17 billion, $12 billion of it in cash, in what was then Asia’s largest leveraged buy-out ever. To secure the debt, his finance people arm-twisted ratings agencies to improve SoftBank’s rating from BB to BBB and secured a loan on very favourable terms: any future cash flow would pay off the interest rate on the debt before dividend payouts had to be made. SoftBank then boosted the number of Vodafone customers by offering subscribers to Son’s Yahoo! Broadband and SoftBank Telecom businesses flashy mobile handsets and undercutting competitors with pricing plans so outrageous that the company was censured by the Japanese Fair Trade Commission. The debt was made good in 2008, when Son won the Japanese distribution rights for the first iPhone from Steve Jobs himself.
Son benefited from the Reagan-era financial engineering that was meant to slow the flood of Japanese exports which had brought US manufacturing, especially the automobile sector, to its knees. In 1985, Reagan’s Treasury secretary, James Baker, negotiated the Plaza Accord, which depreciated the US dollar in relation to the mark, the franc, sterling and the yen. Although Japan acquiesced to the agreement, its government brought in a stimulus to counteract the effect of the accord; the injection of cash, in turn, led to an asset price bubble in the late 1980s. When, in 1989, property prices inevitably crashed, the Nikkei price index followed. The policies that resulted from this, especially the liberalisation of banking, and low or negative interest rates made it easier for new players such as Son to enter the big business leagues and put together enormous syndicated loan packages. Helpfully, the appreciation of the yen against the dollar also meant that Son could buy US technology companies more cheaply.
SoftBank’s most successful investment was not in Silicon Valley, however, but in China. The Chinese e-commerce platform Alibaba was set up by Jack Ma in 1999, and quickly became an online space for B2B, B2C and consumer-to-consumer (C2C) marketing and sales. SoftBank spent just $20 million to acquire 30 per cent of Alibaba’s ownership. When the company was listed on the New York Stock Exchange in 2014, the flotation was the biggest ever in the world: Alibaba’s market capitalisation was valued at $230 billion, or as the Financial Times reported, ‘larger than fellow tech giants Facebook and Amazon, and big US companies such as J.P. Morgan and Procter & Gamble’. SoftBank’s $20 million investment was now worth $70 billion – a 350,000 per cent gain. Because Alibaba was so valuable, Son could sell its shares as needed to shore up SoftBank’s finances.
Son had ‘smelled the money’ in China because Japan had been investing at a vast scale in enterprises there ever since Deng Xiaoping’s ‘opening’ of the Chinese economy in 1978: it was the place to look to for opportunities. The timing of Alibaba’s founding had also been propitious. China had liberalised private ownership and foreign direct investment in the early 1990s and, as poverty rates fell, markets for consumer products expanded. With China’s accession to the World Trade Organisation in 2001, Alibaba was in a position to export Chinese products, taking advantage of extensive state investment in both domestic and transnational logistical infrastructure.
Not everything went Son’s way. The bursting of the dotcom bubble in 2000 and the global financial crash of 2008 posed existential crises for SoftBank. In the mid-1990s, the firm had invested nearly $5 billion in more than fifty technology companies in Silicon Valley, many of them loss-making and soon to be forgotten. In 2000, SoftBank’s value fell to one twentieth of its value the previous year, and Son himself lost 96 per cent of his paper wealth. To raise money, he sold shares in his successful ventures, especially Alibaba. After the 2008 crash, SoftBank was exposed to $750 million of Goldman Sachs’s collateralised debt obligations; more significantly, its acquisition of Vodafone Japan two years earlier had meant taking on $23.8 billion of debt. The company’s subsequent recovery was aided at least in part by the flow of cheap money unleashed by stimulus packages in both the US and China. The people who had lost their homes or savings were not so lucky.
From the start, SoftBank had employed financial wizards, many of them former bankers, most of them US-educated, who put together increasingly complex and opaque loan packages that blurred the line between Son’s personal assets and debts and SoftBank’s corporate holdings and leverage. The company became a conglomerate with evolving functions, from publishing to telecommunications infrastructure to technology distribution, and eventually to venture capital. Son packed the board with loyal friends and decided the direction of acquisitions and investments almost single-handedly, often inspired only by intuition.
