SAFEs Versus Priced Rounds: What the Data Shows

The argument over whether a first round should be done on a SAFE or a convertible note is, on the numbers, finished. In the first quarter of 2026, convertible notes made up a record low of 7 percent of pre-seed rounds and 8 percent of pre-seed dollars among US-based startups on Carta.¹ Everything else was a SAFE. A law firm summary of the same report put the SAFE share at 93 percent of rounds.²

The more useful questions are how long to keep raising on SAFEs before pricing a round, and what the version almost everyone now signs does to the founder's share along the way. The first has decent data behind it; the second is arithmetic, and it is where most of the surprises sit.

How Far the SAFE Has Taken Over

Carta's quarterly pre-seed reports are the best public series on this, with two caveats. They cover US-based companies using Carta, so they describe the American market rather than a European one. And some figures are a share of deals while others are a share of dollars, which can diverge. In the first quarter of 2025, SAFEs were a record 90 percent of pre-seed rounds by count, but only 82 percent of pre-seed capital, which Carta read as a quarter with some large convertible note deals.³ By the first quarter of 2026 the gap had closed, at 7 percent of rounds and 8 percent of dollars for notes.¹

The version of the SAFE has shifted too. In the second quarter of 2024, 84 percent of SAFEs on Carta were post-money.⁴ By the third quarter it was 87 percent, against 43 percent at the start of the decade, and 89 percent of all pre-priced investments that quarter were SAFEs.⁵ Crunchbase News, writing in 2025 from Carta data, reported the same pattern: nearly 90 percent of pre-priced rounds in the third quarter of 2024 on SAFEs, and more than 85 percent of SAFEs post-money.⁶

One more caution, since it catches people comparing reports. Carta's 2026 reports count far more instruments than its 2025 quarterly reports did. The second quarter of 2026 report compares 11,500 instruments with 14,825 in the second quarter of 2025,⁷ while the original second quarter of 2025 report described that quarter as a little over 5,000.⁸ The coverage has clearly been redefined. Shares within a single report are safe to use; totals across old and new reports are not.

What the Post-Money Change Did

Y Combinator introduced the post-money SAFE in 2018, and its own documentation is clear about the reason: the ability to calculate immediately and precisely how much ownership of the company has been sold.⁹ The user guide puts the advantage in one line, that "the amount of ownership sold is immediately transparent and calculable for both the founder and the investor."¹⁰

The mechanism is that the valuation cap is measured after all the SAFE money. A 500,000 dollar SAFE at a 5 million dollar post-money cap is a claim on 10 percent of the company as it stands just before the priced round, regardless of how many other SAFEs are sold later. The guide states the consequence directly: "while the safes are not diluted by each other, the safes will be diluted by the new money raised in the Equity Financing."¹⁰

If SAFE holders do not dilute each other, the existing shareholders absorb each new SAFE. TechCrunch's report on the change in 2018 described it as each incremental dollar raised on post-money SAFEs diluting just the current stockholders, which is often the founders and early employees.¹¹ A Pillar Legal analysis from 2023 is blunter: common holders, typically only the founders, bear the entirety of dilution from other SAFE or convertible note rounds until an equity financing takes place, whereas under the older pre-money SAFE both common stockholders and SAFE holders were diluted together.¹²

In some ways this is a fair trade. The old pre-money SAFE shared dilution, but nobody could say how much of the company had been sold until the priced round arrived. The post-money version tells the founder precisely what each SAFE costs, one SAFE at a time; the guide even notes that raising more than the post-money cap "would result in negative ownership for founders."¹⁰

Stacking, Worked Through

What follows is our own illustrative calculation with invented but plausible numbers, not drawn from any dataset.

Two founders hold 9 million shares between them and there is a 1 million share option pool, so the founders own 90 percent. Over eighteen months they raise three SAFEs, all post-money, cap only, no discount: 500,000 dollars at a 5 million dollar cap, then 750,000 dollars at 8 million, then 1 million dollars at 12 million.

Each looks modest on its own terms, at 10 percent, 9.4 percent and 8.3 percent. Together they are 27.7 percent of the company just before the priced round, and all of it comes out of the founders and the pool. The founders are at 65.1 percent before a priced round has happened, having raised 2.25 million dollars.

