Downgrading to Underperform: GEICO Tripped Our Trigger, and an AI Rotation Lifted the Stock to 1.6x Book, 13% Above Where Berkshire Itself Was Buying

Results vs. Consensus

Berkshire publishes no operating EPS of its own and is thinly covered for its size, so "consensus" is a construct assembled by data providers. Two clustered figures circulated: $5.91 per Class B share from two providers, and $8,821.84 per Class A share (about $5.88 per Class B) from a third. Revenue expectations were wider, from $97.3B to $101.0B. Against those, operating earnings of $12,705M on 2,139.6M average equivalent Class B shares produce about $5.94, a result within 1% of every estimate. The quarter is not a beat or a miss. It is a composition story, and the composition is poor.

Q3 2026 Scorecard

The revenue beat is mostly gasoline. Pilot Travel Centers' revenue rose $5.26B on higher fuel prices, 49% of the consolidated $10.7B increase, at a 1.1% pre-tax margin. Excluding Pilot, revenue grew 6.4%, and much of that came from businesses Berkshire did not own a year ago. The GAAP line is noise by construction: investment gains of $9.41B after tax, versus $17.31B in the prior-year quarter, explain $7.9B of the $8.7B decline in net earnings (the rest is the $780M drop in operating earnings), and Berkshire itself calls the quarterly figure "usually meaningless."

Year-Over-Year Comparisons

Quarter-Over-Quarter Comparisons

Nine-Month Comparisons

Revenue. Total revenues rose 11.2% to $105.6B and beat every estimate, but the increase is low-grade. Pilot accounted for $5.26B of the $10.7B increase on fuel prices, with retail fuel volumes up only slightly. Taylor Morrison, acquired on July 24, contributed roughly $1.1B of home sales in about ten weeks, and OxyChem added $1.3B in a quarter that had no OxyChem a year ago. Insurance premiums earned, the revenue line that carries the most value per dollar, were flat at $22.4B, and GEICO's premiums written fell 0.3%. Organic growth in the businesses that matter most is close to zero.

Margins. The manufacturing, service and retailing collection's pre-tax margin slipped to 8.3% from 8.6%, a year-over-year compression after 0.8 points of expansion in Q2. Industrial products held a 19.2% margin against 19.5%, building products fell to 13.3% from 16.4% as Clayton's homebuilding margins compressed and input costs rose, and consumer products fell to 11.9% from 16.3% against a prior-year quarter that booked Duracell's manufacturing tax credits for 2023 through the third quarter of 2025 at once. In insurance, the margin damage is far larger: GEICO's combined ratio deteriorated 7.1 points, BH Primary's 8.1 points and property/casualty reinsurance's 5.1 points, in a quarter with no catastrophes on either side of the comparison.

Earnings per share. Operating EPS fell 5.0% against a 5.8% decline in operating earnings, the 0.8-point gap being the buyback, which retired 0.89% of the share count in a single quarter. That is the one per-share line moving in the shareholder's favor, and it is moving faster than at any point in our coverage. It is also price-contingent, and the price has since moved 13% above where management was buying.

Segment Performance

Insurance: Underwriting

In Q2 the commercial and reinsurance operations held the segment together while GEICO deteriorated. That support is gone. Management states it plainly:

"Each of our three underwriting groups generated lower underwriting earnings in the third quarter of 2026 relative to 2025."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Both Q3 2025 and Q3 2026 were free of significant catastrophes (the 2025 wildfires were a first-quarter event), so the comparison is clean. Our Q2 note argued that prior-year reserve releases were masking a much larger underlying decline. The third quarter lets us measure it directly, because the 10-Q's nine-month disclosures, set against the six-month figures in the Q2 filing, isolate the quarter.

GEICO: the trigger, and what the stable combined ratio hides

The headline combined ratio of 91.4% looks like stabilization after Q2's 91.2%. It is not. GEICO released $474M of prior-year reserves in the quarter, against $154M across the entire first half. Without that release, GEICO's loss ratio would have been about 80.7%, against 74.6% a year earlier, a gap of 6.1 points that is wider than the 4.9-point gap on the same basis in the first half. On an accident-year basis, the combined ratio was roughly 95.5%. The deterioration did not slow in the third quarter. It accelerated, and the reported number was held flat by a larger draw on past reserving conservatism.

