What is Berkshire Hathaway actually worth?
Berkshire is described as a conglomerate, which undersells how unusual it is. A conglomerate owns businesses. Berkshire owns businesses, and a $298 B portfolio of other people’s businesses, and enough Treasury bills to buy most of the S&P’s smaller members outright.
What follows builds the value from the ground up in seven charts: what you own, what it earns, what it is worth, and what you pay. The five assumptions that actually move the answer are yours to change as you read.
Berkshire is usually described as a conglomerate, but most of its value sits in securities and cash, not in the businesses it operates, which is why valuing it as one company, on one multiple, cannot work.
What you own
Start by sorting what Berkshire owns into three buckets, ordered by how readily each could be turned into something else. Operating businesses (a railroad, a utility group, an insurance underwriter and some seventy manufacturing, service and retail companies) are the least liquid. Investments (quoted equities, bonds and minority stakes) are marked by a market every day. Cash and Treasury bills are liquidity itself.
Valued separately and added together, the two financial buckets outweigh the operating one. That single fact explains most of what is strange about Berkshire, and most of what makes it hard to value with a single multiple.
The takeaway 62% of the estimated value sits in securities and cash, not in operating businesses, and cash and Treasuries alone are 35% of the whole company, at $369 B, very nearly as much as every operating business combined. Berkshire is best read as an insurance-funded investment company that also owns a railroad, not as an industrial group with a portfolio attached.
The float engine
An insurer is paid before it pays out. Premiums arrive today; claims are settled over years. The gap is float, money that will eventually belong to policyholders, which the insurer invests in the meantime for its own account.
If underwriting breaks even, float is free. If underwriting is profitable, float costs less than nothing, you are paid to hold other people’s money. This is the mechanism most summaries of Berkshire skip, and it is the reason a company with $717 B of equity controls $1,222 B of assets without meaningful borrowing.
It is also a diminishing advantage, which is the part the bulls skip.
The takeaway Float is real leverage and it compounded at 8.0% for 26 years, but it fell from 45% to 25% of book value along the way. The engine still runs; it is no longer the main driver. Berkshire’s returns now depend far more on what it does with its capital than on how cheaply it funds itself.
What it earns
Reported net income is close to useless here. Since 2018, accounting rules have forced Berkshire to run unrealised gains and losses on its equity portfolio through the income statement. In 2022 that produced a $22.8 billion net loss in a year when the operating businesses earned a record $30.9 billion. In 2021 it produced an $88.8 billion profit that nobody actually earned.
So build earnings from the parts instead: normalised after-tax profit for each operating business, plus the interest the cash genuinely throws off, plus Berkshire’s share of the earnings of the companies it owns slices of, look-through earnings, in Buffett’s phrase.
The takeaway About $53.2 B of adjusted after-tax earning power, or $24.63 per share. Note where it comes from: a little over half is generated by the operating businesses, and the rest by the portfolio and the T-bills.
Where the cash goes
Earning power is only half the question. The other half is what happens to the cash, because a dollar retained and reinvested badly is worth less than a dollar paid out. Berkshire has paid no dividend since 1967, so every dollar of the $597 B of operating cash flow since 2006 was reinvested by the same few people.
Four destinations: capital expenditure inside businesses already owned, acquisitions of new ones, purchases of quoted shares, and repurchases of Berkshire’s own stock.
The takeaway Most of the cash went into businesses Berkshire already owned, $244 B of capital expenditure, well ahead of the $147 B spent on acquisitions. The recent reversal is what matters for the valuation: net equity sales of $172 B since 2023 are why cash and Treasuries are now the single largest line in the sum of the parts, and why the question “what will they do with it?” dominates everything that follows.
“Despite our conservatism, I think we will be able every year to build the underlying per-share earning power of Berkshire.”
Warren Buffett · 2014 Annual Report
What it’s worth
Now add it up, bucket by bucket. Operating businesses at an earnings multiple each, investments at market less the tax that would fall due on unrealised gains, cash at face. That is the base case. Two adjustments bracket it.
Downside assumes the things that could reasonably go wrong at once: a tenth off the quoted portfolio, the insurance liquidity reserve treated as genuinely unavailable, and parent-company debt deducted in full. Upside assumes the cash gets used, roughly $100 B deployed at the rate Berkshire has historically created value per dollar acquired.
The takeaway $458 to $525 per share, centred on $495. The range is deliberately narrow for a company this size, because two-thirds of the value is marked to an observable market, not estimated.
The full sum-of-the-parts table, every segment, multiple and per-share conversion
| Segment | Revenue | Adj. net profit | Multiple | Estimated value | Per B share | % of base |
|---|
A multiple shown as “, ” means the item is carried at face or market value, not on earnings. Deductions represent the full potential liability, not an expected cash outflow. Per-share figures use 2.16 billion equivalent B shares. Intrinsic value is expressed against 2025 year-end book value per share of $332.35, both are 2025 figures, so that multiple compares like with like. The price-to-book multiple uses the latest reported book value, $348.26 at Q2 2026 (Q2 2026 Form 10-Q), because a current price over a year-old balance sheet overstates the multiple. See the book value history for the full 1999–2025 series.
