Everything has been said about Berkshire's cash pile. 397 billion dollars at the end of the first quarter of 2026, a record, larger than the GDP of Norway. The figure supports almost no statement, though, because it grows with the company and arrives without a comparison.
The more interesting question sits one level down. To build that pile, stock had to be sold — between 2022 and 2024 alone a net 173 billion dollars, 134 billion of it in 2024. What did that cost?
That question has a comparison built in: what would have happened if nothing had been done. Precisely the number that is almost always missing from a return figure.
It has also been answered already. Several times. With very different results.
Three publications, one event
[GRAFIK]
All three describe the same sale of the same shares over the same period. Between the smallest and the largest there is a factor of 3.3.
None of the three states which decisions its number assumes. And in none of them is the interest earned on the sale proceeds deducted.
A fourth report cites 330 billion. That is a different quantity — the value of the position had nothing been sold since 2018. A valuation, not a forgone gain. That is not stated either.
What the filings say directly
Before calculating anything, it is worth looking at what Berkshire itself reported. These are not third-party estimates.
From the third-quarter 2024 filing: sales of equity securities produced taxable gains of 97.1 billion dollars in the first nine months of 2024. In the comparable period of 2023 the figure was 5.4 billion.
From Buffett's 2024 shareholder letter: Berkshire made four payments to the IRS that year totalling 26.8 billion dollars — by his account around five percent of everything corporate America paid. Barron's attributes more than 20 billion of that to the Apple sales.
Also from the letter: holdings of marketable equities fell during the year from 354 billion to 272 billion dollars.
And, decisive for the calculation, a sentence from Buffett himself: they were aided by a predictable large gain in investment income as Treasury bill yields improved and they substantially increased their holdings of those securities.
So the man whose forgone gain is being calculated names, himself, the item missing from all three publications.
The biggest error: gross against net
Saying "the position would be worth 241 billion today, it is only worth 74 billion, so 167 billion was lost" compares two numbers that do not measure the same thing.
The 241 billion is a gross figure. Embedded in it is a deferred tax liability that would come due on sale. With a cost basis around 35 dollars a share and a price far above that today, the liability is substantial.
The 74 billion on the other side is what remains after a sale that actually happened and was actually taxed.
Gross against net. That is not a detail. It is the largest single item in the whole calculation.
Done properly it is either both sides pre-tax — in which case the tax paid has to be added back to the sale scenario — or both sides after tax, in which case the hold scenario carries a deferred liability to be deducted. Mixing them does not work.
Incidentally, Buffett explicitly justified the timing of the sales by expecting higher tax rates in future. The tax was not a side effect of the decision, it was part of it. Leaving it out measures the decision against a target it never had.
The second missing item: what the money earned
The proceeds did not sit in a current account. They went into short-dated government paper yielding somewhere between roughly 3.7 and five percent depending on the period.
At the end of 2025 that meant around 305 billion dollars in such securities, which at prevailing rates produces roughly eleven billion a year. For the first quarter of 2026, reports cite annualised interest income approaching twelve billion.
Over two years that adds up to an amount which, against a claimed loss of 50 billion, is no longer negligible. It is the return on the alternative — and without it the raw figure is not a statement but half a calculation.
The complete form
Forgone gain = (value if held − value of the actual position) − tax paid − interest earned on the proceeds
With the addition that the tax does not disappear in the hold scenario, it is merely deferred, and therefore belongs on both sides.
The only one of the three publications that accounts for any of these items arrives at around 50 billion. The other two, at 130 and 167, compute the gross difference and call it a loss.
The difference between those numbers is not opinion. It is the sum of two items you can either include or leave out.
The end date is a decision too
Apple is currently trading near its high. The same calculation at a different date produces a different result, and in some stretches a different sign.
Calculating today because today is the least flattering moment for Berkshire is choosing the horizon after seeing the outcome. Where only one date is shown, it is worth asking what the number looks like at the others.
A defensible presentation shows the path, not a point.
What the calculation cannot capture
Two objections that hold, and therefore belong here.
Concentration. The Apple position was at one point worth more than the entire rest of the equity portfolio combined. Reducing that is a risk decision, not a return bet — and it was justified that way at the time, not in hindsight. Judging a risk decision by its return measures it against the wrong target. Berkshire still holds roughly 22 percent of its equity portfolio in Apple; this was never an exit.
Optionality. Cash is the right to buy into a dislocation. A single-date calculation does not price a right. As long as no dislocation arrives, it looks worthless — that is the nature of such positions, not a finding about them.
And Buffett himself rejected the reading that the cash pile is a market view: the great majority of the money remains in equities, and Berkshire will never prefer cash-equivalent assets over good businesses.
That is the other side, and it deserves to be taken seriously.
What remains
A real forgone gain remains. It is simply much smaller than the circulating figure, and the difference consists of two items anyone can recompute.
What is striking is something else. The most famous investor in the world is being held to a metric built by exactly the rules he has argued against his whole career: a raw number without a comparison, without deducting costs, at a date chosen after the outcome was known.
The same calculation, done cleanly, produces a smaller and more interesting number. It just does not appear in any of the reports.
Three questions for the next figure like it
Gross or net — and the same on both sides? A gross value against a net value is not a comparison.
What did the alternative earn? Without that deduction the number is the gross difference, not the forgone gain.
Why this date? If only one is shown, ask about the others.