The premium side tells the same story from another angle, and management now says so explicitly:

"Premiums written declined $30 million (0.3%) in the third quarter and increased $259 million (0.8%) in the first nine months of 2026 compared to 2025, reflecting continued growth in policies-in-force, offset by lower average premiums per policy."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

"The loss ratio increases reflected the impact of higher claims frequencies and average severities, in addition to lower average earned premiums per policy."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

That second sentence is new. In the Q2 filing, the loss-ratio increase was attributed only to frequency and severity. Now falling price per policy is named as a cause of the loss ratio itself. GEICO is growing its policy count while charging each customer less, into bodily-injury severity that is still rising 9% to 11% year to date. Frequency trends eased by about a point from the six-month ranges, which is the one encouraging datum. Underwriting expenses rose another 17.0%, and the filing attributes the nine-month expense-ratio increase mainly to commissions and advertising. That is a franchise buying share with price and marketing at the moment its loss costs are rising. In August we asked for a visible pricing response. The filing shows the opposite.

Assessment: Our published downgrade trigger read: "We would move the other way if the combined ratio deteriorates a third consecutive quarter without a visible pricing response." The combined ratio is worse year over year for the third straight quarter of 2026 (+7.5, +7.7 and +7.1 points) and the sixth straight since Q2 2025, worse sequentially (87.3% to 91.2% to 91.4%), and the only disclosed pricing datum is that average premium per policy is still falling. The trigger has fired, and we are honoring it as written. The accident-year arithmetic removes any temptation to argue that the sequential flattening is a turn. GEICO's structural advantage has always been the lowest cost base in the industry turned into price. Today it is spending more and charging less, and the gap between those two lines is coming out of Berkshire's most valuable earnings stream.

BH Primary and BHRG: the offsets turn into drags

BH Primary's pre-tax underwriting fell to $119M from $506M as its loss ratio rose 6.4 points to 66.7% and its prior-year releases shrank to $24M from $190M. Premiums written fell 7.9%, with year-to-date declines at RSUI (11.5%), BHHC (9.7%), MedPro (5.1%) and GUARD (14.5%). Property/casualty reinsurance fell to $779M from $1,065M, its loss ratio up 5.1 points to 57.0%, and management names the market condition directly:

"Otherwise, premium volumes in 2026 were constrained by increased price competition in U.S. property markets."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Reinsurance premiums written still rose 7.0%, helped by the Tokio Marine whole-account quota share, which has now written about $700M since April 1. In the run-off books, Berkshire commuted a 2016 retroactive reinsurance contract for a $1.12B payment with no material earnings effect, which shrinks float slightly but also retires a long-tail liability.

Assessment: Shrinking property books into a softening market is the right underwriting decision, and Berkshire's managers are instructed to do exactly this. But the arithmetic of a soft market is unforgiving: premiums fall, current-year loss ratios rise, and reserve releases from harder-market years eventually run down. BH Primary's release fell 87% year over year in the quarter. The Q2 note's support for the segment, "the reinsurance franchise is still winning the deals its capital entitles it to," still holds for Tokio Marine. It no longer holds for the segment's profit line.

Insurance: Investment Income

Interest income fell 13.9% as Treasury bills reinvested at lower short-term rates, and the effective tax rate on the line rose to 17.6% from 14.4%, which turned a 10.7% pre-tax decline into a 14.1% after-tax one. We said in August that dividends from Q2's roughly $19.8B of net equity purchases would begin offsetting the interest decline from Q3. They did not: dividend income fell 3.0%. Management's posture on the Treasury book is unchanged:

"We continue to believe that maintaining ample liquidity is paramount and insist on safety over yield with respect to short-term investments."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: This is the fifth consecutive year-over-year decline, and the rate of decline has accelerated for two straight quarters, from 7.4% in Q1 to 9.1% in Q2 and 14.1% now. The mechanism is simple and will persist as long as short rates drift lower: roughly $337B of Treasury bills reprice every few months. The dividend offset we expected has not appeared, and with equity buying having stopped in Q3, it will not grow from here. Float of $178.5B remains negative-cost, but the income earned on it is shrinking at a double-digit rate, and that changes the earnings power of the insurance segment as much as GEICO does.

Railroad: BNSF

BNSF produced the largest quarterly profit of our coverage window. Volumes rose 3.9%, average revenue per car or unit rose 10.5% on fuel surcharges and higher yield, and the operating ratio improved even though fuel expense rose 60.6%. Unlike Q2, the after-tax growth is not understated by a tax comparison: the effective rate was 23.7% against 24.2%. The line that proves the operating leverage is labor:

"Compensation and benefits expenses rose less than 1% in the third quarter and in the first nine months of 2026 compared to 2025, reflecting increases from wage inflation, substantially offset by improved employee productivity."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: Flat compensation on rising volume is genuine productivity, and it is the strongest evidence in the filing that the margin work Abel prioritized at BNSF is compounding. Intermodal continues to benefit from West Coast imports and tighter truck capacity, while coal volumes fall on plant retirements and lower natural gas prices. Part of the revenue growth is fuel surcharge that washes against fuel cost, so we would not extrapolate 14.7% revenue growth. But 18.5% pre-tax growth on 3.9% volume is the cleanest operating result in Berkshire this quarter, and BNSF is now the largest single contributor to the gap between the businesses that are growing and the ones that are not.

