SoftBank Group — Company History

A founding that filled the gap in software distribution, and an entry into publishing

In September 1981 the 24-year-old Son Masayoshi (孫正義) established Japan SoftBank at Yonbancho, Chiyoda-ku, Tokyo, and began wholesaling packaged software for personal computers. The company's founding ground is the city of Fukuoka. Personal computers were spreading and the range of software was widening, yet almost no business existed to carry goods from makers to retail shops, so a user had to trawl several microcomputer shops before reaching the title they wanted. Son designed the business as an intermediate distributor that would deliver software the way a bookshop delivers a book. He narrowed his options to a handful of industries on the criteria of low capital and high growth, hired staff, and spent a year and a half studying the market and the technology before committing to the founding. In May 1982 the company launched the monthly magazines Oh! PC and Oh! MZ, entering publishing.

The founding capital was the $453,412 (¥100m) Son had earned from the patent on an automatic translation device he developed while studying in the United States. Software could not serve as collateral, so no bank would lend to a distribution business at its founding, and the loan was executed only against personal guarantees from Sharp's Sasaki Tadashi (佐々木正) and others. To make up for his lack of managerial experience, in 1983 Son brought in Omori Yasuhiko (大森康彦), formerly of Japan Guard Security, as president, and stepped back to become chairman. Revenue grew from $9.2M (¥2bn) in the second year to $18.9M (¥5bn) in the third and $31.6M (¥8bn) in the fourth, reaching $48.2M (¥12bn) in the fifth, the year ended December 1985. Oh! PC, launched the year after incorporation, also succeeded, and publishing grew into a pillar of the business alongside wholesaling. Son returned to the presidency in 1986.

Flotation, and buying an information network

In July 1990 the company changed its name to SoftBank, and in July 1994 it registered its shares with the Japan Securities Dealers Association and went public. Using the fundraising power that flotation gave it as a lever, from December that year it moved on the American trade-show operator Comdex and the Silicon Valley IT publisher Ziff-Davis. The acquisitions were funded with bonds: beginning with a first unsecured straight bond of $106.3M (¥10bn) in February 1995, it issued twelve tranches totalling $2.0B (¥220bn) by November 1996. Piling large acquisitions on borrowings one after another swelled interest-bearing debt to a scale greater than its revenue. In January 1998 it listed on the First Section of the Tokyo Stock Exchange.

On 11 January 1996 SoftBank established Yahoo Japan Corporation as a joint venture with Yahoo of the United States, capitalised at $1.8M (¥200m). SoftBank held 60 per cent, or $1.1M (¥120m), and Yahoo Corporation 40 per cent, or $735,497 (¥80m); Son Masayoshi himself took the post of representative director and president. The business was defined as providing a Japanese-language information search service within Japan and operating a mirror site — an irregular joint venture in which the brand and the technology were borrowed from the American Yahoo while SoftBank held the capital and the management. In April the same year it put a further $91.9M (¥10bn) or so into Yahoo itself in the United States, becoming a larger shareholder than the founders. The American Yahoo was at that point no more than a venture with five or six employees and monthly revenue in the tens of millions of yen.

The market voiced concern that the run of acquisitions was draining the company's financial strength, and in August 1996 the question was put directly: could it stand the strain of one purchase after another? Son Masayoshi replied that there was nothing to worry about, arguing that the acquisitions were not expensive purchases and that he was not chasing easy gains. In the autumn of 1997 the deteriorating results of the acquired companies came to the surface, and clouds gathered over Son's expansionary line. Criticism also gathered around an acquisition scheme that used a company Son owned personally; Son said that the responsibility for the shadowed parts is mine as well, while stating plainly that the acquisitions would continue. Even in November that year the company had got through only by financial contrivance, and with the acquired businesses worsening nothing could be taken for granted.

The internet bubble, and shrinking the balance sheet in the collapse

In October 1999 SoftBank converted into a pure holding company, reorganising itself to hold and oversee the shares of its operating companies. In February 2000 it invested in Alibaba.com Corporation, making it an affiliate and taking a stake that would later become the group's largest unrealised asset. At the peak of the internet bubble, the surge in the American Yahoo's share price briefly lifted SoftBank's market capitalisation to $185.6B (¥20tn). Most of that valuation rested on unrealised gains related to Yahoo, and the gap with the profit its businesses actually generated was wide. Less than six years after the 1994 flotation, a software wholesaler had turned into an investment company with one of the largest market capitalisations in the world.

