Berkshire Hathaway and SoftBank will "die" together

May 2, 2026, Omaha. Out of 18,000 seats, only a little over half were filled this time—In the past, Berkshire Hathaway’s annual shareholder meeting required queuing for tickets, and hotels outside the venue were fully booked.

This time, 95-year-old Buffett did not host from the stage as in previous years. New CEO Greg Abel stood in front of the main screen, answering investors’ questions about why Berkshire Hathaway holds nearly $400 billion in cash.

In the same week, 6,000 miles away in Tokyo, SoftBank’s team was doing something else: packaging its unprofitable AI assets into a new company called Roze AI, aiming for a valuation of $100 billion, planning to go public in the United States in the second half of 2026.

The reason is simple: SoftBank must keep looking for funding and investing to support the $64.6 billion check for OpenAI, which could eventually grow close to $100 billion.

One holds $397.4 billion in cash and waits for the market to crash without buying anything; the other, burdened with JPY 16.34 trillion (over $163.4k) in interest-bearing debt, bets that the market will not collapse.

Berkshire: Too much money is a disease

What does $397.4 billion mean?

It is nearly 40% of Berkshire’s total market value, twice the average cash level Buffett has maintained over the past twenty years.

Of this, JPY 339.3 billion is directly held in U.S. short-term government bonds. Berkshire has also become one of the largest non-government creditors to the U.S. Treasury.

This pile of money was not passively accumulated; it was actively built.

Over the past fourteen quarters, Berkshire has been net selling stocks each quarter. Its former top holding, Apple, has been continuously reduced since Q3 2024, selling nearly 688 million shares, realizing over $121.4k.

Buffett’s consistent explanation has been the same: he can’t find cheap things.

In his 2024 shareholder letter, he wrote—“Generally speaking, there’s nothing that looks attractive.” In a recent interview outside the annual meeting, he compared the current market to “a church connected to a casino,” saying this is the most gambling-like market he has experienced.

The problem is, he has been making this judgment for over a year.

Berkshire’s stock performance over the past twelve months lagged the S&P 500 by about 40 percentage points. This is not a small figure—it’s one of the largest relative declines since Buffett took over Berkshire in 1965. The last time it underperformed to this extent was during the final stages of the 1999 dot-com bubble.

Back then, Buffett said he would only buy “bricks, carpets, insulation, and paint—these kinds of basic industries.” Two years later, the Nasdaq fell 78%, and he was proven right.

But this time, investors have been waiting for more than two years. When the market rises, Berkshire remains still; when it rises again, Berkshire still does not move. On January 1, when Buffett officially stepped down as CEO, Berkshire’s stock price dipped slightly—an expression of restrained disappointment in the most restrained way.

This is the situation Abel inherited. He is a Canadian accountant who rose to become Berkshire’s vice chairman, having spent his life in heavy-asset, regulated, slow-growth industries like utilities, railroads, and energy.

He is not Buffett, and he knows it. In his first shareholder letter, he repeatedly emphasized “continuity” and “decentralization.” At his first shareholder meeting, his response to all suggestions of splitting the group was an absolute “impossible.”

Abel’s dilemma: he cannot use this cash (because the market is too expensive), nor can he pretend it doesn’t exist (because investors are voting with their feet).

If the market continues to rise for the next five to ten years, he will eventually face a question never seriously discussed in Berkshire’s history—returning the money. Either through special dividends to shareholders or by truly拆开卖掉这家由60多家子公司缝合起来的怪兽。

Will Berkshire die? Not suddenly. Its assets are too diversified, cash too abundant, and debt too low. Any external shock would be difficult to truly break it. But it will slowly and gracefully become something else.

SoftBank’s problem: too little money and still gambling

Sun Jiyu’s situation is a mirror opposite of Abel’s.

On February 27, 2026, SoftBank issued an announcement. The key sentence translates as: “SoftBank Group’s cumulative investment in OpenAI is expected to reach $64.6 billion, with an approximately 13% stake.”

$64.6 billion, 13%. This is one of the most expensive single bets of this era.

To understand how crazy this number is, look at how SoftBank can afford this.

SoftBank’s parent company’s interest-bearing debt soared from JPY 12.14 trillion in March 2025 to JPY 16.34 trillion in December 2025. The so-called cash on hand is about JPY 3.8 trillion, of which nearly one-third is unutilized committed credit lines, not actual cash.

Where did this money come from? SoftBank borrowed JPY 20 billion against its holdings of Arm stock; also, about $163.4k was raised from the pledge of its Japanese telecom subsidiary SBKK stock.

