Author: Le Ming
May 2, 2026, Omaha. This time, only a little over half of the 18,000 seats were filled—in the past, you had to scramble for tickets and queue up to get into the Berkshire Hathaway annual shareholders meeting, and hotels outside the venue were impossible to book.
This time, the 95-year-old Warren Buffett was not presiding on stage as in previous years. The new CEO, Greg Abel, stood in front of the main screen, fielding investors' tough questions about why Berkshire was sitting on nearly $400 billion in cash.
That same week, six thousand miles away in Tokyo, Masayoshi Son's team was doing something else: bundling SoftBank Group's yet-unprofitable AI assets, preparing to load them into a new company called Roze AI, targeting a $100 billion valuation and planning a U.S. listing in the second half of 2026.
The reason is simple: SoftBank must continue to find money and continue investing for OpenAI's $64.6 billion check, which could ultimately balloon to nearly a trillion dollars.
One is holding $397.4 billion in cash, buying nothing, waiting for the market to crash; the other is carrying over 16.34 trillion yen (over $100 billion) of parent company interest-bearing debt, betting the market won't crash.
Berkshire Hathaway: Too Much Money Is a Disease
What does $397.4 billion mean?
It's nearly 40% of Berkshire's total market capitalization, double Buffett's average cash level over the past two decades.
Of this, $339.3 billion is directly held in U.S. short-term Treasury bonds, making Berkshire one of the U.S. Treasury's largest non-governmental creditors.
This pile of money wasn't passively accumulated; it was actively built.
For the past fourteen quarters, Berkshire has been a net seller of stocks each quarter. Apple, once its largest holding, has been continuously reduced for four quarters starting from Q3 2024, with cumulative sales of nearly 688 million shares, raising over $100 billion.
Buffett's explanation has always been the same: can't find anything cheap.
He wrote one sentence in his 2024 letter to shareholders—"Generally, nothing appears attractive." In a recent off-site interview during the annual meeting, he described the current market as "a church next to a casino," saying that of all the market sentiments he has experienced, this is the most gambling-like.
The problem is, he's been making this judgment for more than a year.
Berkshire's stock performance over the past twelve months has lagged the S&P 500 by about 40 percentage points. This is no small number—it's one of the largest relative drawdowns since Buffett took over Berkshire in 1965. The last time such a degree of underperformance occurred was during the final stages of the dot-com bubble in 1999.
At that time, Buffett said he would only buy in "cutting-edge industries like bricks, carpet, insulation, and paint." Two years later, the Nasdaq fell 78%, and he was proven right.
But this time, investors have waited more than two years. The market rises, Berkshire doesn't move; the market rises again, Berkshire still doesn't move. On January 1 of this year, when Buffett officially stepped down as CEO, Berkshire's stock price fell slightly—the market expressed a measured disappointment in the most restrained manner.
This is the situation Abel inherited. He is a Canadian accountant who worked his way up to Berkshire Vice Chairman, spending his entire career in heavy-asset, regulated, slow-growth businesses like utilities, railroads, and energy.
He is not Buffett, and he knows he is not Buffett. In his first letter to shareholders, he repeatedly emphasized "continuity" and "decentralization." At his first annual meeting, his answer to all suggestions about splitting up the conglomerate was an absolute "impossible."
Abel's dilemma is: he can neither deploy this cash (because the market is too expensive), nor can he continue to pretend this money doesn't exist (because investors are voting with their feet).
If the market continues to rise over the next five to ten years, he will eventually have to face a question that has never been seriously discussed in Berkshire's history—giving the money back. Either returning it to shareholders as a special dividend, or actually breaking up and selling off this behemoth stitched together from over 60 subsidiaries.
Will Berkshire die? Not suddenly. Its holdings are too diversified, its cash too abundant, its debt too low; any external shock would struggle to truly pierce it. But it will slowly, decently, become something else.
SoftBank's Problem: Too Little Money and Still Betting
Masayoshi Son's situation is the opposite mirror image of Abel's.
On February 27, 2026, SoftBank issued an announcement. The most critical sentence translates to: "SoftBank Group's cumulative investment in OpenAI is expected to reach $64.6 billion, representing an ownership stake of approximately 13%."
$64.6 billion, 13%. It's the most expensive single bet of this era.
To understand the madness of this number, one needs to see how SoftBank managed to afford it.
The parent company's interest-bearing debt has skyrocketed from 12.14 trillion yen in March 2025 to 16.34 trillion yen in December 2025. The parent company's so-called cash is only about 3.8 trillion yen, nearly one-third of which is actually unused committed credit lines, not real cash.
Where did this money come from? SoftBank took out $20 billion from pledging its holdings of Arm stock; and scraped together about $7.7 billion from pledging its holdings in the Japanese telecom subsidiary SBKK.
