Downgrading to Hold: Abel Ended a 14-Quarter Selling Streak and Put $24 Billion to Work, But the Discount We Underwrote Is Mostly Spent and GEICO Is Breaking

Results vs. Consensus

Berkshire is thinly covered for its size and publishes no operating EPS of its own, so "consensus" here is a loose construct assembled by data providers. Three different Q2 figures circulated ahead of the print, from $5.04 to $5.24, clustering around $5.13. Against that, the derived operating result of roughly $6.03 per Class B share was a beat of about 17%. The more useful exercise is not measuring the beat but decomposing it, because the single largest contributor is a line no model forecasts and no operating manager controls.

Q2 2026 Scorecard

The GAAP line is noise by construction and Berkshire says so in the release itself. Investment gains of $12.68B after tax, mostly unrealized marks on a $324B equity book, doubled reported net earnings; the prior-year quarter carried a $3.76B after-tax Kraft Heinz impairment running the other way. Neither belongs in an assessment of the quarter's operations.

Year-Over-Year Comparisons

First-Half Comparisons

Revenue. The $9.3B consolidated revenue increase looks broad and is not. Pilot Travel Centers alone accounted for $4,823M of it, or 51.9%, entirely on higher fuel prices against slightly lower fuel volumes, at a pre-tax margin of 1.9%. That is a pass-through, not growth. The revenue lines that carry real economics are BNSF at plus 14.6%, the industrial products group at plus 27.3%, and the service group at plus 21.1%. Insurance premiums earned, the largest insurance revenue line at $22.5B, grew 1.3%.

Margins. The cleanest margin story in the company is manufacturing, service and retailing, where pre-tax margin expanded 0.8 points to 9.5% on a 15.2% revenue increase. Industrial products carried it, with margin up 2.0 points to 21.2% on aerospace strength at Precision Castparts (pre-tax up 34.2%) and metal-cutting at IMC (up 71.0%). OxyChem added revenue and pre-tax dollars in its second quarter of ownership but diluted the margin rather than lifting it, running a 10.6% pre-tax margin against the group's 21.2%. The offsetting story is insurance, where GEICO's combined ratio deteriorated 7.7 points and consolidated pre-tax underwriting margin fell despite substantially larger reserve releases. Berkshire's earnings mix is shifting away from its highest-return franchise toward its most capital-intensive ones, which is a return-on-capital question even when it looks fine in dollars.

Earnings per share. Operating EPS grew 16.5% against operating earnings growth of 16.3%, the gap being the buyback. That gap is about to widen: Berkshire retired roughly 0.43% of its shares in a single quarter, having retired essentially none in the prior eight. If the second-quarter pace were sustained the annualized retirement rate would run near 1.7%, which for a company of this size is the difference between a share count that drifts and one that compounds. We would not extrapolate the June pace, because the repurchase is explicitly price-contingent and the price has since risen above where management was buying.

Segment Performance

Insurance — Underwriting

Last quarter's underwriting comparison was flattered by an easy catastrophe comp: Q1 2025 carried roughly $1.1B of Southern California wildfire losses that did not recur. That excuse is unavailable here. The wildfires were a first-quarter event, so Q2 2025 and Q2 2026 are both catastrophe-free, and underwriting still fell 14% pre-tax. Worse, it fell while absorbing materially more help from prior-year reserve releases.

GEICO — a two-sided squeeze, not a one-quarter blip

This is the quarter GEICO stopped being a watch item and became a thesis problem. Both sides of the combined ratio deteriorated at once, which is unusual and diagnostically important. On the loss side, management quantifies the drivers precisely:

"Private passenger auto claims frequencies increased in the first six months of 2026 for bodily injury coverage (five to seven percent range) and property damage and collision coverages (three to five percent range) compared to 2025. Private passenger auto average claims severities in the first six months of 2026 increased for bodily injury coverages (ten to twelve 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

Frequency and severity rising together on bodily injury, the longest-tail and most inflation-exposed coverage in personal auto, is the classic signature of a loss-cost cycle turning against the underwriter. Critically, management forecloses the benign explanation: the change in prior-year reserve reductions at GEICO was "relatively insignificant," so this is not an accounting artifact.