What you pay
A valuation is only useful next to a price. The convenient thing about Berkshire is that book value per share, while a poor measure of intrinsic value, is an honest and verifiable one: it is audited, published quarterly, and has compounded at 10.5% for 26 years. That makes the price-to-book multiple a reasonable common language between the market’s opinion and this estimate. The multiples below use the latest reported book value of $348.26 a share (Q2 2026), not last year’s, a current price over a year-old balance sheet reads about a point of multiple too high.
Expressed that way, the three cases above become a band: 1.4× to 1.6× book value. Twenty-six years of history show where the market has actually been willing to trade.
What this valuation is really asserting
Buy at a multiple of book value, sell at the same multiple, and your return is the rate at which book value compounded in between.
Nothing else is needed for that to hold. Not a view on the multiple, not a price target. It follows from the definition, which is what makes the question this analysis exists to answer narrow and checkable: is today’s price above or below the level at which the compounding accrues to the buyer and not to the seller?
Berkshire’s book value per share has compounded at 10.5% a year since 1999, 26 years through two recessions and a pandemic. This analysis assumes approximately 10% a year continues, slightly below the record, because Berkshire is now far larger than it was for most of it and size is the one headwind management has named repeatedly itself. On that assumption the base case of $495 is the price at which a buyer should expect to earn about 10% a year from the business, with no help from a re-rating. Pay above it and the compounding is partly spent on the purchase price; pay below it and the re-rating is added to the compounding.
The most recent half year is consistent with it. Book value per share rose 4.7% in the first half of 2026, from $332.55 to $348.26 (Q2 2026 Form 10-Q), an annualised 9.7%. Of that, 4.3% came from retained earnings and the rest from the share count shrinking 0.5% through buybacks.
The sum of the parts contains no growth assumption anywhere; it is multiples applied to normalised profit. The relationship above is read off that result instead of built into it, which is why it can be stated as arithmetic. Where the scenarios are compared with a current share price they are carried forward at 10% a year for the 0.5 years since December 31, 2025 and labelled as rolled forward.
What this is dated to. The balance sheet is the anchor: asset values, share count and book value come from the 2025 Form 10-K, with book value per share updated to the Q2 2026 Form 10-Q. The assumptions applied to them are not a reading of that year. The earnings multiples, the normalised profit and the rate of compounding are long-run judgements, chosen because they should hold across a cycle and not because they describe 2025. Insurance underwriting, for instance, is carried at a 17-year average and not at whatever the latest year produced. A year of unusually good or bad results moves the anchor; it is not meant to move the estimate by much.
The takeaway At 1.6× book value, the stock sits inside the estimated range. Worth holding in mind: a multiple of book is not a valuation. It works here only because Berkshire’s book value is unusually close to a liquidation floor, a quarter of it is literally Treasury bills.
What would change the answer
Every number above rests on an assumption, and they are not equally important. The chart below moves one input at a time across a defensible range, holding everything else at the base case, and reports the resulting value per share. The longer the bar, the more the answer depends on being right.
The takeaway Whether $369 B of cash and Treasuries eventually buys something worth more than par is the live question for a buyer at today’s price, and it is not one the financial statements can settle.
If any single assumption above looks wrong to you, that is the point, change it in the sliders and watch where it lands. The segment deep dives below set out the evidence behind each one.
The evidence, segment by segment
Each deep dive leads with its conclusions and keeps the full historical data underneath, for readers who want to check the work.
Full SOTP valuation
Every line of the sum of the parts, with yields, growth estimates and the implied multiple on each segment.
Open the full table →Book value per share
The current figure, the four-line derivation behind it, and book value, float and price to book from 1999.
See the derivation →Buffett and Munger
Sixty years, the structure they left behind, and what changes now that Greg Abel is chief executive.
Read the background →Capital allocation
Twenty years of cash flow and deployment, year by year, with the acquisition record annotated.
See the record →BNSF Railway
Volumes, pricing, operating ratio and capital intensity, benchmarked against Union Pacific.
Open BNSF →Berkshire Hathaway Energy
Twenty-seven years of assets, equity and return on equity, the buyback valuations, and what the operations actually consist of.
Open BH Energy →Manufacturing, Service & Retail
The industrial core: revenue, margins and earnings across all seven sub-segments since 2006.
Open MSR →Insurance and float
What float costs, the underwriting result split five ways, and the assumptions the valuation rests on.
Open insurance →GEICO
Twenty-eight years of underwriting against Progressive, with policy counts and the combined ratio split in two.
Open GEICO →The public equity portfolio
What the marketable securities are, what they are carried at, and what they contribute to the valuation.
Open the portfolio →Follow the research
New valuations and research notes, a few times a year. No promotional email, no sharing of your address, unsubscribe in one click.
Delivered by Buttondown. You can also subscribe by RSS.