💰 A Japanese company has, for the first time ever, booked an annual net profit of more than 5 trillion yen. SoftBank Group's full-year result of ¥5.0022 trillion ($32 billion) is roughly 4.3x what it earned the year before, and it puts the company atop the all-time Japanese earnings ranking. But how does this number really look when you set it next to Apple, Alphabet, and Microsoft? And how risky is Masayoshi Son's "all-in on AI" strategy? Here is what the world's largest startup investor just told the market.

The headline number: ¥5 trillion, in context

On 13 May 2026, SoftBank Group (SBG) reported a net profit of ¥5.0022 trillion ($32 billion) for the fiscal year ended March 2026 — the first time any Japanese company has crossed the ¥5 trillion line on a single year's bottom line. Revenue was ¥7.7986 trillion ($50 billion), up 7.7% year-on-year.

To put that in perspective, the previous Japanese record was SBG's own ¥4.99 trillion in fiscal 2021, which had narrowly missed the milestone. Toyota, the country's largest manufacturer, peaked at roughly ¥4.94 trillion in fiscal 2024. So in a sense, this is SBG breaking its own record by less than ¥100 billion — but doing so for the first time decisively above the symbolic ¥5 trillion mark.

SBG is not a manufacturer. The bulk of this profit is not from selling things; it is from the rising paper value of companies the group has invested in. That is an important caveat to remember every time you see this number in a headline.

The Vision Fund, fully recovered

It is easy to forget how dark things looked just three years ago. In fiscal 2023, after a brutal global tech sell-off, the Vision Fund posted billions in losses, and Masayoshi Son famously opened a quarterly briefing with a portrait of the defeated warlord Tokugawa Ieyasu to symbolize his own humiliation.

Fast forward to fiscal 2026, and the Vision Fund is once again the engine of the group. In the January–March quarter alone, the unit booked an investment gain of ¥3.1 trillion ($20 billion). For the full year, SBG has now strung together five consecutive quarters of net profit. The recovery has been driven less by a broad portfolio bounce and more by one very large, very specific bet.

OpenAI: a 6.7 trillion yen windfall

That bet is OpenAI. SBG disclosed that its OpenAI stake generated an investment gain of more than ¥6.7 trillion in this fiscal year alone — a figure larger than the company's entire net profit. Total committed investment in OpenAI now stands at $64.4 billion, for a stake of roughly 13%.

To fund that bet, SBG has been actively reshaping its balance sheet:

  • It sold its entire remaining Nvidia stake for $5.8 billion in October 2025.
  • It sold down T-Mobile US holdings.
  • It expanded a margin loan secured against its Arm shares from $13.5 billion to $20 billion.
  • It took out an additional ¥400 billion loan secured against its domestic telco subsidiary's stock.

In March 2026, SBG also signed a contract with five major banks — Mizuho, SMBC, MUFG, Goldman Sachs and JPMorgan — for up to $40 billion (¥6.3 trillion) in financing to support a planned additional $30 billion investment in OpenAI through 2026.

Arm is the load-bearing wall

Even though Arm Holdings rarely shows up as the lead actor in SBG headlines, it is structurally the most important asset in the group. SBG still owns about 90% of the British chip designer, whose stock has rallied as the AI boom has made its CPU architectures the default in data centers and mobile devices alike.

Arm now plays three roles for SBG simultaneously: a long-term strategic holding, a source of unrealized gains that pad the balance sheet, and — increasingly — collateral for the loans funding the OpenAI bet. If Arm's share price corrects sharply, the margin loan structure could force SBG into more dilutive funding.

How does ¥5 trillion compare to US Big Tech?

This is where the perspective changes. Converted at roughly ¥157 per dollar, SBG's ¥5 trillion is about $32 billion. Set against the US tech giants in their most recent calendar year:

  • Alphabet (Google): roughly $132 billion in net income
  • Apple: roughly $112 billion
  • Microsoft: roughly $108 billion
  • Meta: roughly $83 billion
  • NVIDIA: roughly $73 billion
  • Amazon: roughly $59 billion

In other words, the largest profit ever recorded by a Japanese company is still smaller than what Amazon — the lowest-profit member of the Magnificent Seven — earns. It is roughly a quarter of Alphabet's. The gap is not a slight; it is a structural reminder of how concentrated the global AI profit pool has become at the very top of the US tech stack.

That said, SBG's structure is fundamentally different from any of those names. Alphabet sells ads. Apple sells phones. SBG owns slices of the companies building the AI infrastructure underneath them all, including OpenAI itself and Arm's chip blueprints. The question is how much of that paper position eventually translates to realized cash.

The risk: funding gap and OpenAI dependency

Not everyone is celebrating. Credit research firm CreditSights has estimated that SBG could face a funding gap of around $32 billion (about ¥5.1 trillion) over the next two years, when you stack upcoming bond redemptions on top of committed investments — including the OpenAI top-up and the agreed $5.4 billion acquisition of ABB's industrial robotics arm.

There are other unanswered questions:

  • If OpenAI's IPO is delayed, SBG cannot easily realize the paper gains it has booked.
  • A significant share of "investment income" is mark-to-market — sentiment, not cash.
  • Concentration risk in OpenAI is unusual even by SBG's historically bold standards.

CFO Yoshimitsu Goto framed this on results day in characteristic terms: "We are still looking far ahead. Thirty years from now, this AI revolution is still only just beginning. We want to be a group that stands at the center of it."

Where this leaves Son's global investment strategy

The Ohio "Stargate" project — a $500 billion AI campus on the site of a former uranium enrichment facility, planned in partnership with OpenAI and Oracle — is the most visible expression of where SBG's money is now going. Son has also been in talks with French President Emmanuel Macron about a multi-billion-dollar data center investment in France, with some reports suggesting figures up to $100 billion across multiple projects.

Strip away the slogans, and the strategy is fairly simple: borrow against listed assets (Arm, domestic SoftBank Corp, sold-down Nvidia/T-Mobile), pour the proceeds into AI infrastructure and OpenAI, and wait for either the OpenAI IPO or downstream commercialization to validate the position. The fiscal 2026 results don't prove the strategy works long-term; they prove the paper math currently does.

Closing thought

For Japan, the symbolism of the ¥5 trillion milestone matters. It is the first time any domestic company has crossed it, and it has happened not via cars or electronics but via a Tokyo-listed holding company quietly underwriting much of the global AI infrastructure boom. Whether that turns into the next era of Japanese corporate power or into the most spectacular single-bet bust in the country's history is genuinely undecided.

In Japan, public opinion is sharply split — admiration for the audacity, anxiety about the leverage. How is SoftBank's AI bet being talked about in your country? Do investors there see Son as visionary, reckless, or both?

References