The fall in share prices from the spring of 2000 pushed many of its investees into crisis. SoftBank selected and sold the holdings that had appreciated, using the proceeds to shrink interest-bearing debt while clearing out the assets it had accumulated through the 1990s. In parallel it went ahead with the purchase of Nippon Credit Bank, which had failed, been nationalised and been dealt with using public money. Concerns were raised from the outset that a group centred on an investment company owning a bank would turn it into an in-house lender concentrating credit on affiliated firms. In September 2000 President Homma (本間) of Nippon Credit Bank took his own life, and the wind turned against SoftBank after the acquisition.

Front-loaded investment in broadband, and a move into fixed-line telecoms

In September 2001 BB Technology, a SoftBank subsidiary, launched the commercial Yahoo! BB service under the Yahoo brand, taking the group into broadband. Internet access at the time ran mainly over ISDN, which was slow and expensive. ADSL was emerging as a technology that delivered high speed at low cost over existing telephone lines. SoftBank enclosed users en masse by handing out modems free in the street — the so-called parasol squads — by saturating television with advertising, and by giving gift vouchers to subscribers. It was deliberate front-loaded investment, made in full knowledge that losses would widen, on the premise that once subscribers passed a certain number the business would break through its break-even point.

In the first year of entry, the year ended March 2002, the broadband infrastructure business recorded revenue of $58.3M (¥7bn) against an operating loss of $142.9M (¥18bn). In the following year, ended March 2003, the loss widened to $829.9M (¥96bn) on revenue of $344.2M (¥40bn), and capital expenditure reached $663.4M (¥77bn). Son Masayoshi said at the time: On a monthly basis we will start turning profitable at some point during fiscal 2002. It depends on how much we spend on promotion, but if we win a million-odd users we ought to be able to break through the break-even point. The crossover actually came in the fifth year, the year ended March 2006, with revenue of $2.3B (¥267bn) and operating profit of $177.2M (¥21bn).

In July 2004 SoftBank made the fixed-line carrier Japan Telecom a subsidiary. Japan Telecom had come under Britain's Vodafone as its largest shareholder in 2001, then had its mobile arm separated off, and was struggling in a fixed-line market where NTT held the advantage. The acquisition brought in a corporate customer base and an optical-fibre network, and immediately afterwards SoftBank launched cut-price fixed-line telephony, competing with NTT head-on on call charges. In January 2005 it made the Fukuoka Daiei Hawks a subsidiary. Daiei, rebuilding under the Industrial Revitalization Corporation of Japan, was being pressed to sell a baseball club that was popular but unprofitable. SoftBank was not seen as the favourite at first, but Son Masayoshi invoked the tie of having spent his boyhood in Fukuoka and settled the negotiation by buying the club's franchise rights outright.

Becoming a mobile carrier, and expanding into North America

In April 2006 SoftBank made the Japanese arm of Britain's Vodafone a subsidiary and entered the mobile phone business as a new operator. The total acquisition cost was about $15.0B (¥1.75tn), most of it raised through a leveraged buyout. By arranging a non-recourse loan secured on the assets of the Japanese company being acquired, SoftBank could execute an enormous investment while limiting the effect on its own balance sheet. Son Masayoshi had repeatedly declared in public his intention to enter the mobile market, saying he would get in whatever the timing or the method. The experience of running a telecoms business, gained through the Japan Telecom acquisition, provided the footing for this entry.

From the end of 2011 it put $3.1B (¥250bn) into building out and reinforcing mobile base stations, closing the gap in network quality with its rivals. Operating profit for the year ended March 2012 reached $8.5B (¥675bn), roughly ten times the level at the time of the Vodafone acquisition, and net interest-bearing debt shrank from about $21.5B (¥2.5tn) at the end of June 2006 to $10.0B (¥800bn) by the end of June 2012. A path had opened to repay the heavy borrowings taken on at the acquisition out of the cash the mobile business generated, creating the financial headroom for the next large purchase. In April 2015 the four companies SoftBank Mobile, SoftBank BB, SoftBank Telecom and Ymobile merged, and in July that year the holding company renamed itself SoftBank Group.