The expense side is arguably the more revealing half. Underwriting expenses rose 27.3% in the quarter, driven by commissions and advertising, at a time when premiums written grew just 1.1%. And that 1.1% is itself misleading: it came from commercial auto, a business currently under 5% of premiums written, offsetting lower average premiums per policy in private passenger auto. Put those facts together and the picture is a franchise spending materially more to acquire customers while charging them less per policy, into rising loss costs. That is a market where pricing power has, at least for now, inverted.

Assessment: We flagged the GEICO loss ratio as a new watch item last quarter and said we would monitor the next two prints. One of those two has landed and it is worse, not better: pre-tax underwriting down 34.8% in Q1 became down 45.4% in Q2, and the expense ratio has now joined the loss ratio in deteriorating. GEICO's structural advantage was always the lowest cost structure in the industry converted into price, which bought share profitably. Rising acquisition spend against falling average premium is that advantage running in reverse. One more quarter of this and the question stops being about the auto cycle and starts being about the competitive position. This is the single largest change to our view of Berkshire since we upgraded.

BH Primary and BHRG — the offsets, and what is inside them

The commercial and reinsurance operations did the work of holding the segment together. BH Primary swung to $273M of pre-tax underwriting from $63M, and BHRG rose 40.5% to $913M. Both improvements lean heavily on the reserve releases tabulated above, and both are running smaller books: BH Primary premiums written fell 4.1% in the quarter, with year-to-date declines at RSUI (13.2%), BHHC (5.7%), GUARD (7.5%) and BHSI (2.6%) as several units reduced U.S. property volumes. Excluding a new contract, BHRG property/casualty premiums written fell 5.6%.

That new contract is the quarter's one genuinely new insurance disclosure, and it is a large one:

"We recorded premiums written in the second quarter and first six months of $483 million from a new whole account reinsurance agreement with certain wholly-owned insurance subsidiaries of Tokio Marine Holdings, Inc. ("Tokio Marine"), which commenced on April 1, 2026. Under the agreement, NICO assumes on a quota-share basis a portion of the non-life premiums written and related losses and expenses of Tokio Marine on risks attaching over a ten-year term."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

Assessment: Shrinking the property book into a softening market is correct underwriting discipline and exactly what Berkshire's reinsurance operations are supposed to do; we read the premium declines as a feature. The Tokio Marine treaty is more interesting than its first-quarter contribution suggests. A ten-year whole-account quota share with a major Japanese carrier is a durable, relationship-based use of balance-sheet capacity that only a handful of counterparties in the world could write, and it is the kind of business that scales with Berkshire's float rather than competing for it. It does not offset GEICO, but it is evidence that the reinsurance franchise is still winning the deals its capital entitles it to.

Insurance — Investment Income

Interest income fell 12.8% as short rates drifted lower and the Treasury bill book reinvested at reduced yields. Dividend income was flat, which will change: the roughly $19.8B of net equity purchases made during the quarter will begin contributing dividends from Q3, partially offsetting the interest decline.

Assessment: This line has now declined for four straight quarters and the direction is structural rather than cyclical noise. It is also the single most powerful argument for the deployment that finally happened. A dollar earning a falling short-term yield inside a $325B Treasury bill position is the least productive dollar Berkshire owns, and management is now voting with the balance sheet. The transition is dilutive to reported investment income in the short run and accretive to intrinsic value if the purchases are good ones, which is precisely the trade a shareholder should want. Float of $177.5B remains negative-cost, which is the asset underneath all of it.

Railroad — BNSF

The reported plus 6.3% understates this quarter badly. Pre-tax earnings rose 13.9%; the entire gap is a tax-rate comparison, and the distortion sits in the prior year. Management attributes the 5.4-point effective-rate increase to lower enacted state income tax rates recognized in the second quarter of 2025, a one-time benefit that did not repeat. Underneath, this was the best operating quarter BNSF has produced in our coverage window: volumes up 6.5% across three of four business groups, average revenue per car or unit up 7.6%, and consumer products volumes up 9.3% on higher west coast imports, tightening truck capacity, and what management explicitly calls market share gains.

