BRK-B 10-K Analysis: Why Cash Flow Surged 50% While Earnings Fell 25%

Berkshire Hathaway, the $371 billion-revenue conglomerate spanning insurance, railroads, energy, and manufacturing, reported a 24.8% decline in net income to $67.0 billion in its FY2025 10-K — while operating cash flow surged 50.3% to $46.0 billion. The divergence is the filing's most important signal.

Every earnings headline reported "Berkshire profits fall 25%." The narrative wrote itself: Buffett's final full year as CEO ended in decline, the transition loomed, and the conglomerate's best days were behind it. The headline P/E of 15.7x appeared to confirm the thesis — Berkshire as a cheap, aging collection of industrial assets.

But the 10-K filed March 2, 2026 tells a fundamentally different story. Of the $67.0 billion in reported net income, $39.1 billion — 47% of pre-tax earnings — consists of unrealized investment gains that management explicitly calls "generally meaningless." Strip away this mark-to-market noise and Berkshire's operating P/E jumps from 15.7x to approximately 38x — a 2.4x gap that transforms "cheap conglomerate" into "expensive franchise." The real question isn't whether Berkshire's earnings fell. It's whether the $176 billion insurance float machine, operating at a -5.4% cost (Berkshire was paid $9.5 billion to hold other people's money), justifies that premium — and whether the $369 billion cash fortress represents Buffett's loudest farewell or Abel's greatest opportunity.

The Cash Flow Paradox — Why the Headline Is Wrong

Berkshire Hathaway's operating cash flow surged 50.3% from $30.6 billion to $46.0 billion in FY2025 while GAAP net income fell 24.8% to $67.0 billion. These numbers should not coexist in any normal company. A 25% earnings decline usually signals deteriorating fundamentals. At Berkshire, it signals accounting noise overwhelming operating reality.

The disconnect originates in a single GAAP rule: since 2018, unrealized gains and losses on equity securities must flow through net income. In FY2025, Berkshire's $294 billion equity portfolio generated $39.1 billion in pre-tax unrealized gains — 47% of the $82.5 billion in total pre-tax income. These gains require no cash, no customer, no employee, and no operational achievement. They simply reflect that the equity portfolio appreciated.

"We believe that investment gains and losses, whether realized from dispositions or unrealized from changes in market prices and exchange rates, are generally meaningless in understanding our reported periodic results or evaluating our periodic economic performance."

Management's own dismissal is the starting point for what we call the 4-Layer Earnings Decomposition — a framework that separates Berkshire's $82.5 billion in pre-tax income into four distinct economic engines, each with different persistence and appropriate valuation multiples:

Layer 4 alone represents 47% of reported pre-tax income. Assign it the 0x multiple that management's own language implies, and the headline P/E of 15.7x recalculates to approximately 38x on the ~$27.5 billion in after-tax operating earnings from Layers 1-3. That 2.4x gap between reported and operating earnings is the single most important number in this filing.

The operating businesses buried under that noise are strengthening. BNSF improved its operating ratio by 250 basis points to 65.5% on flat revenue ($23,350 million vs. $23,355 million) — a $591 million operating expense reduction driven by compensation savings ($341 million) and lower fuel costs ($256 million). Industrial products within the manufacturing segment grew pre-tax earnings 13.1% to $6.8 billion, quietly generating more incremental earnings ($791 million) than Pilot earns in total. Even Duracell posted a 2.8% earnings increase, though that required a three-year retroactive advanced manufacturing tax credit catch-up — before those credits, consumer products earnings "declined significantly." Berkshire Hathaway's operating cash flow surged 50.3% to $46.0 billion in FY2025 while GAAP net income fell 24.8% to $67.0 billion, because $39.1 billion of reported earnings consisted of unrealized investment gains that management explicitly called "generally meaningless."

The $176B Float Machine — Berkshire's Structural Moat

Insurance float — the money policyholders pay in premiums before Berkshire pays out claims — is the architectural foundation of the entire conglomerate. In FY2025, that float reached $176 billion, growing from $171 billion in 2024 and $169 billion in 2023. What makes float extraordinary at Berkshire isn't its size alone. It's its cost.