Looking abroad for its next growth, SoftBank considered Asia first, but China's major mobile operators were state-owned and could not be bought, and India was ruled out because monthly revenue per subscription was around two dollars, so it settled on the United States, where revenue levels were closer to Japan's. On 15 October 2012 it announced a deal to acquire 70 per cent of Sprint's shares for $20.1bn. When America's Dish Network made a counter-proposal in April 2013, SoftBank raised the total to $21.6bn and won approval at an extraordinary general meeting on 25 June. On 11 July that year it completed the purchase of about 78 per cent of Sprint for roughly $21.6bn in all, making it a subsidiary and becoming the world's third-largest mobile phone company by revenue, with some 97 million subscriptions.

Into semiconductor design: the ARM acquisition as a turning point

In September 2016 SoftBank Group made the British semiconductor design company ARM a subsidiary. The offer price was 1,700 pence a share, a 43 per cent premium to the closing price of 15 July 2016. The total came to about £24bn, or roughly $30.3B (¥3.3tn), surpassing JT's purchase of Gallaher of Britain as the largest acquisition ever made by a Japanese company. ARM manufactures no semiconductors: it is a fabless company that licenses its designs to others, and 95 per cent of the smartphones sold worldwide carried chips based on those designs. Its operating margin for the year ended December 2015 was 51.6 per cent, above the 25 per cent of the leader, Intel. The money was found from the just under $9.2B (¥1tn) raised that June through sales of Alibaba and GungHo shares and other disposals.

Son Masayoshi placed the aim of the acquisition in seizing the design layer for an age in which semiconductors would be embedded in every device, saying that within twenty years ARM would ship something like a trillion chips a year and draw in the data of the planet. People think SoftBank is a mobile company, but that is only the last ten years, he also said, showing his intention to swap out the core business and lead the next transformation. Of the $30.3B (¥3.3tn) price he remarked that it was only three trillion yen. The ARM business did not settle down financially for some time after the acquisition, however: in the year ended March 2020 it recorded an operating loss of $400.8M (¥43bn) on revenue of $1.9B (¥206bn).

A ¥10 trillion fund, and results that move with share prices

In May 2017 the SoftBank Vision Fund 1, investing mainly in technology companies, began operating. At its first closing on the 20th of that month it had gathered more than $93bn, about $92.7B (¥10.4tn). Committed capital eventually reached $98.6bn, made up of $33.1bn from SoftBank Group and $65.5bn from outside investors led by Saudi Arabia's Public Investment Fund (PIF), so that external money accounted for around two-thirds. SoftBank Group contributed about $8.2bn of its own share in kind, in ARM shares. Where conventional venture capital funds ran hundreds of billions of yen, this was a fund of a different order of magnitude, putting billions of dollars into a single company — Uber, ByteDance, WeWork.

In October 2019 the SoftBank Vision Fund 2 began operating. External money did not come to SVF2, however: of its $59.8bn of committed capital, $57.2bn came from SoftBank Group and only $2.6bn from outside investors. SoftBank Group's share, about a third in SVF1, rose to 96 per cent in SVF2. The form reverted from managing money entrusted by others to investing its own, so that changes in the value of the investees fed through almost directly into the group's profit and loss. As at the year ended March 2024, SVF1's cumulative investment was $102.0bn for a cumulative gain of $16.7bn, while SVF2 showed a loss of $19.3bn on $52.4bn invested.

As the fund business grew in weight, consolidated net profit and loss swung with the share prices of the investees. The year ended March 2021 produced net income attributable to owners of the parent of $45.4B (¥4.99tn), while the year ended March 2022 turned to a net loss of $13.0B (¥1.71tn) and the year ended March 2023 to a net loss of $6.9B (¥970bn). The crisis at the shared-office operator WeWork, and the losses booked when its IPO was pulled, were emblematic of this structure. A further net loss of $1.5B (¥228bn) followed in the year ended March 2024, before the year ended March 2025 returned net income of $7.7B (¥1.15tn) — results moving by hundreds of billions of yen from one quarter to the next.

