Bloomberg LP rents the Terminal to more than 300,000 finance professionals and runs one of the world’s largest newsrooms. It has never published its financials. In July it held early talks with bankers about an IPO or a stake sale, while saying Michael Bloomberg has no plans to sell.
Company values in billions: the scenario model’s range, the probability-weighted value, bankers’ estimate and the price Bloomberg paid to buy back Merrill Lynch’s stake in 2008. The copper line marks the base case.
Written on October 21, 2026. Figures are from company statements and news reports, as linked; they are not audited. Not investment advice. This company is private, so there is no share price; values are for the whole company. The author holds no shares.
Founded in 1981 to sell bond prices to traders, Bloomberg LP now makes most of its money from the Terminal, which costs about $30,000 a year per user and is found on almost every trading desk. Around it sit enterprise data feeds, indices, trading and risk software, and Bloomberg Media: Bloomberg News (about 2,700 journalists), Businessweek, television, radio and Bloomberg.com.
The pieces reinforce each other: the news makes the Terminal more useful, and Terminal revenue pays for the newsroom. Decades of proprietary prices and articles are also valuable training data for AI models, which is one reason an AI lab or exchange could want a stake.
Bloomberg does not report results, so all figures are outside estimates. Revenue has grown slowly for years, mostly through annual price increases rather than more users. The Terminal’s closest rival is LSEG’s Workspace (formerly Refinitiv Eikon); FactSet, S&P Global and newer AI research tools compete in parts of the market.
| Figure | Status | |
|---|---|---|
| Revenue | About $14.4–15B | Outside estimates |
| Profit margins | Not disclosed; widely described as highly profitable | Unknown |
| Debt | Not disclosed | Unknown |
| Terminal renewal rate | Not disclosed | Unknown |
A prospectus would be the first look at margins, renewal rates and how much the newsroom costs. Those numbers decide the value far more than any estimate here.
Michael Bloomberg, now 84, owns an estimated 88% and has said the company will eventually pass to Bloomberg Philanthropies. He recently moved some shares to the foundation, which would reduce taxes if the company listed or sold a stake. In 2008 Bloomberg bought back Merrill Lynch’s 20% stake for $4.4 billion, which valued the whole company at about $22 billion.
In July 2026 Semafor reported early, informal talks with bankers about an IPO or another transaction. The company replied that “Mike controls the firm” and has no plans to sell. There is no filing, date or exchange. A minority stake sold to an AI lab, exchange or asset manager is as likely as a public listing, and might never reach retail investors.
A mature, stable business, so a lower discount rate than the other private companies: 9% a year. Margins are an assumption, because Bloomberg has never disclosed them.
| Bear | Base | Bull | |
|---|---|---|---|
| Revenue in 2030 | $16B | $18B | $20B |
| Operating margin | 22% | 28% | 33% |
| Exit multiple | 18× | 22× | 26× |
| Company value today | $35B | $62B | $96B |
| Probability | 25% | 50% | 25% |
Weighted 25/50/25: about $64 billion. The multiple matters most. Quality data companies such as S&P Global and MSCI have often traded near 30× earnings; at 30× the base case gives about $85 billion.
Every figure in this valuation comes from one model script, rerun and checked against this page each time the site is built. The XLSX model rebuilds it in live formulas: change any blue input and every year’s cash flow, the bridge to the value, the scenarios and the sensitivity table recalculate.
| To justify the price, you need | Against |
|---|---|
| $80B: about $23B of revenue in 2030 at a 28% margin and 22× | About $15B today |
| Or the base case at about 28× earnings | Near what top data companies trade at |
| Risk | Why it matters |
|---|---|
| No deal | Bloomberg says there are no plans to sell; it may never list. |
| Retail access | A stake sale to an AI lab or exchange would leave nothing for individual investors. |
| Control | Even after a listing, the founder and foundation would likely control the vote. |
| AI tools | Cheaper AI research tools could slow Terminal price increases. |
| Unknown numbers | Margins, debt and renewal rates are unpublished. |
Upside: pricing power, recurring revenue and a data archive that AI developers would pay to use.
Bloomberg is dominant, recurring and slow-changing. Near the $80 billion bankers mention, the base case needs a multiple close to the best data companies: fair but not cheap; much above $100 billion it would assume the bull case.
What would change the value: A prospectus with audited margins and Terminal renewal rates would replace the estimates this report relies on.
Private companies are not rated: there is no price at which to buy them. The values above are for the whole company, before the dilution that future funding rounds would bring. How ratings work
Worth owning at a fair price; there is no deal yet.
| What | Figures | Source |
|---|---|---|
| Discount rate inputs | 10-year Treasury 5.15%, equity risk premium 4.5% | Fixed for every report so they can be compared; How ratings work |
| Valuation | Every value, case and sensitivity | This site’s model: XLSX, CSV |
| Checked by hand | Not yet | Reports are added to the audit log as they are revised |
| Input | Bear (10th percentile) | Base (median) | Bull (90th percentile) |
|---|---|---|---|
| Revenue, target year | $16B | $18B | $20B |
| Margin, target year | 22.0% | 28.0% | 33.0% |
| Exit multiple | 18× | 22× | 26× |
| Simulation result | |
|---|---|
| Expected value (average of 4,000 runs) | $63B |
| Likely range, 10th to 90th percentile | $39B to $90B |
| Median | $62B |
Each input is drawn from a split normal: its bear and bull values are the 10th and 90th percentiles, so the two sides can differ. Business inputs share a common factor (correlation 0.6); the discount rate is the same in every case, so it is not drawn. Inputs not listed are held at their base value. How the simulation works.