Invest1 publisher3 min readPublished
A $3B venture fund now needs four of the 81 companies worth $25B
Twenty-one of those 81 carry the number while still private, so the comps a Series A gets underwritten against are marks rather than distributions, and the arithmetic that produced four is sensitive to one ownership input.
The Investor · Invest desk
What happened
- The count of companies worth $25B or more has gone from 6 public and none private, to 60 public and 21 private, a combined 81 in the chart SaaStr is working from.
- SpaceX closed its all-stock acquisition of Anysphere, the maker of Cursor, on August 14 at a value of $60B, issuing about 391 million Class A shares as confirmed in an 8-K.
- Cognition announced a Series E of more than $2B at a $48B valuation on September 8, led by Andreessen Horowitz and Accel with Founders Fund, General Catalyst and Avenir.
- PitchBook's July 2026 data counts 63 active US decacorns against 53 a year earlier and 26 in 2021, with $100M-plus rounds taking 87.5% of the first half's $412.7B.
- Menlo Ventures announced $3B in new capital in June, the largest raise in its 50 years, against a stated Series A entry target of around 20% ownership.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint Ownership sets the bar at the Series A: with the stake a lead actually keeps to exit, a plan that tops out below $25B cannot carry a fund of this size, so it does not get the meeting.
- exposure About a quarter of the comp set is unrealised, because those valuations were set by primary rounds, and a founder is being priced against numbers that only pay if a buyer or a listing eventually honours them.
- precedent The record acquisition required a purchaser whose public stock was days old, so anyone writing a $60B outcome into their underwriting is assuming access to a currency almost nobody has.
- contradiction SaaStr's own two counts disagree, and the 13.5x growth in supply only works from the larger one, so how much the target outcome has actually multiplied depends on which figure the reader takes.
Cognition's May round priced $26B against a $492M run rate, which is 52.8 times, and the September round priced $48B against a run rate the company puts near $900M, which is 53.3 times [14][13][5]. The valuation nearly doubled in a little over three months and the multiple moved about half a turn, so what the $48B records is the revenue arriving rather than private software being repriced. Cursor is a different shape: a $60B all-stock deal at roughly 15 times about $4B of annualized revenue [9][11], 2.05 times the $29.3B mark its Series D set in November [10][9], and 24 times the $2.5B it carried at the start of 2025 [10].
The fund arithmetic is the part worth checking, because one input carries it. A $3B fund needs $9B to $10B gross for 3x, and at 10% terminal ownership a $25B outcome returns $2.5B, which is four of them [19]. Hold 8% instead, the low end of the band SaaStr allows an early lead that defends its position, and each exit returns $2B and the requirement is five [18][6]. Hold 15% and it is three [7], at which point the $2B company is back in the room. Carta's medians say which way the error runs, with founding teams going from 56% at seed to 36% after the A to 16.1% by Series C, and the lead riding the same slide unless it pro-ratas into rounds this cohort prices too large for most to follow [17][18]. Four or five names out of 81 is 4.9% to 6.2% of every company on earth worth $25B, in one fund, in ten years [8].
What that does to a Series A meeting is a resource-allocation claim before it is a valuation claim, since partner time budgeted against four $25B outcomes is time not spent on the $2B outcome, which is SaaStr's own conclusion when it says a credible path to $2B does not come up [22]; the post is written by a co-investor in Owner.com at $2.3B, citing a chart from a growth partner at Menlo, so the irony is doing its own work [23]. The soft spot is the denominator. Twenty-one of the 81 carry $25B while private, about 26% of the set [3][3], marked by the same market that put 87.5% of the first half's $412.7B into rounds of $100M or more, roughly $361B [8][4], and SaaStr's framing concedes the mechanism when it says the returns that used to come from the public market now have to be manufactured while the company is private [21]. The record comp needed SpaceX stock days after listing [12]. If two or three of the 21 clear at or above their marks, the four holds and the threshold is real; if they clear below, the same arithmetic returns a smaller number and much cheaper entries, and the count of 81 will have been the least informative figure in the chart.
What to watch
- Whether any of the 21 privately held $25B companies exits at or above its private mark, the test that will confirm or break the count on its own.
- PitchBook's next tally after 19 companies crossed $10B in the first half of 2026, and whether the $100M-plus share of dollars holds above 87.5%.
- Where terminal ownership for early leads settles inside the 8-12% band, since that is the difference between five required $25B outcomes and four.