Shrinking the balance sheet in the pandemic, and a ¥2.5 trillion buyback

In February 2020 it emerged that Elliott Management of the United States had acquired SoftBank Group shares and was demanding a buyback of up to $20bn, about $20.6B (¥2.2tn), along with an increase in the number of outside directors. The company's shares were then trading below the value of the holdings it owned, such as its Alibaba stake: against a share price of ¥4,621 on 18 May 2020, shareholder value per share calculated from the value of those holdings was estimated at about ¥10,300. Elliott built its position that year to around $3bn, and later, as the buybacks proceeded, disposed of almost all of it.

On 19 March 2020 SoftBank Group's share price fell more than 17 per cent in a single day, and had roughly halved over a month. On the 23rd the company announced that it would sell and monetise up to $42.1B (¥4.5tn) ($41bn) of its assets, applying the proceeds to buybacks of up to $18.7B (¥2tn) and to reducing interest-bearing debt. Together with the $4.7B (¥500bn) announced on 13 March, the buyback programme came to $23.4B (¥2.5tn) in all, a scale equivalent to acquiring and cancelling 45 per cent of the shares in issue. The asset sales were planned over four quarters. After the announcement the shares were bid to the daily limit, rising ¥500 from the previous Friday's close to ¥3,187. As at December 2020 the company reported that $14.0B (¥1.5tn) of the $23.4B (¥2.5tn) had been executed.

For the year ended March 2020, write-downs at the Vision Fund and other items produced a net loss attributable to owners of the parent of $9.0B (¥962bn) on revenue of $49.1B (¥5.24tn) — the largest deficit in the company's history at that point. It was thus carrying out a large-scale buyback while posting a net loss, with the money coming from selling holdings such as its Alibaba shares. Those Alibaba sales proceeded in stages, and in August 2022 SoftBank Group's voting interest in Alibaba fell below 20 per cent, so Alibaba was removed from the scope of affiliates. Twenty-two years after the February 2000 investment, the group's largest unrealised asset had passed outside the consolidation.

The failed ARM sale, and concentrated investment in AI infrastructure

On 14 September 2020 SoftBank Group announced a contract to sell its entire holding in ARM to Nvidia of the United States for up to $40bn. The consideration combined cash and Nvidia shares, and on completion SoftBank expected to hold about 6.7 to 8.1 per cent of Nvidia's shares in issue. ARM's customers, however, included Qualcomm and Apple, competitors of Nvidia, and doubts about its neutrality were raised from the outset. In December 2021 the US Federal Trade Commission sued under antitrust law to block the deal, calling ARM the Switzerland of the semiconductor industry in its complaint. On 8 February 2022 the two companies agreed to terminate the contract, citing regulatory obstacles; the $1.25bn deposit carried no obligation to refund and was recognised as a gain.

ARM listed on its own on the Nasdaq in September 2023. Its closing price on the first day was $63.59, 25 per cent above the offer price, giving a market capitalisation of $65.2bn, about $68.3B (¥9.6tn). That was roughly three times the $30.3B (¥3.3tn) paid in 2016, and SoftBank Group continued to hold about 90 per cent of the shares. By August 2024 ARM's market capitalisation had grown to some $165.0B (¥25tn). Revenue at the ARM business also expanded, from $2.3B (¥300bn) in the year ended March 2022 to $2.7B (¥382bn) and then $3.1B (¥464bn) in the year ended March 2024, with headcount rising to 6,928.

In January 2025 SoftBank Group announced the Stargate Project, to build AI infrastructure for the American AI research and development company OpenAI and its affiliates. The plan is to invest $500bn over the following four years, of which $100bn is to be committed immediately. The initial backers were four parties — SoftBank Group, OpenAI, Oracle and MGX — with SoftBank Group responsible for the finance and OpenAI for the operations. In March the same year it agreed to invest up to a further $40bn in OpenAI, and on the 20th of that month announced the acquisition of Ampere Computing, an American designer of server CPUs, for $6.5bn in total. Son Masayoshi said: The future of artificial super intelligence requires unprecedented computing power.