Then the seed round is priced: 4 million dollars at a 16 million dollar pre-money valuation, so new investors take 20 percent, with the usual requirement that the option pool be topped up to 10 percent of the post-money company and that the top-up sit in the pre-money. At that price every SAFE converts at its cap. After the round, the founders hold 49.1 percent. The three SAFE holders hold 20.9 percent between them, having been diluted by the new money and the pool top-up, as the guide says they will be.

With no SAFEs, the same seed round would leave the founders at 70 percent. Each SAFE takes them lower, to 62.3, then 55.2, then 49.1. None of this was hidden, since every number was calculable the day each SAFE was signed; the surprise founders describe is that nobody added them up, and that the pool top-up in the priced round lands on the founders a second time.

What the Terms Look Like Now

For 2025, Carta reported median caps on post-money SAFEs of about 10 million dollars for rounds of 250,000 dollars to 1 million, and about 15 million dollars for rounds of 1 million to 2.5 million.¹³ Smaller rounds sit lower: in the second quarter of 2025 the median cap for SAFE rounds under 250,000 dollars was 7.5 million.⁸ At the other end, Carta's second quarter of 2026 report says caps on SAFEs larger than 2.5 million dollars can reach 100 million at the 90th percentile.⁷ These are separate round-size bands, and a median from one band says nothing about another.

The standard shape, in Carta's words, is "the post-money SAFE with a valuation cap but no discount."¹³ Carta's own figure for the first half of 2024 was that over 90 percent of SAFEs had a valuation cap, and that when a discount appears it is typically 20 percent.⁴ Crunchbase News, quoting Carta data for January to September 2024, broke that down as 62 percent cap only, 29 percent cap and discount and 9 percent discount only, and separately put uncapped, no-discount SAFEs at about 1 percent.⁶ Those figures come from one write-up and do not quite reconcile, and we could not open the underlying Carta table, so treat the split as reported rather than verified; the headline, that caps are near-universal and discounts a minority, holds across every source we checked. Y Combinator also publishes an uncapped MFN version, with no cap and no discount, whose holder can take the terms of a better SAFE issued later.⁹ On the data it is rare.

When a SAFE Is the Right Call

The case for the SAFE is strong. It is a short standard document, it can close investor by investor rather than all at once, and it avoids setting a price at a stage when nobody can defend one. For a small first round from a handful of angels, a priced round's legal work, board terms and investor rights are a lot of apparatus for the money involved.

Carta's data suggests the market agrees. In its second quarter of 2025 report, a majority of rounds of 3 million to 4 million dollars in the first half of that year were raised on SAFEs or convertible notes rather than priced equity; previously, most companies had switched to priced equity at around 3 million dollars, and Carta said that threshold seemed to have risen to 4 million.⁸ Crunchbase News, writing a few months earlier, put the usual switch to priced equity at 3 million dollars or more, which matches the older threshold.⁶

When a Priced Round Protects the Founder

The same features cut the other way once amounts grow or SAFEs pile up. A SAFE defers the questions a priced round forces, such as total dilution, pool size and investor rights, and deferring them does not make them smaller. The illustration above is what they look like arriving together.

A priced round gives the founder a fixed cap table on the day it closes, with every holder diluted by later rounds on the same basis. The extra legal cost starts to pay for itself once SAFEs are approaching a quarter of the company, once caps have begun rising round by round (so early holders convert far cheaper than later ones), or once the next raise is large enough that investors will insist on a price anyway. The 3 to 4 million dollar crossover in Carta's data is consistent with that, though it describes what founders do, not what is optimal for any one company.

The data has a gap here. We looked for Carta figures on the median time from a SAFE to its conversion, or the share of SAFEs that ever convert, for 2024 to 2026, and did not find any published. There is no benchmark for how long a stack of SAFEs typically sits unconverted, and anyone quoting one should say where it comes from.

Outside the United States

The SAFE is a US instrument written for US corporations. Y Combinator publishes versions for Canada, the Cayman Islands and Singapore,⁹ and not for European jurisdictions, which use their own forms.

In the UK, the common equivalent is the advance subscription agreement. HMRC's guidance says it will not consider one suitable for EIS unless the payment cannot be refunded, the agreement cannot be varied, cancelled or assigned, it bears no interest and it has a longstop date, which HMRC expects to be no more than six months from the date the agreement is entered into.¹⁴ That six-month limit makes it a much shorter bridge than a SAFE. In France, the BSA AIR was adapted from the SAFE into French law by the incubator The Family with SB Avocats, and is a warrant with no interest rate or maturity date that converts at the next capital increase, typically with a cap and a floor.¹⁵ In Germany, early rounds commonly use the convertible loan, and Bird & Bird notes that after a 2022 Zweibrücken court decision, notarising the loan agreement is recommended at least where conversion is mandatory.¹⁶ Spain has its own mechanics, which we leave to a separate piece.