Utilities and Energy: BHE

The U.S. utilities grew 22.8% on a 12.7% increase in electric utility margin and retail volumes up 5.5% year to date, led by MidAmerican at plus 7.2% and NV Energy at plus 6.3%. The comparison is flattered by a $100M PacifiCorp wildfire accrual in the prior-year quarter, and no new accrual was taken this quarter: the cumulative estimate stayed at $2.85B. Corporate interest and other, which carries BHE's corporate interest and the tax credits recognized on a consolidated basis, fell 44.3%, and BHE subsidiaries have issued $4.8B of debt at a 5.8% weighted average rate this year.

"Net earnings increased $194 million (22.8%) in the third quarter and $288 million (16.8%) in the first nine months of 2026 compared to 2025, reflecting increases in electric utility margin and other income combined with higher income tax benefits from recognized production tax credits and the impact of wildfire loss accruals at PacifiCorp of $100 million in the third quarter of 2025, partially offset by increases in interest expense and other energy operating expenses."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: Load growth at MidAmerican and NV Energy is real and is the strongest demand signal BHE has shown in our coverage, consistent with the higher usage and customer growth the filing describes. Adjusted for the wildfire accrual, utility earnings grew roughly 10% to 13%, depending on whether the $100M is stated before or after tax, which is a good regulated-utility result rather than a re-rating event. The segment's wide outcome range is unchanged: the Oregon Supreme Court heard argument in the James appeal on November 3, and the company still cannot estimate a range of additional loss. Pre-tax earnings rose 18.2%, while BHE's income tax benefit fell to $574M from $606M, so after-tax growth was a more modest 8.3%.

Manufacturing, Service and Retailing

The collection decelerated sharply, from 24.1% after-tax growth in Q2 to 11.8%, and the deceleration is wider than the headline suggests because acquisitions are doing more of the work. Industrial products grew pre-tax 26.3%, but that includes OxyChem ($71M pre-tax on $1.3B of revenue, under acquisition-accounting depreciation) and the Acme Brick transfer into Marmon. PCC remains the standout, with pre-tax earnings up 29.6% on aerospace and industrial gas turbine demand, and IMC rose 29.8%. Lubrizol grew 10.3% while raising prices to pass through raw material and energy costs that it says "continued through the third quarter of 2026."

Building products is where the cycle shows. Homebuilding pre-tax earnings rose 6.7% to $575M only because Taylor Morrison joined, and management says Clayton's earnings fell on lower homebuilding gross margins. Other building products fell 12.3% on input costs, even with tariff refunds helping. Consumer products dropped 29.2% against a prior-year quarter that booked Duracell's manufacturing tax credits for 2023 through the first nine months of 2025 in one go; before those credits, the group grew 5.2% year to date. In service and retailing, TTI and aviation services led a 24.5% rise in service earnings, McLane fell 22.0% on lower gross margins as customer losses cut its retail sales, and Pilot's improvement includes a $116M loss on an unnamed business disposition that closed on November 2.

"The impacts of rising raw materials prices are expected to have a negative effect on IMC's earnings over the remainder of 2026 and into 2027."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: This segment is still in good shape, and PCC, TTI and the aviation businesses are among the best operating stories in the company. But the second-half weakening we flagged in August has arrived on schedule. IMC's warning now extends into 2027, TTI's gross-margin caution is repeated word for word, Lubrizol is chasing input costs, and the collection's margin compressed year over year after expanding in Q2. Excluding OxyChem's $71M and Pilot's $200M swing alone, pre-tax growth falls to about 7%, before any adjustment for Taylor Morrison. The collection can grow, but it cannot grow fast enough to replace a 26% decline in insurance earnings.

Other: FX Reverses, Occidental Fills the Gap

"After-tax equity method earnings increased $513 million in the third quarter and $579 million in the first nine months of 2026 compared to 2025 due to increased earnings from Occidental in each period and Kraft Heinz in the first nine months."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: We look through the FX line in both directions, as we have since initiation, and it should not change anyone's view this quarter any more than it did when it added $1.2B to Q2's year-over-year growth. The Occidental pickup deserves the same treatment. Berkshire books Occidental's results a quarter late, and Occidental's June quarter benefited from oil prices high enough to raise fuel costs at BNSF by 60.6% and Pilot's revenue by 48.4%. That is a commodity-price contribution to "operating" earnings that Berkshire's own operating businesses partly pay for, and it will move with the oil price.

