In February we wrote about Apple's record quarter and Warren Buffett's steady exit from it, and we ended on a line we meant: "When the most famous buy-and-hold investor in history spends two years methodically reducing his largest position, the selling is the statement." The piece tracked the position from roughly $175B at its 2023 peak down to about $60B.
Then, in our own six-month capex update, we reported that Apple overtook Nvidia on July 17 to become the world's most valuable company, up 22% year-to-date, and that its restraint on AI capital spending was the reason. We published the answer to our own February question in a different article and did not connect them.
Berkshire Hathaway reports on Monday, August 3. This is the follow-up, and it comes with a complication we also never covered: the man who did the selling no longer runs the company.
Was he wrong? Not a question five months of price action answers
The tempting move is to put Apple's run next to Buffett's exit and call it a bad trade. We are not going to do that, because it would be a category error dressed as a verdict.
Selling down a position over two years is a capital-allocation decision about concentration, valuation and what else the money could do. Judging it by where the stock went in the following five months tests a claim nobody made. Buffett's own record contains several positions that kept rising after he left them and several he held through drawdowns that would have looked like errors on any five-month view.
What we can say precisely is what changed. Apple's most recent quarter — covered in our analysis of the print — was a genuine beat, revenue $109.42B and up 16%, with a caveat: roughly $0.11 of the $2.02 diluted EPS came from a one-time tariff refund. And the reason Apple was rewarded is the one our capex piece identified: it was the only member of the group whose free cash flow was not being consumed by data-centre construction. Restraint was the thesis on both sides of this trade. Buffett was reducing exposure to a company the market later rewarded for restraint.
That is a more interesting outcome than right or wrong, and it is the one the record actually supports.
The number most likely to be misread on Monday
Here is Berkshire's recent record, from our stored Berkshire data:
| Quarter | Revenue | Diluted EPS | P/E |
|---|---|---|---|
| 2025 Q3 | $94.97B | $14.27 | 16.08 |
| 2025 Q4 | $94.23B | $8.90 | 16.21 |
| 2026 Q1 | $93.67B | $4.68 | 14.25 |
Revenue is flat to slightly down across three quarters. Reported earnings per share fell by roughly two-thirds. Read as a normal industrial company, that is a collapse.
It is not one, and it is important to say why before Monday rather than after. Berkshire's reported earnings run the mark-to-market movement of its equity portfolio through the income statement. When the holdings move, reported EPS moves with them, whether or not a single share was sold and whether or not any operating business did anything differently. Buffett spent years telling shareholders to ignore that line for exactly this reason.
We have unpacked this same artifact twice already on other companies — an unrealized equity gain inflating one quarter's earnings at Alphabet, and a multi-billion mark-to-market charge on a government stake at Intel. Berkshire is the purest case of it on the site, because the portfolio is the point.
The 2018 break, which is when this became true
The clearest evidence that Berkshire's EPS stopped describing its operations is in its own series. Splitting the 45 quarters of reported EPS at 2018:
| 2015–2017 (12 quarters) | 2018 onward (33 quarters) | |
|---|---|---|
| EPS standard deviation | $2.68 | $9.21 |
| Quarters with negative EPS | none | six |
| Lowest quarter | $1.63 | −$20.42 |
Standard deviations are population, not sample; on a sample basis they are $2.80 and $9.35, and either way the volatility roughly tripled. The six loss-making quarters are 2018 Q1, 2018 Q4, 2020 Q1, 2022 Q2, 2022 Q3 and 2023 Q3 — and there is not one before 2018. Revenue shows no comparable break. A company whose operations were deteriorating would show it in both lines; Berkshire shows it in one.
One thing we will not do is state the cause from memory. The data establishes when the break happened, not why. The timing is consistent with the accounting change that routes unrealized equity gains through the income statement, but that is a lead rather than a fact we have sourced, and Berkshire's own filings are where it would be settled — the FY2025 10-K and the most recent 10-Q are both indexed in this repo. We are describing a dated pattern and pointing at the documents, which is as far as our own data takes us.
Two comparisons we deliberately avoided, because they would mislead in opposite directions. We did not use a high-to-low spread to contrast the eras: the 2015–2017 spread is 7.23×, which is wider than the post-2018 figure looks, and the post-2018 ratio is undefined because the values cross zero — the natural framing reverses the conclusion. And the Rule of 40, the metric this site is named after, does not apply to a holding company whose earnings swing on portfolio marks; we have written up where that score does not apply and Berkshire belongs in that category, so we are not scoring it here.
It is not Buffett's Berkshire any more
This is the part our own coverage has entirely missed. Per the company history in our data, Warren Buffett retired as chief executive at the end of 2025 and Greg Abel succeeded him on January 1, 2026. Monday will be Abel's third quarter running the company.
This site has never mentioned Greg Abel. Not once, in any article. A chief-executive succession at a company of this size, seven months ago, went uncovered here — and our own ticker page still lists Buffett as Chief Executive Officer in one block while the company history two fields away says he retired. We have corrected that page alongside this piece.
It also sharpens the February story rather than replacing it. The selling was not just a large investor trimming a position; it was, on the timeline in our own data, among the last major capital-allocation moves Buffett made as chief executive. Whatever the exit was arguing, the person who made the argument has handed over the desk, and the results now belong to someone else.
What to watch on Monday
Everything above is reported data through 2026 Q1. What follows are open questions, not forecasts — we are publishing no expected revenue or earnings figure, because we have no consensus number we can source. Berkshire's ratings block here holds no average rating and no average price target at all, and its newest entry is dated mid-2024, so there is nothing to quote even if we wanted to.
- Whether the Apple position moved again. We cannot tell you the current stake. It comes from 13F filings, which this site does not hold, and we are not going to infer it from the share price. If the position changed materially, Monday is where it becomes visible.
- Operating earnings, not reported EPS. The line worth reading is the one that excludes portfolio marks. If reported EPS falls again while operating earnings hold, that is the 2018 pattern repeating, not a business turning down.
- What Abel says, and how much of it sounds like a change. Three quarters in is early, but it is no longer new. Capital allocation under a different chief executive is the genuinely open question at this company, and it has nothing to do with Apple.
- The cash position. Berkshire's optionality has been the bull case for years; whether the new management deploys it differently is the thing the next several years turn on.
The bottom line
Our February call was that the selling was the statement, and we still think the selling said something. What it said is harder than "Apple is overvalued", because Apple then went on to become the most valuable company in the world by doing the thing Buffett has always praised — refusing to spend heavily into a boom. The trade and the thesis point in different directions, and five months of price action does not settle a two-year decision either way.
The part we got wrong was not the call. It was that we made it, published the resolution to it in a separate article five months later, and never told the reader they were the same story — about a company we had, until today, never written about at all.
Berkshire figures are from our stored Berkshire data: a 45-quarter revenue, EPS and P/E series through 2026 Q1, and the company history recording the chief-executive succession. Standard deviations are population figures over the split described. Apple figures are from our own Q3 FY2026 analysis and from the February piece linked above. Berkshire's current Apple stake is not held in this repository and is not asserted here; it is disclosed in 13F filings. No consensus estimate is quoted because the stored ratings block contains none and its newest entry predates this fiscal year.