The dilution arithmetic travels, since a cap is a cap; the legal treatment does not, and investors from different countries tend to arrive with different default instruments.

Before the Third SAFE

The SAFE won because it is cheap and fast at small amounts, and the post-money version made each one honest about its cost, though it leaves the adding up to the founder. Before signing the third one, model the priced round that will convert them, including the pool top-up, and look at the founder line at the end.

That is also the conversation to have early with investors on AngelHive, where a founder may be raising from angels in several countries at once: agree the instrument, and share the pro forma cap table showing where each SAFE lands at conversion, before anyone is asked to sign.

Sources

1. Carta, State of Pre-Seed: Q1 2026, Hamza Shad, 14 May 2026. https://carta.com/data/state-of-pre-seed-q1-2026/

2. Flux, What Carta's Q1 2026 Pre-Seed Report Tells Founders, and Two Things to Stop Doing, Ryan Howell, 28 May 2026. https://www.flux.law/blog/carta-pre-seed-q1-2026-what-founders-should-know

3. Carta, State of Pre-Seed: Q1 2025, Hamza Shad, 20 May 2025. https://carta.com/data/state-of-pre-seed-q1-2025/

4. Carta, State of Pre-Seed: Q2 2024, Hamza Shad and Kevin Dowd, 21 August 2024. https://carta.com/data/state-of-pre-seed-q2-2024/

5. Carta, State of Pre-Seed: Q3 2024, Ashley Neville and Kevin Dowd, 4 November 2024. https://carta.com/data/state-of-pre-seed-q3-2024/

6. Crunchbase News, The Startup Aspirin: SAFEs Ease Fundraising Pain, But Read The Fine Print First, Ling Kong, 29 May 2025, citing Carta data. https://news.crunchbase.com/venture/startup-funding-safes-kong-michelman/

7. Carta, State of Pre-Seed: Q2 2026, Hamza Shad, 13 August 2026. https://carta.com/data/state-of-pre-seed-q2-2026/

8. Carta, State of Pre-Seed: Q2 2025, Hamza Shad, 21 August 2025. https://carta.com/data/state-of-pre-seed-q2-2025/

9. Y Combinator, The SAFE: documents and versions, including the 2018 post-money safe. https://www.ycombinator.com/documents

10. Y Combinator, Post-Money Safe User Guide, version 1.2, February 2023. https://bookface-static.ycombinator.com/assets/ycdc/SAFE%20User%20Guide-a47c6588327d73aa2799e61ed7c2cae9f1a0ee9acfa9c43b62039dc06e715832.pdf

11. TechCrunch, Y Combinator is changing up the way it invests, Kate Clark, 28 September 2018. https://techcrunch.com/2018/09/28/y-combinator-is-changing-up-the-way-it-invests/

12. Pillar Legal, Y Combinator's Post-Money SAFE: Risks for Founders, Greg Pilarowski and Alexandra Ashbrook, 7 November 2023. https://www.pillarlegalpc.com/wp-content/uploads/2024/07/Y-Combinators-Post-Money-SAFE-Risks-for-Founders-Final-2023-11-7-clean-V2.pdf

13. Carta, State of Pre-Seed: 2025 in Review, Hamza Shad, 19 February 2026. https://carta.com/data/state-of-pre-seed-2025/

14. HMRC, Venture Capital Schemes Manual, VCM12025, EIS: advance subscription agreements, updated 30 July 2026. https://www.gov.uk/hmrc-internal-manuals/venture-capital-schemes-manual/vcm12025

15. The Conversation, Un exemple de créativité dans le financement des start-up : la formule des BSA-AIR, Didier Bernard, Grenoble IAE, 13 March 2018. https://theconversation.com/un-exemple-de-creativite-dans-le-financement-des-start-up-la-formule-des-bsa-air-92640

16. Bird & Bird, Form requirements for convertible loans in Germany, 2023. https://www.twobirds.com/en/insights/2023/germany/formerfordernisse-bei-wandeldarlehen