Capital Allocation: The Buyback Doubles, the Equity Buying Stops

Coverage of the print reported that the cash pile rose to $377.4B. That figure includes $20.5B of Treasury bills bought in September and paid for in October, which sit on both sides of the balance sheet. On Berkshire's own net-of-payables basis, the pile fell $6.9B to $352.3B, its second consecutive quarterly decline:

"At September 30, 2026, our insurance and other businesses held investments in cash, cash equivalents and U.S. Treasury Bills (net of payables for purchases in September that settled in October) of $352.3 billion."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

The buyback: twice as large, and at a higher price

Berkshire retired 12,793 Class A equivalents, 0.89% of its shares, in one quarter, after 0.43% in Q2 and essentially nothing in the eight quarters before that. Every month's average Class B price was higher than June's $487.98, and September, the most expensive month, was also the largest at 56% of the quarter. Measured against the book value known at the time of purchase ($348.27 at June 30), management paid about 1.42x to 1.45x; against September 30 book of $357.73, about 1.39x to 1.42x. In August we used management's own purchase level as the floor for a buyer. That floor moved up by roughly 3% this quarter.

"Berkshire acquired $14.4 billion of treasury stock in the first nine months of 2026, including purchases in September that were settled in October."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

The buying continued into October

The 10-Q's cover page gives shares outstanding at October 28: 480,476 Class A and 1,403,805,594 Class B, or 1,416,346 Class A equivalents. At September 30 there were 1,418,900. Conversions of A into B do not change the equivalent count, so the 2,554-share decline, about 0.18% of the company, can only be repurchases. At the October 1–28 closing range of $500.50 to $548.56 per Class B share, that is roughly $1.9B to $2.1B in four weeks. The filing does not give the prices paid, and we do not know how much was bought above $520. We do know management kept buying while the stock climbed from about $500 toward $550.

The equity book: from $19.8B of net buying to a net sale

Derived from the cash flow statement, Berkshire bought about $2.4B of equities in the quarter and sold about $2.8B, a small net sale after Q2's roughly $19.8B of net purchases. The cost basis rose only $1.9B, again all in "commercial, industrial and other," with modest trimming in financials. Concentration eased slightly:

"The fair value of our five largest holdings at September 30, 2026 and December 31, 2025 represented 65% of the aggregate fair value of our equity securities as of each date. The five largest holdings at September 30, 2026 were Alphabet Inc., American Express Company, Apple Inc., Bank of America Corporation and The Coca-Cola Company."
— Berkshire Hathaway, Form 10-Q, Notes to Consolidated Financial Statements

Taylor Morrison closed, and both 2026 deals were bought near net asset value

Taylor Morrison closed on July 24 for about $6.8B (business acquisitions net of cash acquired were $6.5B in the quarter). The 10-Q gives the purchase-price allocation for OxyChem and Taylor Morrison only on a combined basis, but the combined figures are telling: $16.2B of net assets acquired against roughly $16.2B of consideration, with goodwill and other intangibles of only $1.8B. Inventories of $7.6B and property of $7.3B make up most of it, and real estate inventory jumped to $10.6B from $2.4B at year-end. Taylor Morrison produced about $1.1B of home sales in its first ten weeks.

"The third quarter increase reflected earnings from TMHC, partially offset by reduced earnings from Clayton Homes, primarily attributable to lower gross margins on its homebuilding activities."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: The capital-allocation pillar remains intact and the buyback is its strongest expression yet: $9.6B in a quarter, at rising prices, continuing into October. Two deals in 2026 were struck at roughly the fair value of the net assets acquired, with little goodwill, which is disciplined buying. But the composition of the quarter's deployment shifted entirely toward Berkshire's own shares, and the price of those shares has since run 13% beyond what management paid. A buyback is the best signal of management's intrinsic-value estimate that Berkshire gives. The signal says more than $507 in September. It says nothing about $571.

Reading Management Without a Call

Overall management posture: With no call, posture has to be read from the filing and from where the capital went. The capital says management is confident in its own shares and much less interested in other people's, having doubled the buyback while the equity buying stopped. The filing is candid about insurance and still says nothing about price: it names lower premium per policy as a cause of GEICO's loss ratio without describing any response. The quarter also completed the leadership transition, and on the evidence of the September letter that transition is settled rather than in progress.

1. Buffett steps back to Chairman Emeritus, and the transition is declared complete

On September 18, Berkshire named Warren Buffett Chairman Emeritus and elected Howard G. Buffett, a director since 1993, as Chairman. Buffett remains on the board, and Susan Decker remains Lead Independent Director. His letter is unusually direct about who has been running the company:

"He has taken hold of the Chief Executive Officer job in every respect. He has been making the decisions that matter for some time now, and I have not had to think twice about any of them."
— Warren E. Buffett, letter to shareholders, September 18, 2026

"Greg runs the company; Howard will guard its culture and values – both worth more than anything on our balance sheet."
— Warren E. Buffett, letter to shareholders, September 18, 2026

Assessment: This removes the last formal ambiguity in the succession: the repurchase program's requirement that the CEO consult "the Chairman of the Board" now means consulting Howard Buffett, not his father. The market barely reacted, with the stock little changed on September 18 and down 1.5% the next session, which is the right reaction. The succession discount we wrote about at initiation is now fully gone. The stock passed its pre-announcement peak of May 2025 on October 20.

