In the middle of last December, Masayoshi Son appeared by surprise at Mar-a-Lago (Florida, USA) with Donald Trump. The founder of the Japanese group SoftBank promised to invest 100 billion dollars in the United States in the next four years. The president-elect joked that SoftBank might want to double that goal to $200 billion. Son accepted the order and responded with a broad smile: “I'll try.” It may sound like a bluff, but maybe it isn't. The Japanese businessman has been quietly planning one of his characteristic business bets for some time. Son has developed a singular obsession: building the next Nvidia and accessing a billion-dollar market. Son's goal is to have the first batch of chips for use in artificial intelligence (AI) ready in 2026, according to sources close to the tycoon. Son, 67, considers himself an AI pioneer and, however, has seen how Nvidia captures all eyes in this field. The American chip company has a near monopoly on the semiconductors used to train AI models like OpenAI's ChatGPT, which has pushed its market capitalization above $3 trillion. The owner of SoftBank wants to make up for lost time and become its biggest competitor. To achieve this, it has a broad portfolio of start-ups and a 90% stake in Arm Holdings, the British technology company whose chip designs earned a quasi-monopoly for smartphones due to their energy efficiency. With a fortune of more than 15 billion of dollars and aware that some failed investments have clouded his legend in recent years, Son is determined to develop his own chips to bring AI to every place imaginable, beyond data centers. It knows that a weak point of this technology is its high energy consumption and to reduce this dependency it has the collaboration of Arm and the Taiwanese giant TSMC as a manufacturer. In addition, another of the levers to accelerate its growth in the field of AI could be Intel. At the beginning of 2024, Arm proposed the purchase of the American company. Its then CEO, Pat Gelsinger, flatly refused. But the manager is no longer an obstacle after being dismissed a few months ago by the Intel board. Although the roadmap to compete with Nvidia is beginning to be public, along the way there are many doubts to clear up. Among them, the sources of financing for this new bet, since SoftBank has a very large debt. The company plays down the high leverage. SoftBank's chief financial officer, Yoshimitsu Goto, assures that his boss no longer has financial limitations for his ambitions: “We are willing to enter with all the money that Son needs. We can always find a way to deal with investments. “It would not be wise to put limits on a project that has just begun.”
Image change
SoftBank, in addition to entering a million-dollar business, needs an image change. The last time Son made an investment promise to Trump in 2016, the Vision Fund with which he pumped more than $100 billion into hundreds of start-ups around the world, fueled one of the biggest bubbles in the capital sector. risk that are remembered. That bubble ended with the implosion of companies such as the co-working platform WeWork, the construction company Katerra, the smart window manufacturer View, the direct-to-consumer company Brandless and the robotic pizza delivery company Zume. Although the two Vision funds have recovered the worst of their losses, they have damaged Son's legacy. Until then, the Japanese businessman, who went from poverty to rubbing shoulders with the richest in the world, was considered a kind of King Midas with bets that included the launch of iPhones in Japan, a pioneering deployment of broadband in the Asian country and being one of those who first opted for the Chinese e-commerce giant Alibaba. Currently, Arm is the most valuable asset in SoftBank's portfolio: it paid 32 billion in 2016 and now the chip development group is worth 195 billion. Son and Arm CEO Rene Hass want to enter the market. Nvidia's private preserve, AI accelerators. The graphics processing units of the company co-founded and led by Taiwanese Jen-Hsun Huang are far ahead of the competition in terms of capacity and ease of use. However, Son knows that the enormous investment effort required by AI will open a gap for Nvidia's rivals, which alone will not be able to satisfy the growing demand for processors. “There is room for new competitors precisely because the market is going to grow a lot. and to change a lot,” says Chris Miller, author of the book Chip War: The Fight for the World's Most Critical Technology. “It's not surprising that SoftBank, given its ambitions and scale, is also looking at this market,” he adds. When SoftBank bought Arm in 2016, Son called the Cambridge, UK-based company his crystal ball for fortune-telling. the future of technology. The bet was met with misgivings at first, but when the businessman managed to raise funds worth $60 billion from Saudi Arabia and Abu Dhabi, critical voices faded. That gave it great leverage to accelerate the growth of hundreds of start-ups around the world and replicate the mega-successes of the past, such as investments in the aforementioned Alibaba or Yahoo. However, the reality was less triumphalist than the calculations in your Excel sheet. The deployment of so much investment capital in such a short time had led to erroneous selections. The lack of a detailed analysis process (due diligence) on the acquired companies and the fact that the minimum bet was 100 million euros amplified the snowball effect. The result was immediate: in March 2022, the losses incurred by the Vision Fund funds totaled $20 billion. In the fall of that year, Son assumed all responsibilities and removed himself from the front line of management to delegate to his lieutenants. “I couldn't stop crying for days,” she told SoftBank shareholders last June. After the failure of his bets, Son ran the risk of going down in posterity as a mediocre company manager. And that image terrified him. “I want to be an architect of the future, even if I die while drawing the plans,” he acknowledged at that event.