"Float was approximately $176 billion at December 31, 2025, $171 billion at December 31, 2024 and $169 billion at December 31, 2023. The cost of float is measured as the ratio of pre-tax underwriting earnings to float balances. Our combined insurance operations generated pre-tax underwriting gains in each of the three years ending December 31, 2025 and the average cost of float was negative in each year."

A -5.4% cost of float means Berkshire wasn't just borrowing $176 billion for free — it was being paid $9.5 billion annually for the privilege. GEICO contributed $6,824 million in underwriting profit, Berkshire Hathaway Primary Group added $785 million, and Berkshire Hathaway Reinsurance Group generated $1,851 million. Combined, the insurance segment then invested the float alongside shareholder capital across a $529 billion portfolio — a sum exceeding Belgium's GDP — earning an additional $15.3 billion in investment income.

The combined Layer 1 and Layer 2 income — $24.7 billion pre-tax — represents 47.8% of Berkshire's total segment operating income of $51.7 billion. This is not an insurance company with some industrial assets attached. It's a $176 billion negative-cost capital engine with industrial businesses as ballast.

No peer operates at this scale or cost. Aflac's total invested assets ($116 billion) are smaller than Berkshire's cash pile alone ($369 billion). AIG's total assets ($161 billion) are one-seventh of Berkshire's $1,222 billion. Berkshire also maintains a $14.9 billion Japanese Yen carry trade at 1.2% — borrowing in low-rate Yen while earning ~5.3% on U.S. T-bills creates approximately $611 million in annual arbitrage income, while simultaneously hedging its known Japanese equity holdings in five major trading houses.

The critical question is persistence. Float has carried a negative cost for three consecutive years, but a single major catastrophe ($10 billion+ industry insured loss) could push the combined ratio above 100% and flip the cost positive. The filing's disclosed claim payments of $57.3 billion in FY2025 plus $4.4 billion in life benefits demonstrate the operational scale that underwrites — and threatens — this franchise. Berkshire Hathaway was paid $9.5 billion in FY2025 to hold $176 billion of insurance float at a -5.4% cost, generating negative-cost capital at a scale that exceeds the total invested assets of peer insurer Aflac ($116 billion).

Buffett's $369B Farewell — The Capital Allocation Signal

In Warren Buffett's final full year as CEO, Berkshire Hathaway accumulated $369 billion in cash and Treasury bills — 12.3 times its stated $30 billion minimum — while returning exactly zero dollars to shareholders. No dividends. No buybacks. For 18 consecutive months.

"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."

The buyback trajectory tells the story: $9.2 billion (2023), $2.9 billion (2024), zero (2025). The constraint was not liquidity — at $369 billion, Berkshire could have repurchased 35% of its outstanding shares and still held more than twelve times the floor. The constraint was valuation. Buffett concluded that BRK was at or above intrinsic value at every price it traded during 2025. For the investor who has deployed more capital into equity markets than any human in history, 18 months of inaction is a statement.

Then Greg Abel assumed the CEO role and resumed buybacks on March 5, 2026 — within days of the transition. Two interpretations matter: either Abel holds a lower intrinsic value estimate than Buffett (meaning he considers BRK undervalued at prices Buffett rejected, which is bullish) or he is deploying buybacks as a confidence signal during a leadership transition (neutral, fading over time). The filing's capital allocation language is unchanged, but the behavior diverged immediately.

Meanwhile, the $369 billion cash fortress earns approximately $19.5 billion pre-tax at current ~5.3% T-bill yields. That income is embedded in current earnings — and it's vulnerable. Every 100 basis point decline in short-term rates costs approximately $3.7 billion in pre-tax income, or 4.5% of total segment earnings. If rates normalize to 3%, roughly $8.5 billion of current income evaporates. The only risk factor Berkshire escalated from prior filings was geopolitical: "We believe risks of adverse effects from geopolitical events are rising." The cash position is both a shield against that risk and a $17.4 billion annual opportunity cost relative to the 10% total return the market demands.

Berkshire Hathaway accumulated $369 billion in cash and Treasury bills — 12.3 times its stated $30 billion minimum — while returning zero capital to shareholders for all of FY2025, the largest liquidity buildup by a public company in history.