2. The buyback doubled and the purchase price rose every month

July at $496.24, August at $503.99, September at $506.64. The September tranche was the largest of the quarter and the most expensive. The program's terms are unchanged:

"Berkshire's common stock repurchase program currently permits Berkshire to repurchase shares any time that Berkshire's Chief Executive Officer, after consultation with the Chairman of the Board, believes that the repurchase price is below Berkshire's intrinsic value, conservatively determined."
— Berkshire Hathaway, Form 10-Q, Notes to Consolidated Financial Statements

Assessment: Management's conservatively determined intrinsic value sits comfortably above $507 per Class B share. That is useful, and it is higher than we inferred in August. It does not tell us that $571 is below intrinsic value, and the size of September's buying at $507 makes it likely that the purchase level management favors is well below Friday's close.

3. The buying did not stop when the stock rallied

The cover-page share count implies about 2,554 Class A equivalents repurchased between September 30 and October 28, roughly $1.9B to $2.1B, while the stock traded between $500.50 and $548.56.

Assessment: This is the strongest argument against our downgrade, and we take it seriously. Management kept buying in October while the stock traded between about $500 and $549, and some of that buying may have been above $520. If it continues at $570, the floor moves to our price. We will know at the Q4 report. Until then, the only disclosed purchase prices are Q3's, averaging $503.68, about 12% below the current price, and even the highest close from October 1 to 28 was 4% below it.

4. The equity buying stopped after one quarter

About $2.4B of purchases against $2.8B of sales, after Q2's roughly $23.5B of purchases. Net new money again went into the commercial and industrial category, but only $2.3B of it. The equity book rose $11.7B in value, almost entirely from market gains.

Assessment: Q2's buying spree was a single quarter, not a new policy, which is what the bear case on capital deployment said in August. The money that would have gone to equities went into Treasury bills and the company's own shares instead. That is a defensible choice. It is also a statement that management found more value in Berkshire at about $504 than in the market at large, and the market has since closed much of that gap.

5. GEICO names falling price per policy as a cause of its loss ratio

"Private passenger auto claims frequencies increased in the first nine months of 2026 for bodily injury coverage (four to six percent range) and property damage and collision coverages (two to four percent range) compared to 2025. Private passenger auto average claims severities in the first nine months of 2026 increased for bodily injury coverages (nine to eleven percent range) and property damage and collision coverages (zero to three percent range) compared to 2025."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Each nine-month range is about a point lower than the six-month range on frequency and on bodily-injury severity, so loss-cost inflation is easing at the margin. Price per policy is still falling.

Assessment: Easing loss-cost trends would ordinarily let a carrier rebuild margin with modest rate. GEICO is doing the opposite and cutting average premium to grow policies-in-force. That is a deliberate competitive choice, and it means the margin recovery will be slower than the frequency data alone would suggest.

6. GEICO's flat combined ratio was held flat by reserve releases

"Losses and LAE included reductions in the ultimate loss estimates for prior accident years' claims of $628 million in the first nine months of 2026 compared to $495 million in 2025."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Set against the six-month figure in the Q2 filing, this implies $474M of the $628M came in the third quarter alone, worth 4.2 points of GEICO's loss ratio.

Assessment: In Q2, management said the year-over-year change in GEICO's reserve releases was "relatively insignificant" and we credited that candor. In Q3, development became a material support, and it is the reason the reported combined ratio did not rise again. Reserve releases at GEICO are earned from past conservatism, so we are not questioning them. We are noting that the cushion is now being used to hold the reported line, and cushions run out.

7. The soft property market reaches the commercial and reinsurance books

"Premiums written declined $418 million (7.9%) in the third quarter and $575 million (4.0%) in the first nine months of 2026 compared to 2025, reflecting year-to-date declines across several primary group business units, including RSUI (11.5%), BHHC (9.7%), MedPro (5.1%) and GUARD (14.5%)."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: Declining volume at RSUI and GUARD is discipline, not distress. But the operating consequence is that BH Primary and BHRG, which together added $1.19B of pre-tax underwriting in Q2, added only $701M in Q3. The insurance segment no longer has an internal offset to GEICO.

8. Investment income falls at a double-digit rate, and the dividend offset did not arrive

After-tax insurance investment income fell 14.1%, accelerating from a 9.1% decline in Q2, and dividends fell 3.0% despite Q2's equity purchases.

"Float was approximately $178.5 billion at September 30, 2026, an increase of $2.1 billion from December 31, 2025."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: Float grows; the yield on it shrinks faster. For the insurance segment, that combination means lower earnings on a larger asset base, and it puts more pressure on underwriting to carry the segment just as underwriting is weakening.