On March 27, 2026, SoftBank also signed an unprecedented $40 billion bridge loan, led by five banks—JPMorgan Chase, Goldman Sachs, Mizuho, SMBC, and MUFG—later expanded to eight. This is one of the largest bridge loans in Asian history. $30 billion of this money is directly used to follow-on invest in OpenAI. The term is 12 months, meaning SoftBank must repay $40 billion by March 2027.
This is why Son has seemed a bit "not himself" this year: He completely offloaded SoftBank's entire Nvidia stake, raising $5.8 billion. This was a one-time sell-off in October 2025. In a Tokyo speech in early December 2025, he admitted: "I didn't want to sell a single share of Nvidia, but I needed more money to invest in OpenAI. I sold Nvidia crying."
To scrape together money for OpenAI, Son has been "selling the family silver." SoftBank sold its T-Mobile holdings—56.9 million shares in the first three quarters of fiscal 2025, raising $12.7 billion; another 12.5 million shares in Q4, raising another $2.3 billion. Deutsche Telekom was completely sold off. Alibaba was completely sold off. At the end of April this year, SoftBank began organizing a $10 billion margin loan secured by OpenAI equity, with an interest rate as high as nearly 8%.
At the same time, SoftBank has been issuing bonds everywhere: In November 2025, SoftBank issued a 500 billion yen bond with a 3.98% coupon; in April 2026, it followed with a 418 billion yen, five-year 4.97% coupon subordinated bond—the most expensive retail bond in SoftBank's history and the highest coupon in the history of Japanese yen retail bonds for non-financial corporations. This shows that the market is already starting to become "suspicious" of SoftBank's debt.
The credit market's reaction is direct: SoftBank's 5-year credit default swap surged to 355 basis points in early March, an 11-month high.
Son's recent "lifeline" is hoping OpenAI can go public as soon as possible. Otherwise, if debt pressure becomes chronic, it could really blow up SoftBank.
However, while OpenAI CEO Sam Altman advocates for a Q4 2026 listing, CFO Sarah Friar advocates for a delay to 2027. The fact that the CEO and CFO are publicly split on this matter itself tells the market: the company internally isn't sure if it's ready.
One Must Die
Berkshire's death will be gentle.
It won't go bankrupt—its subsidiaries are high-quality cash cows, its debt level is extremely low. Even if the AI capital expenditure bubble bursts, even if data center demand is cut in half, even if the S&P 500 falls 50%, Berkshire's cash pile is enough for it to devour any cheap thing like a glutton over a decade.
Its death is the death of identity—the Buffett-style compound interest myth of "buying good businesses at cabbage prices" may no longer hold in a world where valuations are forever above 30 times P/E, in a world where ten-year Treasury yields have shattered all traditional valuation models.
Abel may perform very well as a "rational CEO"—continuing operations, continuing small buybacks, continuing to make some marginal acquisitions—but Berkshire as a narrative of capitalist discipline died the moment Buffett stopped writing letters. Its body remains, but its soul has left the stage.
SoftBank's death, however, could be violent. It has three death triggers; pulling any one could set off a chain reaction:
The first trigger is OpenAI. If its IPO is delayed until 2027 or even 2028, if Amazon's $35 billion commitment tied to the IPO ultimately fails to materialize, if OpenAI's revenue growth stops for a single quarter—even just one quarter—the valuation of that 13% stake on SoftBank's balance sheet will be revalued downward.
The second trigger is Arm. Arm is currently SoftBank's only truly liquid asset that is still highly valued by the market—market cap about $200 billion, with SoftBank holding 87%.
Arm's royalty revenue rose 26% year-on-year in Q3 FY2026, with data center-related royalties doubling, one of the pillars supporting SoftBank's entire valuation story.
But Arm is also currently the most susceptible target for re-pricing—once it returns from its current forward P/E of 70 times to a "normal" semiconductor company valuation, the coverage ratio for SoftBank's $20 billion Arm-collateralized loan will collapse.
The third trigger is the refinancing itself. The $40 billion bridge loan matures in March 2027. Before that, SoftBank must achieve at least one of the following: OpenAI's IPO, Roze AI's IPO, sell another batch of assets, or issue a bond of similar size to roll it over.
But each path is more expensive than the last—SoftBank's retail bond coupon has already risen from 3.98% in 2025 to 4.97% in 2026.
The probability of each of these three triggers individually is not high—OpenAI will likely go public, Arm likely won't crash immediately, and credit markets likely won't shut their doors to SoftBank overnight. But they have one non-negligible characteristic: they are highly correlated, not three independent events.
If the bubble doesn't burst, Son will ultimately be deified: Roze AI lists at a $100 billion valuation, OpenAI successfully IPOs, and in his 70s he will realize his proclaimed ASI (Artificial Superintelligence) narrative.
And Berkshire, under Abel's prudent management, will be gently, persistently, irreversibly marginalized by the market—until one day, a successor is forced to do what Buffett refused to do his entire life
