The largest holding is not a business Berkshire runs
The quoted portfolio is the single biggest line in the valuation, larger than the railroad, larger than the utilities, larger than all seventy manufacturing and retail businesses combined. It is also the easiest line to value, because the market prices it every day.
Berkshire's largest single holding is not a business it runs, and over the past twenty years the quoted portfolio has stopped being a use of the company's cash and become a source of it.
- 1 $324 B, carried at market. That is the reported figure at June 30, 2026, up from $298 B at year-end 2025. At the year end it was 28% of the base case, more than MSR ($204 B) and BNSF ($110 B) put together. No multiple is being argued over; the market sets the price.
- 2 What you own is $17.8 B of look-through earnings against $4.7 B of dividends. Berkshire’s share of the profits of the companies it part-owns, not the cash it collects. On the carried value that is an earnings yield of 5.6%, the portfolio trades at 18× what it earns, and only 26% of it arrives as cash.
- 3 66% of it sits in five companies. Apple, American Express, Alphabet, Coca-Cola and Bank of America. The other 24 positions share what is left, and the first half of 2026 was the busiest six months in the portfolio’s history: $39.4 B bought and $27.8 B sold.
The one line that needs no estimating
Every other operating business in this valuation requires a judgement about what its earnings are worth. The quoted portfolio does not. It consists of shares in listed companies, and the closing price on December 31, 2025 is what it was worth that day: $298 B.
Two adjustments follow from that. First, a deferred tax liability of $41 B, tax owed on unrealised gains, payable only if the positions are sold, so it is carried as a reduction instead of ignored. Second, in the downside case, a 10% haircut of $30 B, because a portfolio marked to today’s market can be marked to a worse one tomorrow.
A multiple appears anyway, and it is there to be argued with. The valuation applies 20× to $15 B of look-through earnings, which comes to $300 B, close to but not exactly the $298 B market value. The published figure is the market value; the multiple makes the assumption visible. Moving the slider from 14× to 26× scales the carried value accordingly.
A 13F does not show you the whole portfolio
Two numbers are published for the same quarter and they are $24.5 B apart: the Form 13F reports $299.3 B, the balance sheet $323.8 B. Neither is wrong, and the chart below walks between them. One of the three differences is worth a reader’s attention and the other two are bookkeeping.
Berkshire’s Japanese holdings appear in no 13F at all. A 13F covers section 13(f) securities, which in practice means US-exchange listed, so anyone building the portfolio from 13F data alone is silently missing 11.9% of it, of which 94% is the five sogo shosha. This is the adjustment most readers never make, and it is the reason the figure on this page is built from the balance sheet.
The other two differences are bookkeeping and point opposite ways. The $8.6 B of Occidental preferred is in no 13F but sits inside the equity securities line; Kraft Heinz and Occidental common are in the 13F but are held too large to be marked to market, so they sit on a separate $19.9 B line.
Cost is the trading record; fair value is the market’s. Fair value moves every day. Cost moves only when Berkshire buys or sells, so the change in cost since year-end 2025 is what was done. Commercial, industrial and other went from $58.0 B of cost to $82.1 B, a $24.1 B increase, most of the $39.4 B Berkshire spent in the half, and the bucket Alphabet lands in. Consumer products did the opposite: cost fell $3.2 B while value rose $5.5 B, which is what trimming a long-held winner looks like in this table.
| Investee | Owned | Carrying | Market | Look-through |
|---|---|---|---|---|
| Kraft Heinz | 27.5% | $8,760m | $7,692m | ($933)m |
| Occidental | 26.7% | $10,727m | $12,868m | $1,932m |
| Berkadia | 50.0% | $461m | — | — |
| Equity method investments | — | $19,948m | $20,560m | $1,000m |
Kraft Heinz is the one carried above its market price, by $1,068m, or 12.2% of the carrying amount. Berkshire’s representatives left its board in May 2025, so its share of those earnings is now reported a quarter late.
What the portfolio actually holds
The thing to notice first is how little of this is a portfolio in the ordinary sense. There are 29 positions carried at market and the largest five are 66% of the reported line. The 17 positions outside the largest 12 come to $30.4 B between them, less than the Alphabet stake Berkshire did not own at all a year ago.