9. BNSF compounds on productivity

"Average revenue per car/unit increased 10.5% in the third quarter and 7.1% in the first nine months of 2026, primarily from higher fuel surcharge revenue and higher yield. Pre-tax earnings increased 18.5% in the third quarter and 15.5% in the first nine months of 2026 compared to 2025."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: Sixth consecutive healthy quarter and the largest quarterly profit of the series. Higher yield alongside a lower operating ratio while fuel cost rose 60.6% is the profile of a railroad with pricing power and cost control at the same time. BNSF is the segment we would most willingly pay up for.

10. Taylor Morrison joins as Clayton's margins compress

"Site-built units sold increased 87.2% in the third quarter and 30.2% in the first nine months of 2026 relative to the same periods in 2025. These increases were primarily attributable to TMHC, which produced home sales of approximately $1.1 billion from the acquisition date through September 30, 2026."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: The homebuilding platform is now national, bought near net asset value, and arrived just as Clayton's gross margins fell. That is the countercyclical purchase Berkshire intended. The near-term earnings contribution will be modest and cyclical, and the 10-Q discloses no standalone Taylor Morrison profit.

11. Input-cost warnings spread through the industrial businesses

"The increases in selling prices were necessitated by significant increases in raw materials, energy and supply chain costs that began in the latter part of the first quarter and continued through the third quarter of 2026, resulting in increased production costs."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

"Inventory cost and supply chain uncertainties could negatively impact TTI's gross margins in the future."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: In Q2 two businesses carried forward warnings. In Q3, IMC's warning extends into 2027, Lubrizol describes three quarters of cost inflation, other building products cite input costs for a 12.3% decline, and TTI's caution is repeated. Customers at IMC and TTI pulled purchases forward, which flatters the present and borrows from the future. We expect the collection's growth to slow further in 2027.

12. Kraft Heinz: the gap widens and the impairment is again deferred

"As of September 30, 2026, the carrying value of our investment in Kraft Heinz common stock exceeded fair value by $1.4 billion (or 16.2% of our carrying value)."
— Berkshire Hathaway, Form 10-Q, Notes to Consolidated Financial Statements

Assessment: The gap was $1.1B (12.2%) at June 30. Management again judged the decline temporary and added that its intentions "may change in the future and may result in the recognition of an impairment loss at that time." Occidental, the other equity-method holding, now trades $3.3B above carrying value, so the two positions net to a gain. But the Kraft Heinz write-down remains a live charge against a future quarter's net earnings.

13. Wildfire: no new accrual, and the appeal has been argued

"On June 25, 2026, the Oregon Supreme Court issued an order allowing the petition for review and oral argument was held on November 3, 2026."
— Berkshire Hathaway, Form 10-Q, Notes to Consolidated Financial Statements

Assessment: PacifiCorp's cumulative accrual stayed at $2.85B, unpaid liabilities fell to $566M, and the remaining Phase II trials are stayed pending the court. The decision on whether the April Court of Appeals reversal stands is the largest binary event in BHE's outlook. A ruling that upholds the Court of Appeals would materially narrow the James exposure; a reversal would reopen it. The company still says it cannot estimate a range of additional loss.

What They're NOT Saying

1. Anything about GEICO's pricing strategy. The 10-Q now names lower average premium per policy as a driver of the loss ratio and of falling premiums written, alongside growth in policies-in-force. It says nothing about rate filings, planned increases, or whether the price cuts are a deliberate share strategy. We searched the filing for rate-change and pricing language and found none for GEICO. For the most consequential operating decision in the company, that silence is conspicuous.
2. Why the equity buying stopped. Purchases fell from about $23.5B in Q2 to about $2.4B in Q3. The filing gives the totals and an alphabetical top-five list and nothing about the change in posture. The Q3 Form 13F, due in mid-November, will show composition but not reasoning.
3. The prices paid in October. The cover page shows about 2,554 Class A equivalents retired after quarter-end, so buying continued. Whether management paid $505 or $545 for those shares is the most useful unknown for anyone buying at $571, and it will not be disclosed until February.
4. A standalone allocation or profit figure for Taylor Morrison. The purchase-price allocation is combined with OxyChem, and Taylor Morrison's earnings appear only as the reason homebuilding rose 6.7%. A $6.8B acquisition gets one revenue figure.
5. The counterparty in the $1.12B retroactive commutation. The contract dated from 2016 and the earnings effect was immaterial, but the filing names neither the counterparty nor the remaining exposure that was released.
6. What Pilot sold. Pilot booked a $116M loss on a business disposition that closed on November 2. The business is not identified.
7. The names of the four reporting units with thin goodwill headroom. The filing does give aggregates: estimated fair value of about $27.7B against about $26.2B of carrying value, with about $9.2B of goodwill. That is headroom of roughly 5.7%, carried from last year's annual test without a name attached. With building-products margins compressing, it is worth knowing which units these are.
8. Any comment on the leadership change. The September 18 release and letter speak for themselves, but the 10-Q itself never mentions that the Chairman whom the CEO consults on repurchases changed during the quarter.

