Pre-seed vs seed, and what a SAFE costs founders

What is the difference between pre-seed and seed?

Pre-seed and seed are stage labels with no legal definition, and in practice they differ by size and by who sets the price: the median US seed round was about $3 million in 2025, three times its 2018 level, according to Crunchbase data reported by Crunchbase News on 29 October 2026, while venture deals at pre-seed are mostly raised on a SAFE, a short contract that converts into shares at the next priced round.

Datasets draw the line in different places. Crunchbase’s seed and angel category, as its 5 October 2026 funding review defines it, covers seed, pre-seed and angel rounds, plus venture rounds of unknown series, equity crowdfunding and convertible notes of $3 million or less, so pre-seed sits inside its seed totals. The law firm Gunderson Dettmer describes the financings in its Q3 2024 valuation report as running from pre-seed SAFE and convertible note rounds through Series D, so it labels pre-seed by instrument. A figure for “seed” is only usable with the source’s definition beside it.

	Pre-seed	Seed
Instrument	SAFEs in more than 92% of early-stage financings, Q3 2025 (Fenwick)	SAFE, convertible note or priced equity; no neutral split read
Size	Y Combinator’s standard deal invests $500,000; no neutral median read	Median US seed round about $3 million in 2025; lower and upper quartiles $1 million and about $5.6 million (Crunchbase)
Price	Set by a valuation cap on the SAFE	Set by the seed valuation, or by a cap again if the seed is raised on SAFEs
How datasets count it	Pre-seed in Gunderson Dettmer’s reports; inside “seed” in Crunchbase’s totals	Seed

The columns come from different sources, samples and dates, so the table compares tendencies. It does not follow one company through both rounds, which the worked example below does.

How much does a pre-seed or seed round raise?

A seed round is larger than a pre-seed round, and the middle of the distribution says less than the top: Crunchbase’s median US seed round was about $3 million in 2025, yet in its 5 October 2026 review $2.6 billion of the $13 billion of global seed funding in Q3 2026, or 20% of the money, went to seed rounds of $100 million and over.

At the small end, Y Combinator’s standard deal invests $500,000 on two SAFEs: $125,000 on a post-money SAFE for 7% of the company, and $375,000 on an uncapped SAFE with a most favoured nation clause, which converts on the terms of the lowest-cap SAFE the company issues between about the start of the batch and the priced round. The page works an example in which a $15 million post-money cap on the founder’s later SAFEs turns the $375,000 into 2.5% ($375,000 divided by $15,000,000). In that deal part of the price is fixed on signing, and the rest is set by the SAFEs the founder signs before the priced round.

Crunchbase’s spread around that median is wide: in a 26 May 2026 analysis, the lowest quartile of US seed rounds in 2025 was $1 million and the upper quartile around $5.6 million. Unmatched Capital read no neutral median for pre-seed rounds, and the figures above are not comparable one for one: Y Combinator’s is a single investor’s cheque and Crunchbase’s a distribution of a category that includes pre-seed.

Which instrument does each round use?

At pre-seed the instrument is nearly always a SAFE, at least in the US deals law firms report: Fenwick’s analysis of Q3 2025, published on 16 December 2025, says SAFEs comprised more than 92% of early-stage financings, and that convertible notes remain more common after priced equity rounds, with nearly a 30% share of such financings. The post names no sample or definition, so the figure cannot be read as a market census. An Orrick partner puts the pre-seed share at somewhere between 80 and 90 percent in an April 2026 video, without naming a dataset, which makes it a practitioner estimate that agrees with Fenwick’s order of magnitude.

A SAFE, or simple agreement for future equity, is a short contract in which an investor funds a startup now for shares later. Y Combinator’s page says a SAFE is not a debt or a loan and has no interest and no maturity date, and business.gov.au says it was developed as a simple replacement for convertible notes. If the company is sold or dissolved before the SAFE converts, Y Combinator’s page says the SAFE is junior to outstanding debt and ranks like standard non-participating preferred stock: on a sale the holder takes the greater of its money back or what it would receive on conversion, and on a dissolution it is owed its purchase amount. Y Combinator, whose post-money form has been its standard since 2018, defines the valuation cap as the highest valuation at which a SAFE converts, and says the ownership sold equals the investment divided by the cap. The SAFE debuted in 2013, according to Gunderson Dettmer’s summary of a 2018 survey of US and Canadian startup lawyers by John Coyle and Joe Green. The Orrick partners in that video give the reasons for its spread: it takes in money quickly and cheaply, it defers negotiation of shareholder rights to a later round, and, as one of them puts it, founders for the most part do not need startup lawyers.

The instrument at seed is less well documented. Priced equity, SAFEs and notes all appear, and none of the neutral sources Unmatched Capital read gives the share of seed rounds that are priced or the split by round size. A founder cannot assume that seed means a priced round, and an investor cannot assume that a SAFE means pre-seed. What can be said is that the SAFE’s logic runs the same way at either stage: a cap or a discount sets the price now, and a later round fixes the real one.