The operating ratio nonetheless worsened 0.6 points, and the reason is fuel: fuel expense rose 68.1%, or $475M, on higher average prices. Because fuel surcharges also inflate revenue, the effect largely washes in dollars while optically compressing the ratio. Compensation and benefits rose only 1.9% on wage inflation offset by productivity, which is the line to watch for genuine operating leverage.

Assessment: Taking share in intermodal while truck capacity tightens is the most durable thing happening anywhere in this company right now, and it is the fifth consecutive quarter of healthy BNSF results. We would encourage readers not to anchor on the after-tax number: at plus 13.9% pre-tax on plus 6.5% volumes, BNSF is compounding, and the margin opportunity Abel singled out on taking the job is visibly being worked. The coal decline (down 7.9% on plant retirements and cheap natural gas) is secular and already in everyone's model. Adjusted for the fuel pass-through, this is the segment we would most willingly pay up for.

Utilities and Energy — BHE

BHE produced its strongest quarter of the coverage window, with U.S. utilities up 37.6% on an 8.1% increase in electric utility margin and retail volumes up 3.1% year to date, led by MidAmerican at plus 6.0% and NV Energy at plus 4.3%. Pipelines added 32.2% on a general rate case, with colder-weather LNG volumes contributing to the six-month figure but concentrated in Q1. The first-half figures put the quarter in proportion: over six months U.S. utilities grew only 10.9% and total BHE earnings 11.3%, so Q2 carried nearly all of the year's progress and is not a run rate. (The $1,802M six-month figure above is earnings attributable to BHE; the $1,799M shown in the segment tables is the Berkshire-attributable figure, the $3M difference being preferred dividends paid in 2025.) It is worth being precise about the tax mechanics: BHE reported pre-tax earnings of $604M and an income tax benefit of $333M, a negative 55.1% effective rate driven by wind production tax credits. That is a real, recurring feature of BHE's economics rather than a one-off, but it does mean the segment's after-tax growth is levered to a credit regime that is scheduled to phase down.

The blemishes are familiar. Energy operating expenses rose on vegetation management and wildfire prevention, the ongoing cost of PacifiCorp's exposure. Real estate brokerage fell 55.6% on a litigation settlement charge and continues to labor under low housing inventory. Northern Powergrid is lapping unfavorable inflation-linked tariff adjustments.

Assessment: Better than we expected and better than the segment's medium-term run rate, with rate-case timing and tax credits both helping. The structural picture is unchanged: BHE carries the heaviest capital load in the company, with BNSF and BHE together forecast to spend roughly $8.6B over the remainder of 2026, and its earnings quality is more regulated-utility than Berkshire-average. We continue to treat it as the segment with the widest outcome range on wildfire liability and the credit phase-out, and we do not extrapolate a 37.6% utilities quarter.

Manufacturing, Service and Retailing

This is where the quarter was genuinely earned. Industrial products was the engine, with revenue up 27.3% and pre-tax up 41.0% on margin expansion of 2.0 points. Precision Castparts grew revenue 14.4% and pre-tax 34.2% on aerospace and industrial gas turbine demand, IMC grew pre-tax 71.0%, and Lubrizol added 23.4% on price increases that management is explicit were necessitated by rising raw material and energy costs. OxyChem contributed $1.4B of revenue and $149M of pre-tax earnings in its second quarter of ownership, a marked improvement from the first: the six-month pre-tax figure of $121M implies the January-to-March stub ran at a modest loss under acquisition-accounting depreciation and transition costs.

Three caveats keep this from being a clean read. First, Pilot's 143.7% pre-tax increase is a fuel-margin and derivative-timing outcome on a 1.9% margin business, and management warns that fuel price volatility produces earnings volatility here. Second, other building products' 16.6% pre-tax increase is disclosed as being driven by trade-tariff refunds on duties paid in 2025 plus lower restructuring and legal costs; before those items, management states pre-tax earnings fell 8.8%. Third, IMC carries an explicit forward warning that second-half earnings will be hurt by rising raw material costs, and some of its demand came from customers pulling purchases forward.