Book Value and Valuation

Book value per share compounded 2.7% in the quarter and 7.6% year to date, helped by $9.4B of after-tax investment gains and held back slightly by the buyback, which retires shares at about 1.4x book. The stock compounded faster. Measured from our Q2 reference price, BRK-B rose 9.4% while book value rose 2.7%, and the multiple moved from about 1.50x to about 1.60x, the most expensive reading of our coverage window and the top of the 1.2x to 1.6x range we use as Berkshire's decade band. Across our six notes the multiple has run about 1.48x, 1.47x, 1.44x, 1.39x, 1.50x and now 1.60x.

Our twelve-month book value estimate of $380 to $390 assumes 6% to 9% growth: operating earnings net of the buyback's modest dilution to book at prices above book, plus ordinary market returns on a $336B equity portfolio. The scenarios show what the multiple has to do for the stock to keep pace with an ordinary market year. Only the scenario in which the rotation premium persists in full produces a high-single-digit return. Every scenario in which the multiple drifts back toward where management itself was buying produces a flat to negative one.

Assessment: In August we called 1.50x book fair for a business whose non-insurance operations were accelerating and whose largest insurance franchise was decelerating. Since then the insurance deceleration has deepened, the non-insurance acceleration has slowed, operating earnings have turned negative year over year, and the multiple has risen. Book value is compounding at a healthy rate, and the multiple can be grown into. But 1.60x is the multiple of a Berkshire with improving earnings and a market premium for safety, and only the second condition holds today.

Market Reaction

• Pre-print setup: BRK-B closed at $570.89 on Friday, November 6, up 13.6% year to date against the S&P 500's 7.8%, up 15.8% over the trailing twelve months and 12.8% over the trailing thirty days. The 52-week closing range entering the print was $465.40 to $579.09, placing the stock 1.4% below its high.
• Reaction session: none yet. Results and the 10-Q were posted at approximately 7:00 a.m. Central on Saturday, November 7, with markets closed, and Berkshire holds no call. The first session that can trade the quarter is Monday, November 9.
• Into the print: BRK-B set an all-time closing high of $579.09 on November 4, up 3.95% on the day, its largest one-day gain since April 2025. It rose 5.2% in the week of November 2 while the S&P 500 fell 2.7%, including a 4.36% index decline on November 5. From the September 30 close of $497.95, the stock is up 14.6% against an index down 3.6%.
• Since our Q2 note: BRK-B is up 9.4% from $521.80 while the S&P 500 fell 4.9%, a relative gain of about 14 points after we downgraded to Hold.

A flow-driven re-rating, priced before the print

Nothing Berkshire disclosed in October explains a 14.6% rise in five weeks. The company issued no release between September 18 and the November 4 notice of this report's timing. The move coincided with a sharp sell-off in AI-exposed technology, which pushed the Nasdaq into a correction, and the market's explanation is a rotation into a mega-cap with little exposure to artificial intelligence, an enormous Treasury bill position, and a reputation for doing well when markets break. That explanation is credible. It is also, by construction, an explanation about other stocks.

The quarter the market will trade on Monday is weaker than the price it set on Friday. Operating earnings fell 5.8%, insurance earnings fell 26%, and the best-known franchise in the company is cutting price into rising loss costs. The buyback is a genuine positive and will likely dominate weekend coverage. But the stock enters the session 13% above the average price at which Berkshire bought its own shares, and the defensive premium that lifted it can unwind as quickly as it built if the technology sell-off stabilizes.

We should be plain about our own record. We downgraded to Hold in August expecting in-line performance against the index, and the stock has beaten the index by about 14 points since. Part of that was a market event we did not forecast. None of it came from an improvement in the earnings we said were deteriorating, and those earnings deteriorated further. We are making the same fundamental argument at a higher price, and the published trigger that would turn it into a lower rating has fired.

Street Perspective

Debate: Is the re-rating a durable reappraisal or a rotation trade?

Bull view: The market is finally pricing what Berkshire has become under Abel: an active capital allocator with a national homebuilder, a growing chemicals business, a compounding railroad and $350B of dry powder, run by a CEO the founder says has been making the decisions for some time. The AI sell-off simply accelerated a reappraisal that was already under way when the stock passed its 2025 peak on October 20.