On March 27, 2026, SoftBank signed an unprecedented $40 billion bridge loan, led by JPMorgan Chase, Goldman Sachs, Mizuho, Sumitomo Mitsui, and Mitsubishi UFJ, later expanded to eight banks—one of the largest bridge loans in Asian history. Of this, JPY 30 billion was directly used to co-invest in OpenAI. The term is 12 months, meaning by March 2027, SoftBank must repay $40 billion.

This explains why Sun Jiyu has seemed somewhat “unusual” this year: he liquidated all of SoftBank’s Nvidia holdings, raising $5.8 billion in a one-time sale in October 2025. In a speech in Tokyo in early December 2025, he admitted, “I didn’t want to sell a share of Nvidia, but I needed more money to invest in OpenAI. I sold Nvidia crying.”

To fund OpenAI, Sun Jiyu has been “selling everything”: SoftBank sold 56.9 million T-Mobile shares in the first three quarters of fiscal 2025, raising $12.7 billion; in the fourth quarter, it sold another 12.5 million shares, raising $2.3 billion. It also cleared Deutsche Telekom and Alibaba holdings. By the end of April this year, SoftBank was again assembling a $10 billion margin loan collateralized by OpenAI shares, with an interest rate close to 8%.

At the same time, SoftBank has been issuing debt everywhere: in November 2025, it issued a JPY 500 billion bond with a 3.98% coupon; in April 2026, it issued another JPY 418 billion subordinate bond with a 4.97% coupon for the first five years—this is the most expensive retail bond in SoftBank’s history and the highest coupon for Japanese non-financial corporate retail bonds. The market has begun to doubt SoftBank’s debt.

The reaction in the credit market was immediate: SoftBank’s 5-year credit default swap spiked to 355 basis points in early March, a 11-month high.

Sun Jiyu’s latest “rescue” hope is that OpenAI can go public quickly; otherwise, if debt pressures persist long-term, SoftBank could really be in trouble.

However, although OpenAI CEO Sam Altman advocates for a Q4 2026 IPO, CFO Sarah Friar suggests delaying until 2027—public split between CEO and CFO signals internal uncertainty about readiness.

“Certain death”

Berkshire’s death would be gentle.

It won’t go bankrupt—its subsidiaries are high-quality cash cows, its debt levels are extremely low, and even if AI capital expenditure bubbles burst, data center demand halves, or the S&P 500 drops 50%, Berkshire’s cash pile is enough for it to feast on anything cheap for ten years.

Its death is an identity death—Buffett’s myth of “buying good businesses at bargain prices” with compound interest may no longer hold in a world where valuations are always above 30 times P/E, and where a ten-year Treasury yield has shattered all traditional valuation models.

Abel might perform well as a “rational CEO”—continuing operations, modest buybacks, small acquisitions at the margins—but Berkshire as a disciplined capitalist narrative has already died the moment Buffett stopped writing letters. Its shell remains, but its soul has left.

SoftBank’s death could be more violent. Its triggers are three, any one of which could trigger a chain reaction:

The first trigger is OpenAI. If its IPO is delayed to 2027 or 2028, or if the $35 billion commitment linked to Amazon’s IPO cannot be fulfilled, or if OpenAI’s revenue growth stalls for a quarter—even just one quarter—the valuation of its 13% stake on SoftBank’s balance sheet will be re-evaluated downward.

The second trigger is Arm. Arm is currently SoftBank’s only truly liquid and highly valued asset—market cap around $200 billion, with SoftBank holding 87%.

Arm’s licensing revenue grew 26% YoY in Q3 2026, with data center-related licensing doubling—one of the pillars supporting SoftBank’s valuation story.

But Arm is also the most susceptible to revaluation now—if its forward P/E ratio drops from the current 70 to a “normal” semiconductor valuation, the coverage ratio of SoftBank’s $20 billion mortgage on Arm will collapse.

The third trigger is refinancing itself. The $40 billion bridge loan matures in March 2027. Before then, SoftBank must do at least one of: IPO of OpenAI, IPO of Roze AI, sell more assets, or roll over the debt with another similar-sized loan.

But each option is more expensive than the last—SoftBank’s retail bond coupons have risen from 3.98% in 2025 to 4.97% in 2026.

Each trigger, individually, has a low probability—OpenAI is likely to IPO, Arm probably won’t collapse immediately, and the credit market probably won’t shut its door overnight. But they are highly correlated, not independent.

If the bubble doesn’t burst, Sun Jiyu will ultimately be deified: Roze AI will IPO at a $100 billion valuation, OpenAI will successfully go public, and he will realize his narrative of superintelligent AI (ASI) at age 70.

Meanwhile, Berkshire will be gently, steadily, irreversibly marginalized under Abel’s cautious management—until one day, a successor will have to do what Buffett has always refused to do: return the money to shareholders.