Berkshire Q3 2026 Buybacks: What $9.6 Billion Does to Share Count

Berkshire Q3 2026 buybacks totaled about $9.6 billion, according to the company's November 7 earnings release. Berkshire bought its own shares, reducing the shares outstanding. That spending is neither a purchase of the stocks in Berkshire's investment portfolio nor a percentage return earned by someone holding Berkshire stock.

Data through: Berkshire's November 7, 2026 earnings release and quarterly report for the period ended September 30, 2026. Repurchase figures below cover July through September. Published and checked: November 7, 2026, America/Phoenix (MST, UTC−07:00). This article uses issuer disclosures and hypothetical arithmetic; it reports no current share price or post-release stock return.

Berkshire Q3 2026 buybacks: count both share classes

Berkshire's earnings release reports about $9.6 billion of third-quarter treasury-share purchases and about $14.4 billion for the first nine months. The quarterly amount is part of the nine-month total. Adding the two would count the third quarter twice.

In Part II, Item 2 of the quarterly report, printed page 51, Berkshire lists purchases by month and class. Summing those rows gives:

Note 18 specifies that one Class A share converts into 1,500 Class B shares. On that equivalent basis:

``text Class B equivalent shares repurchased = (4,536 × 1,500) + 12,384,959 = 19,188,959 ``

This conversion compares economic share units. The classes have different voting rights, and their market prices need not stay at an exact 1,500-to-one ratio.

Buybacks and earnings per share: isolate the denominator

A repurchase can raise earnings per share even without an increase in total profit. The SEC's financial-statement guide explains the basic relationship between net income and shares. For a simple company with no preferred dividends or dilutive securities:

``text Basic EPS = earnings available to common shareholders / weighted-average common shares outstanding ``

Consider a hypothetical company, unrelated to Berkshire's reported earnings. It starts with 100 million shares and earns $200 million over a year. With no share-count change, EPS is $2.00.

Now suppose it repurchases 5 million shares on the first day of that year. Assume the repurchase has no effect on earnings, no other shares enter or leave circulation, and the remaining count stays at 95 million for the whole year.

EPS rises about 5.26%, even though total profit stays flat. Dividing the old share count by the new count gives the per-share change: 100 / 95 − 1.

A later repurchase has less effect on that year's denominator. If the same company has 100 million shares for the first half and 95 million for the second half, the simplified half-year weighting gives 97.5 million average shares and EPS of about $2.051. The year-end count is still 95 million. Actual calculations use the relevant timing rather than this half-year shortcut.

The assumption of unchanged profit isolates the arithmetic. A company spends cash to buy shares. It may then earn less interest on that cash or incur financing costs. An increase in EPS from fewer shares alone cannot establish whether management created value.

Shares repurchased, year-end shares, and average shares answer different questions

For a generic issuer, write down three separate quantities before calculating a buyback percentage:

• Gross repurchases: how many shares the company bought during the period.
• Net outstanding shares: how many remain on a particular date after repurchases, issuance, and other changes.
• Weighted-average shares: the time-weighted denominator for the period's EPS, with the relevant basic or diluted definition.

Suppose a hypothetical issuer starts with 100 million shares, buys back 5 million, and issues 3 million. Its ending count is 98 million, a 2% net reduction. Calling the gross 5% repurchase a 5% decline in outstanding shares would overstate the change.

For a company with multiple classes, keep the conversion basis consistent across dates. A holder's conversion from one class into another can change the raw counts without representing a buyback. Inspect the share roll-forward before attributing a decline in one class to cash spending.

An authorization also differs from completed purchases. Use the period's actual repurchase disclosure to measure execution; an announced spending limit describes permission. Neither supplies the return on the stock.

Berkshire stock and a Berkshire holdings snapshot are separate comparisons

Berkshire buys its own shares in a buyback. An investor comparing Berkshire's stock with SPY is studying the market-price path of an ownership interest in Berkshire. A replay of selected stocks that Berkshire disclosed answers a narrower allocation question.

Equity Replay's Berkshire snapshot guide explains the selection and normalization behind that example. The public portfolio page does not reproduce Berkshire's entire business, cash balance, liabilities, or repurchase activity. Its result cannot stand in for the return on Berkshire shares.

To inspect the issuer's stock, open Compare Stocks and select the supported Berkshire share class and a reference. Record the exact symbol, shared closing dates, and data-through label. Keep the same class throughout the comparison.

The release came on Saturday, November 7. A chart ending on Friday, November 6 cannot show a subsequent market reaction to that release. Wait for available closing data after the announcement before labeling any interval “post-release.” Even then, the comparison cannot isolate the effect of buybacks from other news.

The two-stock comparison guide explains date alignment and equal starting values. Do not add repurchase spending as a cash distribution to the price return of a shareholder who did not sell; the company paid the sellers of the repurchased shares.

Equity Replay applies starting weights to historical closing-price movement under its replay methodology. It excludes dividends, fees, taxes, deposits, withdrawals, trade execution, and rebalancing. It is an educational historical simulator, not investment advice, a broker, or a record of trades, and it does not reconstruct a buy-and-hold account result. These examples provide no prediction or instruction to buy, hold, sell, or rebalance a security.