The soft spots are consistent with a two-speed economy. Clayton Homes pre-tax fell 3.5%, McLane's retail sales declined 8.9% year to date on customer losses, Forest River weakened, and management repeats that several retailing businesses face "sluggish customer demand." Housing-linked and lower-income-consumer businesses are struggling; anything touching aerospace, electrical infrastructure or industrial capital spending is booming.

Assessment: The best underlying performance in the company and the clearest evidence that the operating collection is in good hands. We would haircut the reported 24.1% for tariff refunds, production tax credits and Pilot's fuel windfall, and we would treat the industrial products momentum as the durable part. The forward-looking disclosure that matters is IMC's raw-material warning and TTI's caution that inventory costs could compress gross margins, both of which point to a tougher second half in exactly the businesses that led the first. This segment is now also where the next chapter is written: Taylor Morrison joins building products, giving Berkshire a national site-built homebuilding platform at the trough of an affordability cycle.

Other — The FX Line Does It Again

Assessment: For the second consecutive quarter, a non-cash remeasurement of yen- and euro-denominated debt is the largest single contributor to operating-earnings growth. We look through it entirely, in both directions. It is worth noting that Berkshire keeps adding to the exposure that creates this line: the company issued ¥272.3B (about $1.7B) of senior notes in April at a 2.4% weighted average rate, maturities running to 2056. That is excellent financing and a sound strategic decision. It also guarantees that this line will keep manufacturing and destroying a billion dollars of reported "operating" earnings at the whim of the dollar, and that anyone anchoring on the headline growth rate will keep being misled.

Capital Allocation: The Quarter the Thesis Paid Off

We upgraded Berkshire on the argument that capital was about to move. In Q1 that argument was validated symbolically, with a $235M buyback and a closed OxyChem deal. In Q2 it was validated at scale. On every measure that matters, this is the most aggressive capital-deployment quarter Berkshire has run in years.

The buyback carries a price signal, and it is below today's price

Two things stand out. First, the buying accelerated into a rising price: June accounted for 84% of the quarter's repurchases at an average Class B price roughly 2.5% above May's. Management was not running a passive limit order at a fixed multiple of book; it was leaning in. Second, and less comfortably for a buyer today, the average prices paid of $476.01 and $487.98 sit at roughly 1.37x to 1.40x the June 30 book value of $348.27 per Class B share. Friday's close of $521.80 is about 7% above the highest monthly average management paid.

The repurchase framework is unchanged and worth reading in management's own words, because it defines the floor:

"Berkshire's common stock repurchase program permits Berkshire to repurchase its Class A and Class B shares at prices below Berkshire's intrinsic value, as conservatively determined by Berkshire's Chief Executive Officer after consultation with the Chairman of the Board. We are not committed to a minimum or subject to a maximum repurchase amount. We will not repurchase our stock if it reduces our consolidated cash, cash equivalents and U.S. Treasury Bills holdings to below $30 billion. Financial strength and redundant liquidity will always be of paramount importance at Berkshire."
— Berkshire Hathaway, Form 10-Q, Management's Discussion and Analysis

The equity deployment: where roughly $20 billion went

Cost basis is the honest measure of capital committed, because fair value moves with the market. It rose $21.1B in the first half, and the entire increase (and more) landed in "commercial, industrial and other," up $24.1B, while consumer products cost basis fell $3.2B on net selling. The five largest holdings at quarter-end were Alphabet, American Express, Apple, Bank of America and Coca-Cola, together 66% of the equity book against 65% at year-end 2025 and roughly 61% at the end of Q1.

Taylor Morrison: the second acquisition in seven months

Berkshire agreed to acquire Taylor Morrison on May 31, secured shareholder approval on July 22, and closed on July 24 for $72.50 per share, roughly $6.8B of equity value and about $8.5B of enterprise value. It joins the building products group and will be integrated with Clayton Properties Group, the collection of 15 regional and local site-built homebuilders Berkshire has assembled over the past decade. The 10-Q notes the deal closed too close to the filing date to estimate acquired asset values, which will appear in the Q3 statements.

