Private-market intelligence
The most valuable private companies
Moonberg tracks the most valuable private companies, from $94B (Anduril) to $1.10T (Anthropic), each marked at its latest NPM Price — with year-end IPO odds and per-share estimates below.
Updated 8 Aug 2026 · every figure sourced & dated
Companies
Valuations are the latest NPM Price (Nasdaq Private Market) marks; IPO-by-year-end odds are implied by prediction markets. See how each figure is computed.
How Moonberg computes these numbers
Every figure is synthesized from public filings and licensed market data, so you can judge it for yourself. None of it is an appraisal, a price you can transact at, or investment advice.
Data sources
- NPM Price — Nasdaq Private Market
- The headline valuation mark. A daily private-market valuation series we read and re-display; we do not adjust it. This is the basis for the current valuation, the per-share estimate and the valuation-history chart.
- Polymarket & Kalshi — prediction markets
- The year-end outlook and IPO odds are implied by traded prices on these prediction markets. We host and transform their data; we never mirror it in real time.
- SEC filings — Form N-PORT & Form D
- Mutual-fund holdings (N-PORT) disclose fair-value marks for private positions; we aggregate these into a filer consensus where enough independent filers agree. Form D records the last primary round.
- Secondary marks — Yahoo Finance / Forge
- A secondary (employee/early-investor) share price. We cite it and disavow it as a company valuation — it is a different, narrower number, shown only for context and to scale per-share estimates.
- News — GDELT & Google News
- Headlines for the catalyst feed. These are cited and linked only; they are never load-bearing and never move a valuation.
How each figure is computed
Current valuation — the latest NPM Price mark, shown as-is with its as-of date, compared to the last primary round to show a premium or discount (current ÷ last-round valuation − 1), which may be positive or negative.
Year-end outlook — prediction markets price a ladder of "will the valuation exceed $X by December 31" contracts. Each contract's price is the market-implied probability the mark reaches at least $X, so the ladder is a survival curve. We repair any non-monotonicity with isotonic regression (the pool-adjacent-violators algorithm, PAVA) and read off the even-odds median — the level the market prices at 50/50. Because these contracts resolve on whether a level is ever reached, that median is a level the mark touches during the year, not a year-end spot.
IPO odds & timing — we read two distinct markets and never blend them: the odds a company is listed (publicly trading) by a date, and the odds an IPO is announced/confirmed by a date. We also surface a conditional debut-cap read — the market-implied valuation if it lists.
Per-share estimate — a company's valuation divided by an estimated diluted share count (itself estimated from the last round's implied valuation and secondary price, since the exact count is not disclosed). It scales a holding to a dollar figure; an IPO issues new shares and dilutes existing holders, so it is not a quote.
Valuation history — the NPM Price daily series over the trailing year, rendered as a server-side chart with a dated milestone table. We do not fabricate intermediate round valuations we do not hold.
When data is thin
Not every company has the same depth of signal. We show a section only when its underlying data clears a quality bar — enough independent filers for a holdings consensus, a liquid enough market for an outlook, a fresh enough mark. When the data is not there, we omit the section rather than pad it, and where a market is too coarse for a precise read we present an indicative range instead of a false point estimate. If a company has no genuine, unique signal at all, we do not publish a page for it.
Limitations & caveats
- "By year-end" is a touch question, not an end-state. A "will it exceed $X by Dec 31" contract resolves yes if the level is ever reached — a different, higher probability than being above $X exactly on Dec 31.
- Secondary marks are not the company's valuation. A secondary share price reflects a narrow, often-restricted transfer market and is shown for context only.
- Share counts are estimates. Private companies do not disclose exact diluted counts; the per-share figure is explicitly an estimate.
- Data can be stale. If a source has not refreshed within its window we suppress or flag the figure rather than present old data as current.
Freshness
A page's visible "Updated" date — and the machine-readable date search engines read — changes only when a figure you can see actually changes, not every time the underlying data is re-checked. Day-to-day rounding that does not change a displayed number does not move the date.
An estimate, not an appraisal or investment advice. Moonberg does not broker, sell, or hold any security.