The Permanent Capital Stress Test — Pilot, PacifiCorp, and the Cost of "Never Sell"

Berkshire's "buy and never sell" model compounds winners over decades. It also traps losers indefinitely. Three positions in the FY2025 10-K quantify the cost of permanence — and the structural defenses that contain the damage.

Pilot Travel Centers is the most dramatic deterioration. Pre-tax earnings collapsed from $968 million in FY2023 to $614 million in 2024 to $190 million in 2025 — an 80.4% two-year decline. On $42.2 billion in revenue, Pilot now operates at a 0.45% margin. For context, Pilot earns less in a full year than Berkshire's insurance segment generates in investment income in five days.

"Pilot's pre-tax earnings declined $424 million (69.1%) in 2025 compared to 2024. Earnings in 2025 were negatively impacted by lower wholesale fuel and in-store gross margins and higher selling, general and administrative expenses, primarily due to higher employee compensation and benefits, insurance and maintenance costs, as well as by charges from adjustments to certain fuel-related balance sheet accounts."

The "fuel-related balance sheet accounts" language is telling — it implies asset write-downs or inventory adjustments beyond operating margin compression. This is not a cyclical dip; the margin has compressed from roughly 1.87% to 0.45% over two years. McLane, Berkshire's distribution business with comparable revenue ($51.0 billion), generated $694 million in pre-tax earnings — 3.7 times Pilot's contribution on similar scale.

PacifiCorp's wildfire exposure presents a different risk profile. Cumulative booked losses stand at $2.85 billion, with $1.7 billion already paid. External estimates range as high as $55 billion. But the accrual trajectory is sharply declining: $1.9 billion (2023), $346 million (2024), $100 million (2025) — a 95% decline from the initial recognition peak. An Oregon Department of Forestry Report found PacifiCorp's lines "did not contribute to the overall spread" of the Santiam Canyon fires, strengthening the defense narrative.

The critical structural defense is the non-recourse debt firewall. Of Berkshire's $129.1 billion in total consolidated debt, approximately $83.5 billion sits at BHE and BNSF subsidiaries.

"Berkshire does not guarantee any borrowings of BHE, BNSF or their subsidiaries."

This single sentence means PacifiCorp's liabilities are legally isolated within BHE. The cost of that isolation is visible in borrowing rates: BHE subsidiaries issued $4.3 billion in new debt at 6.2% weighted average, compared to the parent's 1.2% on Yen borrowings. That 500 basis point spread represents approximately $215 million in annual incremental borrowing costs — the price Berkshire pays for structural separation. Even in the tail scenario ($55 billion, 7.7% of equity), the non-recourse structure means these liabilities cannot migrate to the Berkshire parent through debt channels, though reputational risk could force a voluntary backstop.

Kraft Heinz rounds out the trifecta with a $5.0 billion goodwill impairment driving equity method losses of $619 million — a known problem that Berkshire absorbs through accounting rather than cash outflows. At a combined ~$984 million in goodwill impairments across manufacturing ($788 million) and service/retailing ($196 million) segments, the pattern of historical acquisition overpayment extends beyond Kraft Heinz into the broader portfolio. The impairments are not existential — they represent 1.9% of total segment pre-tax earnings — but they quantify the ongoing cost of Berkshire's permanent capital philosophy. Pilot Travel Centers' pre-tax earnings collapsed 80.4% over two years to $190 million on $42.2 billion in revenue, producing a 0.45% operating margin that makes Berkshire's 2023 acquisition the worst-performing major deal in the conglomerate's recent history.

What You're Actually Paying — The 38x Reality

At approximately $1.05 trillion, every financial terminal shows Berkshire Hathaway trading at 15.7x earnings, 1.46x book value, and 9.8% ROE. The GAAP numbers place BRK-B squarely in line with insurance peers: Aflac at 16.1x, AIG at 15.6x, Unum at 18.1x. On the surface, Berkshire looks correctly priced within the peer band.

That surface is an illusion. At ~38x operating earnings — after stripping the $39.1 billion in "meaningless" mark-to-market gains — BRK trades at a 2.4x premium to the ~16x peer average. The question is whether the premium is justified.