"Today marks an important step forward as Taylor Morrison joins Berkshire. This best-in-class national homebuilder will lead our vision for a unified site-built homebuilding operation. Together, we will help more Americans achieve their dream of homeownership."
— Greg Abel, CEO, Berkshire Hathaway (acquisition completion release, July 24, 2026)

Assessment: This is the quarter Berkshire's capital allocation stopped being a question and became a fact. Roughly $24B was deployed during the quarter through net equity purchases and repurchases, with a further $6.8B committed to Taylor Morrison three weeks after the period closed. The headline $31.9B decline in cash and Treasury bills overstates the deployment, because about $16.5B of it is the unwind of a payable for unsettled Treasury purchases; net of payables the pile fell $14.3B. Management chose to buy a homebuilder while its own housing-linked businesses were reporting weak demand, which is the correct direction of travel for a countercyclical buyer. Buying a national homebuilder into a soft affordability market, and bolting it onto an existing 15-builder platform, is a recognizably Berkshire move: unglamorous, scaled, and priced off distress rather than momentum. Abel has now deployed roughly $16B on two acquisitions and $4.8B on buybacks in seven months, against essentially nothing in the preceding two years. We do not think the market has fully absorbed how different an operator this is. That is the bull case. The problem for us is that it is now partly in the price.

Reading Management Without a Call

Overall management posture: Berkshire says nothing on a call because it holds none, so posture must be read from the filing and from where the capital went. On that evidence the posture shifted decisively from defensive to offensive during the quarter, with the buying accelerating month by month rather than tapering. The written disclosure, by contrast, is notably candid on the deterioration: management quantifies GEICO's frequency and severity precisely, volunteers that reserve development was not the cause, and separately strips out the tariff refunds flattering a building products line. That is a company describing a problem accurately rather than managing around it, which is worth something.

1. The 14-quarter selling streak ended, and the scale is what surprises

Derived from the cash flow statement against the prior quarter's actuals, Berkshire bought roughly $23.5B and sold roughly $3.7B of equities in Q2, for net purchases near $19.8B. For a company that had been a net seller in every quarter since 2022, this is not a marginal adjustment in posture; it is a reversal. The sales figure is arguably the more striking half: $3.7B of sales in a quarter, against $24.1B in Q1, means the disposal program essentially stopped at the same time the buying started.

Assessment: Two consecutive quarters ago the open question was whether anything would ever be done with the cash. That question is now answered. The follow-on question, which we cannot answer from this filing, is whether the purchases were good ones, and that will only be visible over years.

2. The buyback's price signal is more informative than its size

Zero repurchases in April, $741M in May at $476.01 per Class B share, and $3,787M in June at $487.98. Management bought five times as much in June as in May, at a higher price, having bought nothing at all six weeks earlier.

Assessment: The escalating pace at rising prices tells you management's conservatively determined intrinsic value sits comfortably above $488 per Class B share. The absence of April buying tells you the framework is genuinely price-contingent rather than programmatic. Both facts are useful. Neither supports paying $521.80.

3. GEICO's expense ratio is the disclosure that changes our view

A deteriorating loss ratio in personal auto is a cycle. A deteriorating loss ratio plus a 27.3% increase in underwriting expenses on commissions and advertising, while average premium per policy falls, is a competitive-position question. GEICO is spending materially more to win business it is charging less for, into rising claims costs.

Assessment: This is the first datum in years that makes us question whether GEICO's cost advantage is still converting into profitable share. We are not calling the moat broken on one quarter. We are saying the burden of proof has moved.

4. Reserve releases are doing more work than the headline admits

Favorable prior-year development of roughly $877M pre-tax in the quarter, against roughly $13M net adverse a year ago, and consolidated underwriting still fell $354M with no catastrophes in either period.

Assessment: Berkshire's reserving is famously conservative and these releases are legitimately earned. But they are a stock, not a flow. When a segment needs $890M of incremental development swing to still print a decline, the underlying trend is roughly $1.2B worse than the reported one, and that is the number we are modelling forward.

5. Insurance investment income is in a structural, four-quarter decline

Pre-tax investment income fell 7.9%, with interest income down 12.8% and dividends flat, on lower short-term rates and Treasury bill reinvestment.

Assessment: This is the mechanical justification for everything in the capital-allocation section. As Treasury bills reprice lower, the opportunity cost of holding them collapses and deployment becomes structurally attractive rather than a matter of managerial preference. Expect the dividend line to start offsetting the interest line from Q3 as the new equity positions season.