Three of these positions would be the whole story at most investment firms. American Express is 22.2% of American Express, bought for $1,287m and now worth $51.3 B (40 times the money), and Berkshire has not sold a share of it in this entire record. Coca-Cola, at $1,299m of cost, is the same story. DaVita is 45% of a listed company.
| Holding | Value | Weight | Owned | P/E | Earnings | Dividend |
|---|---|---|---|---|---|---|
| Apple | $66.0 B | 20.9% | 1.6% | 33.2× | $2,001m | $243m |
| American Express | $51.3 B | 16.3% | 22.2% | 20.5× | $2,723m | $555m |
| Alphabet | $37.8 B | 12.0% | 0.9% | 34.5× | $2,131m | $90m |
| Coca-Cola | $32.5 B | 10.3% | 9.3% | 24.9× | $1,331m | $1,028m |
| Bank of America | $27.5 B | 8.7% | 6.8% | 13.1× | $2,196m | $665m |
| Chevron | $14.0 B | 4.4% | 4.2% | 16.2× | $872m | $604m |
| Chubb | $11.7 B | 3.7% | 8.8% | 11.8× | $992m | $135m |
| Moody’s | $11.2 B | 3.5% | 14.1% | 28.4× | $407m | $100m |
| Mitsubishi Corp | $10.7 B | 3.4% | 10.6% | 20.9× | $583m | $266m |
| Mitsui & Co | $8.2 B | 2.6% | 10.2% | 15.5× | $589m | $206m |
| Itochu | $8.0 B | 2.5% | 8.9% | 14.5× | $497m | $151m |
| DaVita | $6.4 B | 2.0% | 45.0% | 17.2× | $381m | — |
| All 29 positions | $315.6 B | 100.0% | — | 17.7× | $17,786m | $4,700m |
The other 17 positions
| Holding | Value | Weight | Owned | P/E | Earnings | Dividend |
|---|---|---|---|---|---|---|
| Delta Air Lines | $5.4 B | 1.7% | 8.7% | 16.9× | $287m | $45m |
| Marubeni | $4.6 B | 1.5% | 9.6% | 14.3× | $339m | $104m |
| Sumitomo Corp* | $4.4 B | 1.4% | 9.7% | 12.3× | $371m | $108m |
| Sirius XM | $3.7 B | 1.2% | 37.1% | 8.7× | $1,439m | $135m |
| Tokio Marine* | $2.5 B | 0.8% | 2.9% | 18.0× | $137m | $75m |
| VeriSign | $2.3 B | 0.7% | 9.9% | 27.3× | $84m | $29m |
| Kroger | $2.2 B | 0.7% | 6.4% | 11.1× | $69m | $57m |
| Ally Financial | $1.2 B | 0.4% | 8.8% | 10.1× | $131m | $34m |
| Lennar | $1.2 B | 0.4% | 5.6% | 16.6× | $73m | $27m |
| New York Times | $1.1 B | 0.3% | 9.7% | 27.0× | $38m | $12m |
| Capital One | $0.6 B | 0.2% | 0.5% | 9.9× | $52m | $10m |
| Louisiana-Pacific | $0.4 B | 0.1% | 8.1% | 68.2× | $4m | $7m |
| Nucor | $0.4 B | 0.1% | 0.8% | 19.8× | $24m | $4m |
| Macy’s | $0.2 B | 0.1% | 2.8% | 10.6× | $21m | $6m |
| Constellation Brands | $0.1 B | 0.0% | 0.4% | 12.8× | $7m | $3m |
| NVR | $0.1 B | 0.0% | 0.4% | 18.7× | $4m | — |
| Jefferies | $0.0 B | 0.0% | 0.2% | 13.9× | $2m | $1m |
| Occidental preferred | $8.6 B | — | — | — | $680m | $680m |
What these columns are. Value is what the position was worth at June 30, 2026, and Weight is its share of the portfolio. Owned is the percentage of the company Berkshire holds. P/E is the share price against earnings per share. Earnings and Dividend are Berkshire’s share of the company’s net income and of the dividends it paid over the twelve months ended June 30, 2026. An asterisk marks Sumitomo and Tokio Marine, the two positions carried on a corrected basis.