Bear view: A nearly 15% move in five weeks with no company news is a flow, and flows reverse. Berkshire's YTD outperformance was built almost entirely since late September as money left technology. If AI-exposed stocks stabilize, the haven premium goes with it, and what remains is a company whose operating earnings are falling.

Our take: The bears have the better of this one. The bull narrative is true, but it was equally true at $500 in September, when management was buying. The incremental 14% has a market explanation, not a Berkshire one. We think about half of the move is reappraisal that can hold and half is a defensive premium that will fade.

Debate: Does a $9.6B buyback at $504 validate $571?

Bull view: Management doubled the buyback, paid more each month, and kept buying into October as the stock climbed toward $550. A CEO whose repurchases are bound by a conservatively determined intrinsic value would not do that unless intrinsic value were well above the price. The buyback sets a floor that keeps rising.

Bear view: The buyback validates the price management paid, not the price the market charges. September's $506.64 average sits 11% below Friday's close, and the program's own language says purchases happen only below conservative intrinsic value. Management has never shown it will buy at 1.6x book.

Our take: The October share count is the bulls' best evidence, and it is why our conviction on the downgrade is moderate rather than high. If the Q4 filing shows heavy buying above $560, we will have been wrong about where management's value sits, and we will say so. On current disclosure, management's demonstrated level is the Q3 average of about $504, with October's prices not yet known.

Debate: Is GEICO's deterioration cyclical or competitive?

Bull view: Loss-cost trends are easing, with frequency and bodily-injury severity each down about a point from the six-month ranges, and GEICO is growing its policy count. Taking share in a soft patch and repricing later is how GEICO built its franchise. A 91.4% combined ratio is still an underwriting profit.

Bear view: GEICO is cutting price per policy while severity rises 9% to 11% and expenses rise 17%. Before reserve releases, its combined ratio was roughly 95.5%. The reported stability came from a $474M release that is three times the first half's total. That is a competitive-position problem being disguised by past conservatism.

Our take: We now lean firmly bear. The share-gain strategy may prove right over several years, but its cost is visible today and its payoff is not, and the accident-year loss ratio worsened rather than improved in Q3. We would need two quarters of year-over-year accident-year improvement, or evidence that average premium per policy has turned up, before treating this as a cycle that has peaked.

Model Update

Valuation framework: We continue to anchor on price to book, recognizing that reported book understates value because a growing share of Berkshire sits in operating businesses carried at depreciated cost. Book value per Class B share of $357.73, compounding at 6% to 9% a year, reaches $380 to $390 in twelve months. We think the fair multiple for a Berkshire whose operating earnings are falling while its insurance earnings shrink is about 1.45x to 1.50x: the level at which management itself bought in Q3, plus a small premium for a balance sheet that will be valuable in a weaker market. That implies a fair value of roughly $550 to $585, or a twelve-month return of about −4% to +2.5% from $570.89. We expect an ordinary market year to beat that, which is the definition of our Underperform.

Thesis Scorecard Post-Earnings

The Q2 checklist, graded

The Q2 triggers, graded as published

• Upgrade, condition one: a return toward 1.35x to 1.40x book. Not met. The stock moved to about 1.60x.
• Upgrade, condition two: GEICO rate actions stabilising the loss ratio while acquisition spend normalises. Not met. The reported loss ratio held at 76.6% only with a larger reserve release, the expense ratio rose to 14.8% from 14.6% in Q2, and no rate action is disclosed.
• Downgrade: the combined ratio deteriorates a third consecutive quarter without a visible pricing response. Met, on both the year-over-year and the sequential measures, with average premium per policy still falling.

Overall: The thesis weakened on every insurance pillar and held on capital allocation and continuity. The business remains superbly financed and well run, BNSF is excellent, and the buyback is the most shareholder-friendly use of capital in our coverage. But the earnings that make Berkshire distinctive, underwriting profit on negative-cost float and the investment income it earns, fell 26% in a catastrophe-free quarter, and the reported figures understate the decline because reserve releases filled part of the gap. The investment case we downgraded to Hold has moved further against the price.

Action: Downgrading to Underperform from Hold, conviction moderate. We are acting on the trigger we published in August, at a price 9.4% higher than when we published it. We expect BRK-B to trail the S&P 500 over twelve months as the defensive premium fades and insurance earnings keep shrinking. The main risk to the call is that the technology sell-off deepens and Berkshire's haven premium widens further. We will move back to Hold if the stock returns to about 1.45x the latest reported book value, roughly where management bought in Q3, or if GEICO shows two consecutive quarters of year-over-year improvement in its accident-year loss ratio with average premium per policy rising. If the price returns to 1.45x while GEICO keeps deteriorating, those two signals net to Hold, not higher. Outperform requires both a price at or below about 1.40x book and the GEICO evidence. If the Q4 filing shows heavy repurchases above $560, we will reassess, because that would put management's intrinsic-value estimate above today's price.