6. BNSF is compounding, and the after-tax number hides it

Pre-tax earnings up 13.9% on volumes up 6.5% and revenue per unit up 7.6%, with the reported plus 6.3% after tax entirely explained by a one-time state tax benefit in the year-ago quarter.

Assessment: Intermodal share gains against tightening truck capacity are the most durable competitive development in the company this quarter. If anyone builds a sum-of-the-parts on Berkshire, BNSF's contribution should be rising.

7. The industrial recovery is real but partly subsidised

Manufacturing pre-tax rose 26.8% with industrial products up 41.0%. Inside that, trade-tariff refunds on 2025 duties, advanced-manufacturing production tax credits at Duracell, and gains on business divestitures at Marmon each contributed. Management is explicit that other building products pre-tax fell 8.8% before such items.

Assessment: Aerospace, electrical and industrial capital-spending exposure is genuinely inflecting; the refunds and credits are not repeatable. We would carry roughly two-thirds of the reported growth forward.

8. Two explicit forward warnings, both pointing at the second half

Management states that IMC's second-half earnings "are expected to be negatively impacted by the rise in raw materials costs," and separately that "inventory cost and supply chain uncertainties could negatively impact TTI's gross margins in the future." Both businesses were among the quarter's strongest performers, and part of IMC's demand came from customers accelerating purchases.

Assessment: Berkshire almost never guides. When it volunteers a forward negative on two named businesses in the same filing, it is worth writing down. Pulled-forward demand plus rising input costs is a recipe for a weaker second half in precisely the units that led the first.

9. Kraft Heinz: the impairment was deferred, not resolved

At June 30 the carrying value of Berkshire's Kraft Heinz stake exceeded its fair value by $1.1B, or 12.2% of carrying value, and management concluded no impairment charge was required. The year-ago quarter carried a $3.76B after-tax write-down of this same holding.

Assessment: A 12.2% deficit is meaningful but not automatically an impairment under the other-than-temporary framework, and management's judgment here is reasonable. It is nonetheless a live liability against reported earnings, and a reminder that the equity method holdings carry write-down risk the marked-to-market portfolio does not.

10. Financial-strength discipline is being maintained through the spending

Berkshire's own debt fell $2.3B to $20.4B, with $3.3B of maturing debt repaid against $1.7B of new yen notes issued in April at a 2.4% weighted average rate. The $30B cash floor on repurchases is restated. Float rose $1.1B to $177.5B at negative cost.

Assessment: The deployment is being funded from operating cash flow and the Treasury bill pile, not from leverage. Whatever else changes under Abel, the balance-sheet conservatism has not.

What They're NOT Saying

1. Any explanation of what was bought. Roughly $23.5B of equity purchases, and the 10-Q discloses only an alphabetical list of the five largest holdings and a category cost-basis table. Investors will not learn the composition until the Form 13F. For a deployment of this magnitude and this much thesis significance, a quarter of blind time is a real information gap.
2. Anything at all about GEICO's pricing response. Management quantifies the frequency and severity increases with unusual precision but says nothing about rate filings, planned price increases, or when the loss ratio is expected to stabilise. In a business where average premium per policy is falling into 10% to 12% bodily-injury severity inflation, that silence is the most conspicuous omission in the filing.
3. Whether the buyback continued into July. The repurchase table stops at June 30, and the stock has risen roughly 7% above June's average repurchase price since. Given that the June acceleration was the quarter's clearest signal, whether it survived the rally is the single most valuable unknown for a buyer today.
4. Any framing of Taylor Morrison's economics. The deal closed three weeks before the filing and Berkshire discloses only price and timing, citing impracticability on the purchase-price allocation. No revenue, no earnings, no return expectation, no commentary on how a national homebuilder fits alongside Clayton at this point in the housing cycle beyond a single sentence in a separate press release.
5. Why the equity sales stopped so abruptly. Sales fell from $24.1B in Q1 to $3.7B in Q2. Something changed in the assessment of the existing book, and nothing in the filing addresses it.
6. The wildfire liability trajectory. BHE's energy operating expenses rose on vegetation management and wildfire prevention, and the filing carries the usual litigation disclosures, but there is no updated framing of PacifiCorp's aggregate exposure. This remains the largest unquantified liability in the company.
7. Anything about succession below Abel, or about Buffett's role. The second quarterly report of the new era passes without a word on the management structure, the investment-management arrangements, or the Chairman's involvement in capital allocation. The repurchase framework still formally requires the CEO to consult the Chairman, which is a governance detail with a finite life and no disclosed plan.
8. The four reporting units near impairment. The filing repeats that four reporting units' fair values did not exceed carrying values by at least 20% at the last annual review. They are not named, and against $83.1B of goodwill that is a standing, unquantified risk that gets restated each quarter without progress.