The 4-Layer sum-of-parts provides the answer:

At ~$1.07 trillion, the sum-of-parts aligns within 2% of the $1.05 trillion market price. An investor buying at current prices pays a correct price for the float franchise, a fair price for operating businesses, and face value for the cash — with zero margin of safety. Buffett's 18-month refusal to repurchase shares is precisely consistent with this math.

The ~38x operating P/E implies the market expects operating earnings to grow approximately 8-10% annually for five years, compressing the multiple to ~22x at the end of the period, to deliver a 10% total return. The filing shows actual segment pre-tax earnings grew approximately 6.6% year-over-year ($51.7 billion vs. ~$48.5 billion), driven by insurance (+5.3%), BNSF (+7.9%), and industrial products (+13.1%). Meeting the embedded growth expectation requires sustained performance at or above the FY2025 run rate, with no deterioration in float economics.

At approximately $500 per share, the market implies Berkshire's float machine will persist at negative cost, Abel will deploy cash at returns above T-bill yields, and PacifiCorp's wildfire exposure will resolve within the current accrual trajectory. The filing supports the first two assumptions but complicates the third — and provides zero cushion if any of the three falters. Berkshire Hathaway's operating P/E of approximately 38 times — after excluding $39.1 billion in mark-to-market investment gains — represents a 2.4-times premium to the 15.8 times average across insurance peers Aflac, AIG, and Unum Group.

What to Watch

Five metrics will determine whether the FY2026 10-K confirms the current thesis or breaks it:

1. Cost of float (quarterly): If any quarter shows an underwriting loss (combined ratio above 100%), the negative-cost-of-float thesis requires revision. A major natural catastrophe ($10 billion+ industry insured loss) is the most likely catalyst. Bull: float exceeds $180 billion at continued negative cost. Bear: any quarter with positive cost.

2. Capital deployment pace: The $9.5 billion OxyChem deal (January 2026) is only 2.6% of available cash. If Abel deploys more than $50 billion at 8%+ ROIC within the next four quarters, the optionality thesis strengthens materially. Bull: more than $50 billion deployed at 8%+ returns. Bear: cash grows beyond $400 billion with sub-5% deployment returns.

3. Pilot margins: FY2025's 0.45% operating margin is either a cyclical trough or a structural impairment. Bull: margin recovers above 1.0% ($420 million+ pre-tax on ~$42 billion revenue). Bear: margin stays below 0.5% for two more quarters, confirming secular decline.

4. PacifiCorp trial verdicts: The declining accrual trajectory ($1.9 billion to $346 million to $100 million) is encouraging, but any single James trial verdict exceeding $500 million would reverse the trend and stress-test the BHE non-recourse firewall. Bull: total FY2026 accrual below $100 million. Bear: single verdict exceeding $1 billion.

5. T-bill rate sensitivity: At $369 billion in short-term instruments, every 100 basis point rate decline costs $3.7 billion in pre-tax income. If the Fed cuts to 3% by year-end, approximately $8.5 billion of current earnings disappear — unless Abel has redeployed the cash into higher-returning assets. Bull: rate cuts offset by capital deployment. Bear: rates fall 200 basis points with cash still parked in T-bills.

Frequently Asked Questions

Why did Berkshire Hathaway's net income fall 25% in FY2025?

Berkshire's net income declined 24.8% to $67.0 billion primarily because unrealized investment gains fell from $49.3 billion to approximately $39.1 billion — non-cash mark-to-market adjustments that flow through GAAP earnings since 2018. Management explicitly states these gains are "generally meaningless in understanding reported periodic results." The underlying operating businesses strengthened: operating cash flow surged 50.3% to $46.0 billion, BNSF improved its operating ratio by 250 basis points, and insurance underwriting generated a $9.5 billion profit.

What is Berkshire's cost of float, and how can it be negative?

Insurance float is money policyholders pay in premiums before Berkshire pays out claims. In FY2025, Berkshire's float was $176 billion, growing from $171 billion in 2024. The cost of float is calculated as pre-tax underwriting gain or loss divided by average float. In FY2025, Berkshire earned a pre-tax underwriting profit of $9,460 million on $176 billion float, yielding a -5.4% cost. A negative cost means Berkshire was paid $9.5 billion to hold $176 billion of other people's money, then invested those funds to earn an additional $15.3 billion. Float has been at negative cost in each of the three years ending December 31, 2025.