Son saw himself as a visionary architect of new technologies. Barber, for many years the editor of the Financial Times, thinks Son is interested not in turning profits but in growing his business and acquiring ever more novel tech companies. His self-aggrandising ‘mine is bigger than yours’ attitude impressed both Trump and Mohammed bin Salman. Barber recounts a meeting between Son and the soon-to-be US president at which Trump showed Son how to replicate his signature combover. Son and bin Salman shared fantasies of a high-tech, automated future: bin Salman committed $45 billion of Saudi Arabia’s sovereign wealth fund to Son’s venture capital operation, the Vision Fund. Son also raised money from Abu Dhabi by enticing Crown Prince Mohammed bin Zayed with the promise of ‘a new world of remote, round-the-clock surveillance’. When the authorities in charge of the Saudi investment fund asked for more information about SoftBank’s figures, and the nature of the relationship between SoftBank and the Vision Fund, Son’s executives shut them down: ‘You are a limited partner. You are not the driver, you are in the back seat of the car.’
In 2019, SoftBank launched Vision Fund 2, this time attempting to raise money from Apple, Microsoft and FoxConn (all of which promised money, but never paid up). SoftBank’s first fund had percentage returns in single figures, far below the riches promised by Son; its second posted only losses. Many of the investments made by both funds were spectacular failures. Meanwhile, to keep SoftBank afloat, Son accepted cash from investors like Paul Singer of Elliott Management, better known as the vulture capitalist preying on Argentina’s unpaid sovereign debt, and the supply-chain financing tycoon Lex Greensill, who hired David Cameron to lobby the UK government for business and was later disgraced and bankrupted.
The biggest Vision Fund bet was WeWork – essentially a real estate company that leased commercial buildings in big cities, fixed them up with communal workspaces, cafés and bars, and rented them out at a profit. But it had a charismatic CEO, Adam Neumann, an Israeli who spoke of WeWork as a capitalist kibbutz and threw raves for his clients. Son offered to invest $4 billion, and when other SoftBank executives expressed doubts, insisted that WeWork was going to be his next Alibaba. When news of the investment emerged in 2017, WeWork’s projected value shot up to $20 billion. Goldman Sachs, which hoped to secure WeWork’s lucrative IPO, flattered Neumann with a much higher valuation of $96 billion – more than Goldman Sachs itself – and enticed him with the promise of a ‘path to $1 trillion’.
WeWork wasn’t worth anything like that much. As an article in the Financial Times in 2018 pointed out, its biggest competitor, IWG, which had ten times as many properties, was valued at only $4 billion. The tech-bro hype surrounding WeWork, as with most SoftBank investments, pumped the value of the stock based on very few fundamentals and a lot of hot air. When, in 2019, WeWork’s IPO was dramatically derailed by Neumann’s erratic behaviour (he insisted on kabbalistic ‘energy cleansing’ sessions for executives and alarmed flight crew on a borrowed Gulfstream jet by packing large amounts of marijuana in cereal boxes) as well as by more serious concerns over company governance, the schadenfreude was deafening. ‘“Charismatic CEO” is a term that should strike fear into any investor’s heart,’ one analyst said. WeWork’s failure cost SoftBank $11.5 billion in equity losses, with another $2.2 billion in debt outstanding.
Son was an early supporter of gig economy apps and the latest developments in AI, with the Vision Fund 2 making significant investments in the field. Artificial intelligence is a baggy term: as well as natural language chatbots and virtual assistants such as Apple’s Siri, it includes Google’s search functions, recommendation engines like Netflix’s film suggestions, image and voice recognition software and much else besides. Although people had used most of these applications without fuss for years, AI puffery escalated when OpenAI released ChatGPT to the world in 2022. That ChatGPT hallucinated (i.e. made things up), that its responses were only as good as the material on which it had been trained, and that OpenAI had used copyrighted material without acknowledgment or recompense, seemed not to matter to investors or to many users. Like many other recent technologies (data mining, cryptocurrencies, blockchains and NFTs), the publicity around generative AI was used to inflate the value of companies working on anything related to it. (Grok, the rapidly and shoddily developed AI on Elon Musk’s X, has allowed his company XAI to reach a reported value of $75 billion.)