Book Value and Valuation

Book value per share compounded 3.3% in the quarter, helped by both retained earnings and the buyback's accretion. The stock compounded faster. Since our Q1 note, priced off the May 4 reaction close of $468.52, BRK-B has appreciated 11.4% to $521.80, and the multiple has expanded from roughly 1.39x book to roughly 1.50x. The shares closed Friday within 0.5% of their 52-week closing high of $524.61, against a 52-week closing low of $461.47.

The relative picture is more nuanced and is the strongest remaining argument for the bulls. Berkshire is up 3.8% year to date against the S&P 500's 13.3%, so it still lags a market that keeps making highs. But that gap has narrowed sharply: at the Q1 report Berkshire was down 5.9% year to date against a market up 5.6%, an 11.5-point deficit that has since compressed to 9.5 points while the absolute price rose 11%.

Assessment: Our upgrade thesis had a specific, measurable edge: buy a fortress compounder at 1.39x book alongside a company that was itself repurchasing stock at that level, creating a self-reinforcing floor a fairly-valued stock does not enjoy. At 1.50x, with management's last disclosed purchases struck 7% lower, that floor has moved meaningfully below the market. Book value is compounding at roughly 2 to 3% a quarter, so the multiple can be grown into. But growing into a multiple is a very different proposition from buying a discount, and we should be honest about which one we now own.

Market Reaction

• Pre-print setup: BRK-B closed at $521.80 on Friday, August 7, up 3.8% year to date against the S&P 500's 13.3%, up 13.1% over the trailing twelve months and 5.5% over the trailing thirty days. The 52-week closing range entering the print was $461.47 to $524.61, placing the stock 0.5% below its high.
• Reaction session: none yet. Results and the 10-Q were posted at approximately 7:00 a.m. Central on Saturday, August 8, with markets closed, and Berkshire holds no earnings call, so there is no after-hours print or post-call move to report. The first session in which this quarter can be traded is Monday, August 10.

The absence of a reaction is itself worth a comment. Berkshire is the only mega-cap that systematically releases into a closed market, which removes the reflexive first-hour repricing that dominates most earnings and forces the print to be digested over a weekend. Practically, this means the quarter will be traded against two days of reflection rather than two minutes of headline reading, which historically has produced muted moves and a slower absorption of segment detail.

The setup entering that session is materially less favourable than the one we wrote about in May. Then, the stock was a persistent laggard trading near the bottom of its valuation range with the company newly buying its own shares. Now it has rallied 11% into the print, sits half a percent from a 52-week closing high, and carries a headline result whose growth rate is two-thirds currency and whose most important operating franchise deteriorated sharply. The capital-allocation news is genuinely good and may well carry the tape. The underwriting news is genuinely bad and is the part a weekend of reading tends to surface.

Street Perspective

Debate: Does the end of the selling streak re-rate Berkshire structurally?

Bull view: The multi-year defensive crouch is over. A CEO who will deploy $24B in a quarter across equities and buybacks, and another $6.8B on an acquisition three weeks later, deserves a higher multiple than one who hoards Treasury bills, because the cash drag that justified the conglomerate discount is being actively dismantled. The re-rating from 1.39x toward 1.50x is the start, not the end.

Bear view: One quarter of aggressive buying is not a policy, and the market has already paid for it. The cash pile is still $365B, or roughly 29% of total assets, and a single quarter of deployment barely dents it. Paying a higher multiple for capital deployment before knowing what was bought is paying for a press release.