Why did Berkshire stop buying back stock in 2025?

There were zero share repurchases in all of FY2025, following a declining trajectory: $9.2 billion (2023), $2.9 billion (2024), then zero. Berkshire's buyback policy requires shares to trade below intrinsic value with a minimum $30 billion cash floor. With cash at $369 billion (12.3 times the floor), the constraint was valuation, not liquidity: Buffett concluded BRK was at or above intrinsic value for 18 consecutive months. Greg Abel resumed buybacks on March 5, 2026, within days of assuming the CEO role.

What happened to Pilot Travel Centers' earnings?

Pilot's pre-tax earnings collapsed 69.1% in FY2025 to $190 million, down from $614 million in 2024 and $968 million in 2023 — an 80.4% two-year decline. On $42.2 billion in revenue, this represents a 0.45% operating margin. The filing cites lower wholesale fuel margins, higher SGA expenses, and "charges from adjustments to certain fuel-related balance sheet accounts," suggesting inventory or asset write-downs beyond margin pressure.

How large is PacifiCorp's wildfire exposure, and is it contained?

PacifiCorp has recorded cumulative estimated wildfire losses of $2.85 billion, of which $1.7 billion has been paid. The filing acknowledges "reasonably possible significant additional losses." External estimates put total potential exposure at up to $55 billion. However, the accrual trajectory is declining sharply: $1.9 billion (2023) to $346 million (2024) to $100 million (2025). Critically, Berkshire does not guarantee BHE subsidiary debt, creating a structural firewall — PacifiCorp's liabilities are legally isolated within BHE.

What is Berkshire's real P/E ratio?

Berkshire's GAAP P/E is approximately 15.7 times ($1.05 trillion market cap divided by $67.0 billion net income). However, $39.1 billion of that net income consists of unrealized investment gains that management calls "meaningless." Excluding these and adjusting for equity method earnings quality, operating after-tax earnings are approximately $27.5 billion, yielding an operating P/E of approximately 38 times — a 2.4-times gap that is the most important valuation nuance in the filing.

How does Berkshire compare to insurance peers like Aflac and AIG?

Berkshire's GAAP P/E (15.7 times) appears in line with Aflac (16.1 times) and AIG (15.6 times), but Berkshire's total assets ($1,222 billion) are 7.6 times AIG's. BRK's cash alone ($369 billion) exceeds Aflac's total assets ($116 billion). No peer reports negative-cost float at Berkshire's scale, and no peer returned zero capital to shareholders in FY2025. The meaningful comparison is ROE: Aflac leads at 13.3%, followed by BRK at 9.8%, AIG at 7.5%, and Unum at 6.6%.

What does Abel's buyback resumption signal about Berkshire's future?

Greg Abel resumed share repurchases on March 5, 2026, within days of Buffett's retirement. Two interpretations: (1) Abel has a different, lower intrinsic value estimate than Buffett, meaning he sees BRK as undervalued at prices Buffett rejected — bullish for shareholders; or (2) Abel is using buybacks as a confidence signal during transition. The filing's capital allocation language is unchanged: "Financial strength and redundant liquidity will always be of paramount importance at Berkshire."

Is Berkshire's $369B cash position too conservative?

At $369 billion, Berkshire holds 12.3 times its stated $30 billion floor and 30.2% of total assets in liquid form. The cash earns approximately $19.5 billion pre-tax annually at ~5.3% T-bill yields. But the market's required return implies the $369 billion should generate ~$36.9 billion at 10%, leaving a $17.4 billion annual opportunity cost gap. However, the option value of deploying $369 billion during a market dislocation — as Berkshire did in 2008-2009 — could far exceed interim opportunity cost. Watch for deployment signals: if Abel deploys more than $50 billion at 8%+ ROIC within two years, the optionality thesis strengthens.

What is Berkshire's Yen carry trade strategy?

