Most new US “venture funds” are a single deal

Where these numbers come from

A Form D is the notice a US private fund files when it starts selling interests under a Regulation D exemption. It is free, official, filed in real time, and it states the fund's name, its state, its target size, how much has been sold and how many investors are in. We have been collecting them since April and the corpus is committed to the repository, so every figure below is reproducible rather than asserted.

Three decisions shape every number here, and each one changes the answer:

• A fund is one entity, not one filing. Funds file an amendment at every close, so counting filings would invent new funds out of old ones. We key on the SEC's Central Index Key and date each fund from its first notice only.
• We date by the filing, not by the launch. A fund that started raising in June but filed in July sits in July. This is a property of the source, not a choice we can make differently.
• Only complete months are compared. April and August are partial and excluded from everything below. May, June and July have filings on every one of their 20, 21 and 22 US business days, so the comparison is not resting on a gap.

Two-thirds state a target under $5m

Across the three complete months, 2,544 entities filed a first notice as a venture capital fund. Sorted by the target they stated:

• No stated target — 476 (18.7%). Open-ended and evergreen vehicles. The size is genuinely unknown, not withheld.
• Under $5m — 1,646 (64.7%).
• $5m – $25m — 203 (8.0%).
• $25m – $100m — 114 (4.5%).
• $100m – $500m — 88 (3.5%).
• Over $500m — 17 (0.7%).

What is actually in that bottom band

A vehicle raising under $5m from a handful of people is a syndicate or an SPV — one company, one cheque, a wrapper created to hold it. On the SEC form it is indistinguishable from a $1.5bn institutional fund filed the same week: there is one box, and both tick it. The filers are not misreporting; the taxonomy has one slot where the market has two very different things.

That is not a hypothetical pairing. In the three days around 8VC's $1.5bn fund, filed on 29 June, 130 vehicles targeting under $5m filed alongside it — and on the form, they are the same kind of thing. The same held for the other two $1.5bn funds in these months: 112 and 115.

Our own read is that 70–71% of each month's cohort is a single-deal vehicle, and that share held at 71%, 70% and 71% across the three months independently — a mostly-noise rule would not sit that still. But that figure is a judgement of ours, from the vehicle's name and its average cheque, and we flag rows rather than deleting them. No filing is dropped from any total on this page. A count that depends on an unpublished heuristic cannot be argued with, which makes it worse than one that can.

This is why the size bands lead and the headline count does not. A reader can see the composition and disagree with our line without having to trust it.

Eighty per cent of the vehicles hold three per cent of the money

Counting entities and counting capital give almost opposite answers. Of the 2,068 vehicles that stated a target, the $42.0bn they are collectively chasing splits like this:

• Under $5m — 79.6% of the vehicles, 3.0% of the capital ($1.3bn).
• $5m – $25m — 9.8% of the vehicles, 4.8% of the capital ($2.0bn).
• $25m – $100m — 5.5% of the vehicles, 12.2% of the capital ($5.1bn).
• $100m – $500m — 4.3% of the vehicles, 46.1% of the capital ($19.4bn).
• Over $500m — 0.8% of the vehicles, 33.8% of the capital ($14.2bn).

Which number you count decides what you conclude

Five per cent of the vehicles hold eighty per cent of the money. The 105 funds targeting $100m or more account for 79.9% of everything being raised; the 1,646 under $5m account for 3.0%. The fifty largest funds alone are 59.8% of all targeted capital.

The vehicles we flag as single-deal are 70% of the count and 7.9% of the capital — which is the same finding stated from the other side, and it is why the distinction matters. Count entities and the market looks like a crowd of small vehicles. Count dollars and it is a few dozen institutions, with a long tail attached that barely moves the total.

Neither view is wrong. But a headline built on the count describes a syndication market, and a headline built on the dollars describes a fund market, and they are not the same market — so any figure quoted without saying which one it counted is not telling you much.

The cheque changes by 240×. The number of investors barely moves.

Dividing what each vehicle had sold by how many investors were in it, at the moment it filed:

• Under $5m — median cheque $25k, median 12 investors (n=1,590).
• $5m – $25m — median cheque $567k, median 13 investors (n=175).
• $25m – $100m — median cheque $2m, median 12 investors (n=61).
• Over $100m — median cheque $6m, median 10 investors (n=29).

Why that ladder is the clearest thing in the data

The cheque rises 240-fold from the bottom band to the top. The number of names on the filing does not move at all — it is about a dozen everywhere.