Like preceding booms, particularly in cryptocurrencies, the new AI technologies relied on a vast amount of computing power, which in turn demanded vast amounts of electricity and water. Within months of ChatGPT providing plagiarism aids to sleep-deprived students everywhere, US tech companies were investing in ever more numerous and enormous data centres. Some made deals with energy companies to bring coal-burning and nuclear power plants – including Three Mile Island – out of mothballs. The California-based AI chip-maker Nvidia surpassed $1 trillion in value in 2023. Son, who had owned and then sold shares in Nvidia before the surge in its price, must have kicked himself.
The Silicon Valley companies that spend the most on AI are Meta, Microsoft, Amazon, Google and OpenAI, all of which hoard chips and energy resources. The massive capital investment in AI infrastructure has come with eye-wateringly high stock valuations, and management consultants at McKinsey are predicting it to lead to an increase in US GDP of between 5 per cent and 13 per cent by 2040 (more sceptical accounts predict only a 1 per cent rise). Pessimists foresee significant job losses in ‘knowledge’ sectors such as medicine, software development, accounting – the list is long. Sam Altman of OpenAI has called for a ‘change to the social contract’. Larry Ellison of Oracle has touted a more sinister use of AI: ‘Citizens will be on their best behaviour because we are constantly recording and reporting everything that’s going on
One of the final acts of the Biden White House, on 14 January, was to issue an executive order: ‘Advancing US Leadership in Artificial Intelligence Infrastructure’. The order directed the departments of defence and energy to lease federal land to AI companies to enable them to build new data centres and secure cheaper energy. An executive order issued by Trump on 23 January embraced not only AI but crypto, and called for the US to assume global leadership in the development of AI, ‘free from ideological bias or engineered social agendas’.
Two days earlier, Trump had launched Stargate, a private joint venture between SoftBank, OpenAI, Oracle, Nvidia and Arm (a British chip-maker 90 per cent owned by SoftBank). Although most journalists focused on the Silicon Valley big beasts, the Stargate consortium also includes an Emirati AI fund, MGX, which only months earlier had launched a partnership with Microsoft and BlackRock, the world’s biggest asset management firm, to build data centres and power plants. Stargate will invest $100 billion in AI infrastructures immediately, and $500 billion by 2029. The announcement not only boosted the value of the companies involved in Stargate, but also the stock of energy companies.
Four days after Stargate was announced, DeepSeek’s R1 chatbot was launched, the ‘side project’ of a Chinese hedge fund owner, Liang Wenfeng. Unlike the Silicon Valley versions, it’s open-source and seemingly much cheaper to develop (DeepSeek claimed $5.5 million, though that figure has been disputed). Liang was quoted as saying that ‘more investment doesn’t necessarily result in more innovation,’ and DeepSeek’s results are as good or better than those of other models. On 27 January, Nasdaq dropped by 3 per cent. Bloomberg reported that the world’s five hundred richest people, Nvidia’s CEO among them, lost $108 billion. A number of energy companies saw their share price fall by a quarter. Nvidia alone lost $589 billion in market capitalisation. In all, tech stock lost $1 trillion that day. Son’s response was to fall back on fearmongering: ‘Depending on the country of origin, very dangerous situations can occur if we use this technology incorrectly, and this could be a trigger for a very bad situation for humanity.’
The protagonist of Dostoevsky’s novel The Gambler stakes his life and the woman he loves on just one more game, one more throw of the dice, driven by ‘self-intoxication, by your own fantasy
Like most other biographers of tycoons, Barber makes this a rags-to-riches tale, stressing Son’s Korean lineage and his early outsider status. He doesn’t ignore Son’s hubris, self-regard, recklessness and ego, though it’s clear he is charmed by his chutzpah, self-confidence and ability to bounce back again and again. But there is something more mundane – and far more dangerous – at work in Son’s irresistible rise. Like other billionaires, he can afford to gamble thanks to cheap money, unconstrained corporate power and the ascendancy of a technopolitics that seeks solutions to intractable problems not in the slow and collective work of politics but in technological systems that can only reproduce and exacerbate inequalities.
Send Letters To:
The Editor
London Review of Books,
28 Little Russell Street
London, WC1A 2HN
letters@lrb.co.uk
Please include name, address, and a telephone number.