How do pre-seed and seed instruments differ outside the United States?

Outside the United States the SAFE is a local adaptation or is absent: Y Combinator’s own SAFE tool supports only US-incorporated companies, tells founders elsewhere to work with a local lawyer, and lists separate valuation-cap versions only for companies formed in Canada, the Cayman Islands and Singapore.

Place	What the evidence shows	Source
United States	Post-money SAFE is the Y Combinator standard; Fenwick puts SAFEs at more than 92% of early-stage financings (Q3 2025)	Y Combinator; Fenwick
Canada, Cayman Islands, Singapore	Y Combinator publishes cap versions for companies formed there	Y Combinator
United Kingdom	An advance subscription agreement for SEIS or EIS must name a longstop date for issuing the shares (details below)	HM Revenue & Customs; GOV.UK
Austria	SAFEs are novel, there are no standard forms, and the Supreme Court has not ruled on whether notarial form is needed	Wolf Theiss, for the IBA
Finland	Finnish corporate law does not recognise SAFEs as a category; capital loans and subscription warrants are used	Castrén & Snellman
Australia	A SAFE can be an eligible venture capital investment for venture capital limited partnerships and early stage venture capital limited partnerships if it is a convertible note that is not a debt interest	business.gov.au

The United Kingdom shows the clearest difference. The Seed Enterprise Investment Scheme gives tax relief to individuals who buy new shares, and most companies can raise at most £250,000 through it in their lifetime, according to GOV.UK guidance updated on 6 April 2026. HM Revenue & Customs will not treat an advance subscription agreement as suitable for SEIS unless the payment cannot be refunded, the agreement cannot be varied, cancelled or assigned, it bears no interest, and it has a longstop date by when the shares must be issued (HMRC manual VCM33025, updated 29 October 2026). It expects that longstop to be no more than six months from the date of the agreement and is unlikely to give advance assurance for a longer one, but a longer longstop does not by itself make the investment ineligible for SEIS relief, and advance assurance is optional. The EIS page (VCM12025) says the same for EIS.

A SAFE commonly has no maturity date, as business.gov.au describes it; the UK instrument must say when the shares will be issued. The consequence, Unmatched Capital’s inference, is that a UK pre-seed investment relying on SEIS relief has a date on it: the agreement must name a longstop, and a founder who wants HMRC’s advance assurance should expect that date to fall within six months, by when the next round must arrive or the shares be issued on the agreement’s own terms.

Elsewhere in Europe local law shapes the instrument. A Wolf Theiss article for the International Bar Association of March 2022 says SAFEs are still quite novel in Austria, that agreements transferring shares in a limited liability company must be in notarial form, and that no Supreme Court case settles whether a SAFE needs it. Castrén & Snellman wrote in December 2025 that several European countries have developed local versions, naming the UK, Sweden, Norway, France and Germany, and that each must be tailored to its own legal system. Readers outside the United States should treat the arithmetic below as the test and the documents as local.

How much dilution do pre-seed and seed cost a founder?

A post-money SAFE costs founders its cheque divided by its cap, and the seed then takes its share of what is left, as Y Combinator’s SAFE page states the rule (ownership sold equals the investment divided by the cap): in a hypothetical where a $3 million seed sells 20%, a $1 million SAFE on a $10 million cap leaves the founders with 72.0% against 80.0% if they had gone straight to seed. The table is Unmatched Capital’s own calculation, and no real company is described.

Assumptions: the founders own 100% before the SAFE; there is one SAFE round, with no discount and no option pool (a pool would lower the founders’ real figures); the seed raises $3 million, about Crunchbase’s median US seed round, and sells 20%, a round number we chose, making a $15.0 million post-money and a $12.0 million pre-money valuation; and each SAFE converts at its cap, because every cap is below $12.0 million. The caps and cheques are chosen for the example, since no neutral source read gives median caps by round size.

Path	SAFE holders at conversion	Founders after seed	SAFE holders after seed (the founders’ cost, in points)	Seed post-money at which founders keep 80.0% (seed raises $3 million)
Straight to seed	none	80.0%	none	$15.0 million
$750,000 SAFE, $10,000,000 cap	7.5%	74.0%	6.0	$22.2 million (1.5 times)
$1,000,000 SAFE, $10,000,000 cap	10.0%	72.0%	8.0	$27.0 million (1.8 times)
$1,000,000 SAFE, $8,000,000 cap	12.5%	70.0%	10.0	$35.0 million (2.3 times)

A 20% seed leaves everyone with 80% of what they had. The last column is a counterfactual about ownership only. It gives the seed post-money valuation at which the founders hold the same 80.0% as on the straight-to-seed path with a $3 million seed, and it ignores the cash the SAFE puts into the company. To run it with your own numbers, add up each SAFE’s cheque divided by its cap: that is the share the SAFEs own at conversion, if each is a post-money SAFE without a discount that converts at its cap; any other SAFE needs its own calculation, as the next paragraph shows. Divide the share founders would keep without the SAFEs (80% here) by one minus the SAFE share, and subtract the result from one: that is the most the seed can sell. Divide the seed cheque by that share for the post-money valuation. For the $1 million SAFE on a $10 million cap, 0.80 divided by 0.90 is 0.889, so the seed can sell 11.1%, and $3 million divided by 0.111 is $27.0 million. The price climbs as the SAFEs sell more: SAFEs totalling 15% would need 3.4 times the $15.0 million, and once they own as much as the seed sells, no valuation restores the founders’ share. If the seed sold 15% of a $20 million post-money, the same $1 million SAFE would cost 8.5 points and need $54.0 million, 2.7 times.