Our read, and it is a read: this is a first-close snapshot rather than a fact about how funds are structured. A large fund files early with its anchor investors in and fills the rest of the book over the following year, so the dozen names in the top band are the first dozen, not the final list. The bottom band's dozen, by contrast, is probably the whole thing. We cannot see the difference from the filing, and nobody should take these as final LP counts.

What survives the caveat is the cheque size, which is the number that actually separates a fund from a syndicate: $25,000 is a person writing an angel cheque; $6m is an institution.

One firm can be several filings

There is a second reason a raw count overstates the market, and it runs in the same direction. A single fundraise is routinely split across two or more legal vehicles — usually to separate classes of investor, or a main fund from its opportunity fund. Filed on the same day, by the same firm, for the same raise:

• Benchmark, the US venture firm, 30 June — two vehicles, $750m and $1.25bn.
• Union Square Ventures, 16 July — a $500m main fund and a $400m opportunity fund.
• Crosslink Ventures, 29 July — two vehicles of $350m each.
• FCA Health Innovations, 15 May — two vehicles of $175m each, one of them tagged “QP” for qualified purchasers.

Nearly three-quarters of the large funds file before the first dollar

Of the 105 funds targeting $100m or more in the three complete months, 76 (72%) filed with no sale having occurred at all. Of the ones that had sold something, the median had 40% of its target already committed.

A Form D is due within 15 days of the first sale, so filing with nothing sold is filing early, not late — and it is a choice, made by most of the large funds and by almost none of the small ones.

Our read, offered as ours: a large fund wants the exemption on the record before it begins talking to institutions, and the filing costs it nothing. A syndicate has no such sequence — the deal exists first and the vehicle is created to hold it, so by the time anyone files, the money is in. We cannot see intent in a filing and this explanation is inference, not evidence.

The practical consequence is worth knowing: a large fund's Form D announces an intention, and a small one records a completed transaction. Reading both as “money raised” double-counts the optimism.

New York's median fund is larger than California's

Taking only the 219 funds that stated a target of $25m or more — the ones that are recognisably institutional — and grouping by the state on the filing:

• California — 78 funds, $14.37bn targeted, median $91m.
• New York — 36 funds, $7.46bn targeted, median $138m.
• Massachusetts — 16 funds, $4.54bn targeted, median $58m.
• Texas — 9 funds, $2.27bn targeted, median $80m.
• Cayman Islands — 6 funds, $1.89bn targeted, median $325m, the largest median anywhere.

What we think that shows

California forms more than twice as many institutional funds as New York and targets roughly twice the capital — and its median fund is a third smaller. Our read: California carries the long tail of new seed managers, and New York skews later-stage, and towards crossover investors — public-market funds that also buy into private rounds — where a fund is bigger by construction. Massachusetts sits lower again on the median while appearing repeatedly at the very top of the size table, which is what a small number of very large firms looks like in a median.

The Cayman figure is the one to be most careful with: six funds is not a distribution. Offshore feeders exist to take institutional and non-US money, so a high median is what you would expect, but at n=6 we are describing six filings rather than a jurisdiction.

If you are raising, this is the number that matters

Strip out the syndicates and the vehicle-splitting and one figure is left that bears on a founder: 114 funds in the $25m–$100m band formed over three months — about 38 a month. That band is where new institutional seed funds live, and a fund that closes this year is writing first cheques for the next two or three.

Fund formation is the earliest public signal there is of who will be investing, and it runs a year or more ahead of any deal announcement. It is also the half of the market almost nobody watches, because it is unglamorous, US-only and buried in a regulatory filing nobody reads for pleasure.

The count of institutional-scale funds also rose across the three months — 68, 73 and 78 at $25m and above. Three data points cannot separate a trend from a season, and we are not going to call it one. Ask us again in six months.

Limits, stated plainly

• The fund type is self-declared. Every vehicle here chose “Venture Capital Fund” itself. We do not second-guess any individual filing.
• The SPV share is our judgement, not a fact. It comes from the vehicle's name and its average cheque. The size bands are the filers' own stated numbers and carry no judgement of ours at all — prefer them.
• United States only. A Form D has no UK or European equivalent; there is no way to run this analysis on the funds a British founder is most likely to meet.
• Targets are not money raised. A stated target is an intention. Where we quote what has been sold, we say so.
• Dated by filing, not by launch. A fund that began raising in one month and filed in the next lands in the next.
• April and August are excluded. Our coverage does not span either month completely, and a partial period sitting beside a complete one reads as a collapse.