Nearly four dollars earned for every one collected
Berkshire’s income statement records the dividends the portfolio pays it. Over the trailing year that was $4.7 B. But Berkshire’s share of what those companies actually earned was $17.8 B, an earnings yield of 5.6% on the carried value against a dividend yield of 1.5%. Both figures are as-reported GAAP results for the twelve months ended June 30, 2026, taken whole, unadjusted, so a year with a large write-off in it shows up here.
The difference is not an accounting curiosity. 26% of those earnings reach Berkshire as cash; the other 74% stays inside Apple and Amex and Alphabet and is reinvested there. A shareholder who reads only the income statement sees the smaller number and concludes the portfolio earns 1.5%. What is owned is the larger one.
Where the earnings come from is not where the value sits. Apple is 21% of the portfolio and 11% of its earnings. American Express is 16% of the value and the largest single contributor at 15%. Bank of America, now less than half the position it was two years ago, still earns more for Berkshire than Apple does.
Why this is the number the valuation uses. The sum of the parts applies a multiple to look-through earnings, not to dividends received, because a dollar retained by Apple and reinvested at Apple’s returns is worth at least as much to a Berkshire shareholder as a dollar paid out and reinvested at Berkshire’s. That is Buffett’s own argument from the 1980s letters, and it is the reason the portfolio is worth more than its dividend stream implies.
The five Japanese trading houses
This is the position that looks least like the rest of the portfolio and most like Berkshire itself: five conglomerates of operating businesses and minority stakes, accumulated from 2019 and funded largely by issuing yen-denominated bonds, so the position carries its own currency hedge. What the five are and how they work has its own section of this site, the sogo shosha. What matters here is what they do to Berkshire’s portfolio.
At June 30, 2026 the five were worth $35.9 B against $15.4 B of cost, 2.3× the money, and Berkshire’s share of their earnings was $2.4 B. That last figure is the one worth sitting with: the trading houses are 11.4% of the portfolio by value but 13.4% of its earnings. They earn more than their size because they are among the cheapest things Berkshire owns, the five together on 15× earnings against 18× for the portfolio as a whole, and they pay out 35% of what they earn.
Berkshire said it intended to go above the 10% ceiling it had originally set itself, and it has: Mitsubishi and Mitsui are both over it now. It has also bought something new. Tokio Marine, the Japanese property and casualty insurer, first appears in the first quarter of 2026, the first Japanese holding outside the five, and an insurer bought by an insurer.
| Company | Value | Cost | Owned | P/E | Earnings |
|---|---|---|---|---|---|
| Mitsubishi Corp | $10.7 B | $4.2 B | 10.6% | 20.9× | $583m |
| Mitsui & Co | $8.2 B | $3.5 B | 10.2% | 15.5× | $589m |
| Itochu | $8.0 B | $4.2 B | 8.9% | 14.5× | $497m |
| Marubeni | $4.6 B | $1.6 B | 9.6% | 14.3× | $339m |
| Sumitomo Corp* | $4.4 B | $1.9 B | 9.7% | 12.3× | $371m |
| Five houses | $35.9 B | $15.4 B | — | — | $2,380m |
| Tokio Marine* | $2.5 B | $2.5 B | 2.9% | 18.0× | $137m |
The takeaway Buy, never sell, and let the dividends pay the coupons. The trading houses are the clearest surviving example of how Berkshire used to buy, a large stake in an unglamorous business at a single-digit multiple, financed in the currency of the asset, and then left alone for seven years. The rest of the portfolio has turned over heavily in that time. This part has not.
What was bought and what was sold
A portfolio total tells you what the market thinks. The share count tells you what Berkshire decided. Every position’s holding is recorded here quarter by quarter from September 2019 to the present, accumulated forward from the 13F deltas, so it ties to today’s holding exactly. The three positions below are carried back further, to 1Q18, because the decision that matters most started before the record does.
Two decisions dominate it. Apple peaked at 1,010 million shares and is now 228 million, Berkshire sold 77% of the largest position any investor has ever held, most of it in two quarters of 2024. Bank of America peaked at 1,033 million in 2Q21 and is now 483 million, and unlike Apple it is still being sold: it has fallen in 8 consecutive quarters and did so again in both quarters of 2026.