Berkshire increased Japanese Yen-denominated debt from $12.6 billion to $14.9 billion at a weighted average rate of 1.2%. Borrowing at 1.2% in Yen while investing in U.S. T-bills at ~5.3% creates approximately $611 million in annual arbitrage income on the JPY position alone. The Yen debt also serves as a natural hedge for Berkshire's known Japanese equity holdings in five major trading houses.

How does the BHE non-recourse debt firewall protect shareholders?

Total consolidated Berkshire debt is $129.1 billion, of which approximately $83.5 billion sits at BNSF and BHE subsidiaries. The filing states: "Berkshire does not guarantee any borrowings of BHE, BNSF or their subsidiaries." PacifiCorp's wildfire liabilities are legally contained within BHE. The cost of this isolation is visible in borrowing rates: BHE subsidiaries issued $4.3 billion at 6.2%, compared to the parent's 1.2% to 3.6%. The 270-500 basis point spread costs BHE roughly $112-215 million annually on new issuance.

What are the biggest risks to Berkshire's investment thesis?

Three risks dominate: (1) Rate sensitivity — $369 billion in T-bills at ~5.3% generates ~$19.5 billion pre-tax; every 100 basis point decline costs ~$3.7 billion. (2) Float cost reversal — negative-cost float requires sustained underwriting profitability; a major catastrophe ($10 billion+ industry loss) could flip the cost positive. (3) Capital deployment failure — if Abel cannot deploy $369 billion at returns above the ~5% T-bill rate, the cash fortress becomes value-destroying. The $9.5 billion OxyChem deal is the first test case, but it's only 2.6% of available cash.

Methodology

Data Sources

This analysis is based on Berkshire Hathaway's 10-K filed March 2, 2026 (FY ending December 31, 2025), accessed via the MetricDuck filing intelligence pipeline (5-pass LLM extraction) and direct filing text sections. Pipeline metrics were extracted from XBRL financial statements and tagged [PIPELINE]. All filing quotes are verbatim from MDA, Notes, or Risk Factors sections. Derived calculations include explicit formulas documented inline. Market data (share price ~$500, market cap ~$1.05 trillion) is sourced externally and carries approximately ±3-5% accuracy. Peer data for AFL, AIG, UNM, and ASB is drawn from MetricDuck's automated extraction pipeline for FY ending December 31, 2025.

Limitations

• Market cap and valuation metrics are manually sourced. BRK-B's dual-class share structure (Class A ~$750,000/share, Class B ~$500/share) breaks the automated pipeline extraction. All valuation ratios (P/E, P/B) are derived from an external market cap estimate of approximately $1.05 trillion, introducing potential inaccuracy of ±3-5%.
• Operating earnings are estimated. Berkshire does not report a single "operating earnings" line item. The $27.5 billion figure is derived by subtracting after-tax investment gains ($31.3 billion at 80% of $39.1 billion) and adjusting for equity method earnings quality from GAAP net income. Different analysts may define operating earnings differently.
• PacifiCorp external estimates are speculative. The $55 billion maximum exposure figure comes from external analysis (Insurance Journal), not from the filing. The filing states only that "significant additional losses" beyond $2.85 billion are "reasonably possible" without quantifying the upper bound.
• Peer comparison is inherently limited. No public company is comparable to Berkshire Hathaway. The assigned peers (AFL, AIG, UNM, ASB) capture only the insurance dimension. Berkshire's railroad, energy, manufacturing, and retail segments have no analogues among these peers.
• Apple portfolio concentration is not quantifiable from the 10-K. Berkshire's $294 billion equity portfolio is known to be concentrated in Apple, but individual holding details are disclosed in 13-F filings, not the annual report.
• Post-period events including the OxyChem acquisition ($9.5 billion, closed January 2, 2026), Abel's buyback resumption (March 5, 2026), and the PacifiCorp federal settlement ($575 million) are not reflected in the 10-K's financial statements.

Disclaimer:

This analysis is for informational purposes only and does not constitute investment advice. The author does not hold positions in BRK-B, AFL, AIG, UNM, or ASB. Past performance and current metrics do not guarantee future results. All data is derived from public SEC filings and may contain errors or omissions from the automated extraction process.