A cap above the seed’s pre-money valuation gives the SAFE no protection. With a $15 million cap and a $12.0 million pre-money seed, a $1 million SAFE converts at the seed price and owns $1 million divided by $12 million, or 8.3% before the seed and 6.7% after.

Can a founder skip pre-seed and raise a seed round directly?

A founder can skip pre-seed, and in the worked example doing so saves 8.0 points of ownership on a $1 million SAFE at a $10 million cap if the seed sells 20% either way, but ownership is one side of the trade. The pre-seed money wins those points back in ownership only if it lifts the seed post-money valuation by the factor in the last column, about 1.8 times for that SAFE, with the seed cheque held at $3 million. No source Unmatched Capital read gives the share of companies that skip pre-seed, or shows whether companies that raised on SAFEs later priced their seed higher.

Unmatched Capital’s reading is that a pre-seed SAFE buys time and a better position for the seed round, paid for in ownership at a rate the cap fixes, and that a founder can price that purchase before signing. The strongest counterargument is that pricing the seed better is the purpose of pre-seed money: it funds the product, the revenue or the hires that make the seed possible at a higher valuation, and a founder who skips it may raise a smaller seed, a lower-priced one or none. That holds. Treat the break-even as a reference point. A founder who would otherwise raise later, or not at all, can rationally give up more points for better odds of raising. The table turns it into a test, because the post-money valuation at which the points come back is a number a founder can compare with what the funded milestone is expected to do to the next valuation. Where the alternative to the SAFE is no company, the test does not apply, and the comparison between caps is the only one left.

The evidence on progress argues for realism about that milestone. Crunchbase found, in its 29 October 2026 analysis, that only 24% of companies that raised a $1 million-plus seed in 2023 had moved past seed, and just 16% of the 2024 cohort, down from 55% or higher through 2020. The younger cohort has had less time to progress, so the 16% will rise, but the series shows that a seed does not by itself carry a company to the next round. A SAFE’s points are paid up front, and the uplift they are meant to buy arrives later, if at all.

A second objection is the clock. In the United Kingdom the advance subscription agreement behind a SEIS investment must carry a longstop date, which HMRC expects within six months and beyond which it is unlikely to give advance assurance, so a founder who wants that assurance has about six months for the next round to arrive before the shares must be issued on the agreement’s own terms. A SAFE, which commonly has no maturity date, can sit open until a priced round arrives. Where there is no standard form, as the Wolf Theiss article says of Austria, drafting the instrument is a cost that a standard SAFE largely avoids, on the Orrick partner’s account.

If the money runs out before the next round, a founder may need a bridge round. After seed, the next step is covered in Series A versus Series B.

What this means if you are choosing between them

A founder weighing a pre-seed SAFE against waiting for a priced seed should work out two numbers before signing: the points sold, which is the investment divided by the cap added up across every SAFE, and the seed post-money valuation at which the founders keep the same share, using the formula under the table with the founder’s own figures. A post-money SAFE makes the first number easy to calculate, as Y Combinator says, and its example of $500,000 at a $5.5 million cap (about 9%) plus $500,000 at an $8.3 million cap (about 6%) sells about 15%. A founder outside the United States should first ask a local lawyer which instrument the jurisdiction supports, and a UK founder using SEIS should ask an adviser how an advance subscription agreement and its longstop date fit the plan.

The points sold can then be set against what the next round is likely to sell. In the hypothetical it is 20%, which is an assumption and not a benchmark; SAFEs that have sold close to that leave little room, and if more investors are lining up, the company has raised a seed round on SAFEs and should ask whether to price it.

For an angel writing into a post-money SAFE, the stake at conversion is the cheque divided by the cap, and the priced round that follows dilutes it. A $25,000 cheque on a $10 million cap buys 0.25%, which falls to 0.20% after a seed selling 20% (an Unmatched Capital calculation, with no option pool). A cap above the seed’s pre-money valuation gives the angel no step-up, because the SAFE then converts at the seed price.

The same arithmetic gives an angel a paper return. At the worked example’s seed post-money valuation of $15.0 million, the $1 million SAFE on a $10 million cap is 8.0% of the company after the seed, which is $1.2 million, or 1.2 times the money, against 1.0 times for the same cheque on a $15 million cap, which converts at the seed price. These are paper values at the seed price, before later dilution and failure.

For the wider picture of how funds and returns work, see venture capital. None of this is investment, tax or legal advice.