Our take: The bulls have the better argument on direction and the bears on price. The behavioural change is real, larger than we forecast, and unlikely to reverse given where short rates are heading. But a re-rating that has already occurred is not an investment case, and we cannot evaluate the quality of $23.5B of purchases from an alphabetical list of five names. We would rather own this argument at 1.40x than at 1.50x.

Debate: Is GEICO's deterioration cyclical or competitive?

Bull view: Personal auto is a cycle business and this is a cycle. Loss costs run hot, carriers file for rate, margins recover with a lag. GEICO has navigated worse (2022's combined ratio was far uglier) and came out with better share and better margins. The combined ratio is still 91.2%, which is underwriting profit.

Bear view: The expense ratio is the tell. Loss-cost inflation alone would show up on one side of the ratio; spending 27% more on commissions and advertising while average premium per policy falls is what a business does when it is defending share against better-priced competition. Two consecutive quarters of accelerating decline, from down 35% to down 45%, is a trend.

Our take: We lean bear on this one now, having leaned bull last quarter. The loss-ratio move is defensibly cyclical; the simultaneous expense-ratio move is not, and the combination of falling price per policy and 10% to 12% bodily-injury severity is not a configuration that resolves quickly. GEICO's rate machinery is still the best in the industry and we expect it to respond, but rate filings take quarters to earn in. We would budget for at least two more weak prints, and the absence of any management commentary on the pricing response makes it harder to underwrite the recovery.

Debate: Is Berkshire still the right defensive holding for a narrowing market?

Bull view: A fortress balance sheet, $365B of dry powder, negative-cost float and a proven willingness to act into dislocation is exactly what you want against a market whose leadership keeps narrowing. Berkshire still trails the index by nearly ten points year to date, so the hedge is not yet expensive.

Bear view: The defensive discount has largely closed while the market kept rallying, so the hedge now costs more and the ballast is thinner. And the specific franchise that made Berkshire defensive, an insurance operation that earns through cycles, is the one currently deteriorating.

Our take: The bull case here is the reason this is a Hold and not something worse. Berkshire remains a superbly financed collection of businesses with a capital allocator who has just demonstrated he will act, and if the market breaks, this is where you want to be standing. But defensiveness is a function of price as much as balance sheet, and we bought the discount rather than the narrative. With the discount roughly two-thirds closed and the earnings quality deteriorating at the margin, in-line performance against the index is the honest expectation.

Model Update

Valuation framework: We anchor on price to book, adjusted for the fact that a growing share of Berkshire's value sits in operating businesses carried at depreciated cost rather than in a marked-to-market securities portfolio, which biases reported book downward over time. Book value per Class B share of $348.27 compounding at roughly 2 to 3% a quarter supports a book value approaching $380 to $390 by mid-2027. Berkshire's decade-long multiple range has broadly run from 1.2x to 1.6x, and the stock is now mid-range at 1.50x. Against a business whose non-insurance operations are accelerating and whose largest insurance franchise is decelerating, mid-range is the right place for it to sit. That combination produces a twelve-month expected return roughly in line with the market, which is the definition of our Hold.

Thesis Scorecard Post-Earnings

Overall: The thesis worked and, in working, consumed itself. Both bull pillars we upgraded on resolved in the same quarter: capital allocation turned decisively (confirmed, emphatically) and the valuation discount closed (challenged, and largely spent). Meanwhile the bear point we flagged as a watch item last quarter escalated from emerging to materializing, and a new one has been added on reserve dependency. The business is in good hands and the non-insurance operations are performing better than we modelled. The investment, as we originally framed it, has been realised.

Action: Downgrading to Hold from Outperform. We are taking the rating down on price and on insurance earnings quality, not on the franchise or on management. Berkshire at 1.50x book with a deteriorating crown jewel and a rising share of earnings coming from currency translation is a market-performer, not a market-beater. We would revisit Outperform on either of two developments: a return toward 1.35x to 1.40x book, which is where the company itself becomes a buyer again, or clear evidence that GEICO's rate actions are stabilising the loss ratio while acquisition spend normalises. We would move the other way if the combined ratio deteriorates a third consecutive quarter without a visible pricing response.