Wells Fargo is the third line, and the only one that reaches zero. It was a thirty-year holding and one of the positions most associated with Berkshire. The line opens at 457 million shares in 1Q18 and is already falling: the selling had begun before this record does, and what the chart shows is the back half of a position that was close to half a billion shares. It reaches nothing by 1Q22. It is included because a disposal that finishes looks different from one still running, and both Apple and Bank of America are still running.
And what replaced them
The cash did not sit still. In the six months to June 30, 2026 Berkshire bought $39.4 B of equities and sold $27.8 B, the heaviest buying in the record, and almost all of it one name. Alphabet went from nothing in mid-2025 to 106 million shares and $37.8 B, which makes it the third largest holding in the portfolio three quarters after the first share was bought. Nothing in this record was built that fast.
Delta Air Lines is the stranger entry. Berkshire owned the four large American carriers and sold all of them in April 2020, with Buffett saying publicly that he had been wrong to buy them. Delta is back, at 8.7% of the company. So is a cluster of smaller new positions: the New York Times, Macy’s and Tokio Marine, none of them large, all of them bought in the last three quarters.
And on the other side, Chevron fell from 130 million shares at the year end to 84 million, all of it in one quarter, while Constellation Brands went from a 2Q25 peak of 13 million shares to 0.6 million, bought in 2024 and substantially gone by 2026, which for Berkshire is a fast admission.
Buying and selling, 2006 to 2025
Stepping back from the positions to the cash: the chart below shows net purchases and sales each year, below the line is buying, above it is selling.
Three episodes stand out. 2008 and 2011, when Berkshire bought heavily into weakness. 2022, the most aggressive buying since the crisis. And 2024, when it sold $134.1 B net, a single year larger than the twenty-year total of either acquisitions or buybacks.
The takeaway The selling was the signal, and it has stopped. Berkshire does not hold cash by accident, and $172.1 B of net sales since 2023 was the same decision, seen from the other side, as the $369 B of Treasury bills on the balance sheet. The first half of 2026 turned that around: $11.6 B went back into equities, most of it into one company. The upside case in this valuation turns on what the rest of that cash gets deployed at.
What the portfolio does to the valuation
Because the portfolio is 28% of the base case and carried at market, a reader buying Berkshire is buying a quoted portfolio at its quoted price, plus a group of operating businesses, plus a cash pile. The first of those three needs no faith at all.
That is also the argument against paying a large premium to book value for the whole company. 62% of the estimated value is already marked to market or held in Treasuries, assets whose value is known, not estimated, so there is less room for the valuation to be pleasantly wrong than a conglomerate usually offers.
The one place that reasoning does not reach is the $19.9 B of equity method investments, which the sum of the parts carries at $20 B on a separate line. Those are not marked to anything: Kraft Heinz is carried 12% above its market price and Occidental 20% below it. They happen to roughly offset, which is luck, not method.
Two quarters on, it earns twice as much more as it is worth
The sum of the parts carries the portfolio at $298 B on $15 B of look-through earnings. Both have moved since, and not by the same amount.
| Year-end 2025 | June 30, 2026 | Change | % | |
|---|---|---|---|---|
| Market value | $297.8 B | $323.8 B | $26.0 B | 8.7% |
| Look-through earnings | $15.0 B | $17.8 B | $2.8 B | 18.6% |
| Cost basis | $85.4 B | $106.5 B | $21.1 B | 24.7% |
| Unrealised gain | $212.4 B | $217.3 B | $4.9 B | 2.3% |
Market value is up $26.0 B, or 8.7%, and the earnings behind it are up $2.8 B, 18.6%, better than twice the rate of the price. That is the more interesting half: the portfolio did not simply re-rate, it rotated into companies that earn more per dollar invested, which is what buying $37.8 B of Alphabet while selling Apple and Bank of America does to a weighted average. The earnings yield on the carried value is 5.6% against the 5% the valuation assumes.
The effect on the valuation Rolling the portfolio forward would add roughly 2.3% to the base case. The $26.0 B of extra market value is worth about 2.4% of the $1,069 B estimate on its own, less about $0.9 B of additional deferred tax on the larger unrealised gain, call it $12 a share. It is directional, not a restatement: the other lines in the sum of the parts have moved too, and they are only re-struck once a year on a full annual report. The published figures stay where they are until then.
Download the series (.xlsx) Berkshire quoted equity portfolio, with sources and the date it was struck written into the file. This page’s series only; there is no combined download.