Private companies have no continuous price. Every reference point — a primary round, a tender, a broker quote, a fund NAV — measures something different.
Data coverage has improved, but observations remain sparse, lagged and often unverifiable.
Tender prices are usually set by the company, not by a market; most are struck at or above the last primary round.
A listed share has one price at any moment, quoted continuously and settled through a regulated market. A private company share has no such thing. It has a series of reference points, each produced by a different process, for a different purpose, at a different time. The growth of the secondary market has multiplied those reference points without making them equivalent, and the first discipline of private-market analysis is knowing what each one measures.
Five reference points, five meanings
Most valuations quoted for a private technology company trace back to one of the following:
- The last primary round. A negotiated price for newly issued preferred stock, usually carrying liquidation preferences and other rights that common stock does not have. It is a headline number, not a per-share value for every class.
- A company-run tender offer. A single price at which a company allows employees and early investors to sell, typically to approved buyers. Gunderson Dettmer found that the price matched the most recent preferred financing in 60% of the tender offers it reviewed [1].
- Broker or marketplace quotes. Indicative bids and offers for common stock or for interests in vehicles holding it. Useful for direction; often thin, and not always backed by completed trades.
- Completed secondary trades. The closest thing to a market price, but usually private, small relative to the cap table, and sometimes structured.
- Fund NAVs and adviser surveys. Portfolio-level marks reported with a lag, which is why LP portfolio pricing is quoted as a percentage of NAV rather than as a value.
Table 1: What each reference point actually measures
| Reference point | Set by | Main limitation |
|---|---|---|
| Last primary round | Company and lead investor | Prices preferred stock with rights common stock lacks |
| Company tender offer | Company, informed by recent rounds | Not an open-market clearing price; access is controlled |
| Marketplace quote | Broker or platform participants | Indicative; may not reflect completed transfers |
| Completed secondary trade | Buyer and seller | Sparse, private, sometimes structured or forward-based |
| Fund NAV | Fund manager, with auditor review | Reported with a lag; portfolio-level, not single-asset |
Source: Argent Bluebook, drawing on the cited references.
What the 2025 data showed
Adviser data gives a reasonable read on portfolio-level pricing. Jefferies reported that LP portfolios traded at an average of 90% of NAV in the first half of 2025, with buyout portfolios well ahead of venture [2]. On volume, the market set a record in 2025: Evercore put the year at about US$226 billion [3]. Neither figure, however, tells an investor what a single share in a specific company is worth, and no aggregate can substitute for the terms of an individual trade.
Why tenders are not market prices
Company-run tenders are the most visible pricing events in late-stage technology, and they are frequently misread. The price is generally anchored to the last primary round rather than discovered through competitive bidding, and participation is capped: Gunderson Dettmer found a median limit of 20% of vested holdings for current employees [1]. A tender clearing at par therefore says that the company chose to transact at its last round price, not that an open market cleared there.
The limits of aggregation
Three cautions apply to any private-market data set. First, survivorship: platforms report on the companies they can trade, which skews toward names in demand. Second, structure: some reported transactions are interests in vehicles or forward contracts rather than registered share transfers, and these carry different risks and different prices. Third, verification: unless a transfer has been approved and recorded by the issuer, a quoted price may not correspond to anything that can actually change hands.
That last point is not theoretical. In July 2025, OpenAI publicly stated that tokenised products marketed as exposure to its shares were not its equity and that any transfer of its equity requires company approval [4]. Whatever one makes of the products, the episode showed how far a quoted price can drift from the issuer's own records.
Using the data well
The practical approach is triangulation rather than dependence on a single number: compare the last primary round, any recent tender, observable secondary indications and listed comparables, and treat divergence between them as information about liquidity and structure rather than as noise. Where a gap cannot be explained, the appropriate response is usually more diligence, not a wider spread.
Better data has made private markets more legible. It has not made them continuous, and it has not removed the need to ask what, precisely, a quoted price refers to.