The 'eureka' moment
As part of their promised rematch, and behind the scenes, Son was working with Haas to transform Arm. For most of its history, the company had focused on being a mere license seller, charging pennies on the dollar for each device that used its technology. But under the leadership of Haas, and with the support of Son, Arm began to move up the value chain, becoming a comprehensive chip developer. In June 2024, Son had his eureka moment at four in the morning. He sent a text message to the chat he shared with his closest managers in a state close to ecstasy. Hours later, during the annual shareholders meeting, he compared his euphoria to solving a complex equation after having worked on it day and night for a year. “Trust me. «Let's make it,» was his enigmatic message, without going into details. Later, he urged chat members to review any business plan every week. They would have to be flexible and prepared for any scenario. The new adventure of this technology visionary was named Izanagi in honor of the Japanese god of creation and life. The battle for Nvidia's scepter had just begun. Son's ambitions to build his own chip empire have received an additional boost thanks to his relationship with OpenAI CEO Sam Altman. The two had maintained contact after their first meeting in 2019, when Son offered OpenAI $1 billion in investment, a proposal that never materialized. Each Nvidia chip costs tens of thousands of dollars. In response to its high price and the long waiting process to receive a graphics processing unit, Son and Altman considered creating their own AI chips. The SoftBank founder suggested they would need $3 billion to do it successfully, the OpenAI guru raised the figure to $7 billion, according to The Wall Street Journal. As for technical support, Son turned to Arm. Since its founding in 1990 by a small group of engineers, the company has grown to become the owner of the most widely used microprocessor technology in the world. The idea is that Arm engineers can repeat the success they had 35 years ago, guided by Son's vision of what an AI-adapted chip needs. Haas's ambitions to transform Arm into a different chip company and more large coincided with those of Son. However, the two businessmen have not always agreed. Haas, who previously worked on Nvidia computing products, is fully aware of the challenges posed by chip design and the industry's manufacturing difficulties. After receiving repeated denials from Haas, a frustrated Son at one point demanded to speak to other Arm executives, to which Haas refused. Despite the differences, the two continue to talk daily, and Haas has become the person in charge of bringing the Japanese's great ideas down to earth. However, SoftBank is not the only one that aspires to a slice of the growing pie. AI chips. Californian giant AMD, which argues its chips are better than Nvidia's, plans an upgrade to its accelerators (AI accelerator chips are specialized microprocessors designed specifically to accelerate complex artificial intelligence calculations) next year. For its part, Amazon, which invented cloud computing operations with custom-built hardware, is also using its economies of scale to compete with Nvidia. Jeff Bezos' engineers are working to deliver their AI accelerator Trainium to data centers in the first few months of 2025.
Great competition
Chip industry veterans have watched SoftBank's efforts to advance in the field despite this growing competition with bemusement. Son's business skills have served him well throughout his career selling software, publishing magazines, deploying mobile networks and supporting young entrepreneurs, but he has never been put to the test in the realm of nanometers. and the clean rooms of chip manufacturing. Even Nvidia almost went under several times before reaching the top of this business. In the end, SoftBank's chances of success largely depend on Arm's intellectual property. Other weapons in Son's arsenal include SoftBank's customer base—although it's only a fraction of Amazon's—in a country like Japan, which is eager to pay big bucks to catch up in AI. The technology group also operates renewable energy projects in the US that help power Google's data centers and other hyperscalers. Son also has a team of investment experts in his Vision Fund manager, hunting for the companies and technologies he needs to fight the battle in AI. During the SoftBank shareholder meeting, one of the attendees asked about Bloomberg's report on the Izanagi project and how much progress had been made in its development. Son smiled. “Let's not talk about specific things,” he said. “We don't want to show our cards too soon.” The game for the throne of AI has only just begun and the enfant terrible of technology is confident of having a good hand this time.
