The Peak It Knows

On the first Monday of June 2026, SoftBank Group passed Toyota to become the most valuable publicly traded company in Japan. Its market capitalization reached roughly ¥47.2 trillion — about $296 billion — pulling ahead of the carmaker that had held the crown, with only the briefest interruption, for a generation. The stock had nearly doubled in a year, up more than 90%, carried aloft by a single conviction: that Masayoshi Son, SoftBank's founder, has positioned his company at the dead center of the artificial-intelligence revolution.

It is a genuine triumph, and the market is right that Son has made an extraordinary bet. But there is a detail in the record books that should make anyone celebrating this milestone go very quiet, and it is this: the last time SoftBank was the most valuable company in Japan was at the peak of the dot-com bubble in 2000. Toyota had not been dethroned in more than two decades, and the only previous time it happened was the exact top of the last technology mania — the one after which SoftBank's stock fell roughly 99%, and Masayoshi Son personally lost an estimated $70 billion, more money than any individual had ever lost in the history of the world.

Son has stood on this summit before. He knows, better than any person alive, what is on the other side of it. The question that should haunt this moment is whether he has learned the lesson the last peak taught him — or whether he has simply found a bigger ledge from which to make the same leap.

The man who lost the most money in history

To understand the stakes, you have to understand the protagonist, because SoftBank is not really a company. It is Masayoshi Son's personality rendered as a balance sheet.

Son's career is the most volatile in modern finance, a sine wave of vertiginous highs and near-fatal lows. In 2000 he was, for a brief shining moment, reportedly the richest person on earth, his SoftBank shares having ridden the internet bubble to absurd heights. When the bubble burst, those shares collapsed by some 99%, and Son's paper fortune evaporated by around $59–70 billion — a sum so staggering it earned him, for years, the unwanted distinction of having lost more money than anyone who has ever lived. He rebuilt through telecoms. Then, in 2017, he raised the $100 billion Vision Fund — the largest technology fund in history — and proceeded to author some of the most spectacular blunders of the era, none more famous than WeWork, the office-rental company he valued at $47 billion and which nearly collapsed into bankruptcy after a humiliating failed IPO. In the fiscal year ended March 2022, the Vision Fund posted a loss of around $27 billion. Twice now, Son has taken his company to the brink of ruin by believing too hugely, too early, with too much leverage.

He is not a cautious man who occasionally gambles. He is a gambler who occasionally wins so enormously that the wins rewrite his reputation between the losses. This is the temperament now steering a $180 billion bet on AI.

The bet, and its concentration

Son's comeback — and it is, on paper, the largest comeback in business history — rests on a fortune now reportedly larger than at any previous point in his life, built almost entirely on one position: OpenAI.

Through commitments approaching $65 billion, SoftBank is on track to own roughly 13% of OpenAI by October — a stake that, at the trillion-dollar valuation OpenAI is targeting for its IPO, would be worth on the order of $130 billion on a look-through basis. SoftBank's Vision Fund reported gains of about $43.9 billion in its most recent fiscal year, and nearly all of that increase was attributed to its OpenAI mark. Son is also chairman of Stargate, the $500 billion AI-infrastructure venture; SoftBank has poured money into AI data centers, energy, and a roughly €75 billion investment in France; and it still owns the majority of Arm, the British chip designer whose blueprints sit inside most of the world's AI servers and smartphones.

Read the bull case and it is dazzling: SoftBank is the purest large-cap proxy for AI on any public exchange, an investor that got into the single most important private company of the age early and at scale. Read the same facts as a risk officer and the picture inverts. SoftBank's surging value rests overwhelmingly on a paper mark on an illiquid, private stake in a company — OpenAI — that loses $14 billion a year, will not be profitable until the end of the decade, and has not yet sold a single public share. The $43.9 billion "gain" is not cash. It is an accounting estimate of what a private holding is worth, and it can be marked back down as fast as it was marked up. SoftBank's net worth has become a leveraged bet on a single unprofitable company's continued ascent — which is, almost precisely, the structure that destroyed it in 2000, scaled up by a factor of ten.

The leverage underneath

What makes the echo of 2000 more than poetic is the debt, because Son does not make these bets with cash alone.

SoftBank has long funded its ambitions through borrowing against its assets, and the AI pivot is no exception. Apollo recently expanded a net-asset-value loan against the Vision Fund 2 to a record $5.4 billion — the largest such facility known in the market — a structure that borrows against the very portfolio marks that are themselves estimates. At the end of 2025, SoftBank was reported to be scrambling to find $22.5 billion to meet its OpenAI funding commitments before year-end, hardly the posture of a company swimming in spare capital. And the SoftBank holding company has, for years, traded at a persistent conglomerate discount of 40–50% to the stated value of its own assets — the market's standing verdict that it does not fully trust either Son's marks or his leverage. The company insists its loan-to-value ratio remains disciplined, below 25%. But the assets in that ratio's denominator are themselves the soaring, unrealized, private valuations of the AI boom. Leverage measured against a bubble's prices is only conservative until the prices move.

This is the machinery that turns a great bet into a dangerous one. When the underlying marks rise, borrowing against them looks prudent and the equity compounds magnificently. When they fall — as private AI valuations surely will at some point in any cycle — the leverage works in reverse, margin against estimates that no longer hold, exactly when liquidity is scarcest. It is the 2000 structure and the WeWork structure wearing a newer, more fashionable logo.

The peak it knows

None of this is a forecast that SoftBank collapses, and it would be foolish to bet against Masayoshi Son's capacity to be spectacularly right; he has been before, and OpenAI may indeed become the defining company of the century and vindicate every yen. SoftBank's Arm stake is real and valuable, its OpenAI position may prove the investment of a generation, and a conglomerate discount means the shares are arguably cheap relative to the assets even now. The bull case is not naïve. It is a bet on the most aggressive technology investor of the modern age being right about the most important technology of the age — which is not the worst bet one could make.

But the symbolism of June 1 is too precise to wave away. A company returns to the exact summit it last occupied at the top of the dot-com bubble, on the strength of a leveraged, illiquid, paper-marked bet on a single unprofitable company, steered by the man who once lost more money than anyone in history doing a structurally identical thing — and the market treats the milestone as pure vindication rather than as the most ominous rhyme in finance. Tops are not rung by bells. They are rung, sometimes, by a company reclaiming a record it has held exactly once before, at exactly the wrong moment, and by a crowd cheering the achievement instead of reading the date beneath it.

Masayoshi Son knows this peak. He has been here. He, of all people, knows what the air feels like just before the long way down — and he has bet a hundred and eighty billion dollars, and a great deal of borrowed money, that the view is different this time. It may be. But the most expensive four words in markets are the ones his entire comeback is wagered against, and the ones every investor reaching the summit on his coattails should say aloud before they buy: this time is different.