YC's post-money SAFE: the dilution math the Brazilian founder needs to do before signing

SAFE (Simple Agreement for Future Equity) has become the de facto standard for early-stage fundraising, and Y Combinator's official documentation doesn't hide why: it's a short, free contract that, as a rule, has a single negotiable term (the valuation cap) and automatically converts into shares once a priced round arrives. Created by Carolynn Levy at YC in 2013, the instrument has, according to the accelerator itself, already backed more than US$15 billion raised by portfolio companies.

The problem isn't the SAFE. It's how easy it is. When the Brazilian founder signs a post-money SAFE to receive a check from an angel or a foreign fund, they're making a cap table decision that only charges its price two or three years later, at the Series A. And most sign while looking at the check amount, not the dilution math built into it.

Post-money is a promise of percentage, not of valuation

Here lies the change that YC standardized in 2018, which a lot of people still misread. In the post-money SAFE, the stake sold is fixed and calculable at the moment of signing: ownership = amount invested ÷ valuation cap.

The documentation's own examples make this explicit. On a post-money cap of US$6.7 million:

| Check | Post-money cap | Stake sold | |---|---|---| | US$500K | US$6.7M | ~7.5% | | US$800K | US$6.7M | ~12% | | US$1M | US$6.7M | ~15% |

The strategic reading is the reverse of what founders usually do. You don't start from the valuation to figure out how much you're selling. You start from how much of the company you're willing to give up and how much you want to raise, and the cap is the consequence: US$1 million for 15% implies a post-money cap of ~US$6.7 million (US$5.7 million pre-money). The math is simple, which doesn't mean it's harmless.

The difference between pre- and post-money changes who absorbs the dilution from subsequent SAFEs. In the post-money version, the percentage promised to each SAFE investor is protected against dilution caused by other SAFEs issued afterward. In other words: all the dilution from stacking SAFEs falls on the founder, it isn't shared with earlier investors. This is the opposite of the old pre-money model, where the first angels diluted alongside the founders with every new SAFE. Post-money is more transparent for the investor precisely because it's less forgiving toward the founder.

The silent stacking

The concrete risk for the BR founder isn't one SAFE. It's several. YC's documentation shows you can raise at different caps: US$500K at a US$5.5 million cap (~9%) plus US$500K at a US$8.3 million cap (~6%) add up to the same ~15%. The catch is that, in the practice of distributed fundraising (several angels coming in over different months, at rising caps), each SAFE locks in its own percentage. You need to add them all up before knowing how much you have left.

The classic case of regret: a founder raises US$300K here, US$200K there, plus a lead check of US$500K, each at a cap that seemed reasonable at the time. The Series A arrives and the sum of the SAFEs has already converted into 30%, 35% of the cap table, before the option pool and before the new check. In the post-money model, that percentage doesn't dilute among earlier investors when the next SAFE comes in: the one who bleeds is the founder.

It's worth noting the honest counterpoint: post-money isn't a trap, it's a design choice that prioritizes predictability. The investor knows exactly what they bought, and that speeds up closing (part of the instrument's value). The mistake isn't in the contract, it's in signing without keeping a live cap table spreadsheet that adds up all the open SAFEs. YC's free tool, the SAFE calculator cited in the documentation, exists precisely to run these numbers before sending the document.

MFN: the clause that decides who's protected

The clause that confuses founders the most is MFN (Most Favored Nation). An MFN SAFE has no cap and no discount: it automatically adopts the best terms of any SAFE you issue afterward. It works for the angel who wants in now without fighting over valuation, trusting that they'll inherit the best future terms.

The point that YC's FAQ makes crystal clear, and that almost no one operates well, is which date the MFN uses: it's the later of the SAFE's signing date and the receipt of funds. Direct practical consequence:

Close your lower-cap SAFEs (with the wire already received) before issuing an uncapped MFN. You sign last and control the date.

-- Y Combinator, SAFE FAQ

Translating for whoever's deciding: if you issue the MFN and only later close a SAFE with an aggressive cap, the MFN investor gets that aggressive cap. If you close and receive the wire for the cheaper SAFEs first, the MFN doesn't retroactively apply to them. The order and timing of the wires directly affect dilution. For a Brazilian founder raising from a foreign investor, where the international wire can take days and arrive out of order, this isn't a legal detail: it's money out of your percentage.

What the BR founder needs to check beforehand

There are two traps specific to the Brazilian context that the documentation flags without naming Brazil.

First, jurisdiction. The "Send a SAFE" tool and most of the forms assume a company incorporated in the US. YC offers versions for Canada, the Cayman Islands, and Singapore, each requiring local counsel. A Brazilian company without an offshore structure (the classic Delaware C-Corp or Cayman entity on top) doesn't have an off-the-shelf SAFE ready to go, and YC itself recommends a lawyer licensed in the country of formation. Signing a US SAFE with a Brazilian Ltda. (the country's standard private limited company structure) underneath is an invitation to headaches at conversion.

Second, accredited investor status. The FAQ is explicit: the SAFE investor must be accredited. If they aren't by the time of the priced round, the startup may have to rescind the SAFE and return the money. A Brazilian individual angel entering a SAFE through a US vehicle needs to meet US accreditation rules, which isn't always the case.

Two operational reminders that cut down on rework: pro rata doesn't go inside the SAFE, it goes into a standardized, optional side letter (it's negotiable, and US$500K justifies pro rata far more than US$10K does); and an investor who wants to put in more money later shouldn't amend the old SAFE, but should sign a new SAFE on the same terms.

The implication for builders

The SAFE solves for cost, speed, and simplicity, and that's why it became the standard. But it shifts onto the founder the responsibility of doing the math the instrument doesn't do for you. The practical rule: never sign a SAFE without a cap table spreadsheet that adds up all the open SAFEs and projects the conversion at the next round, including the option pool the Series A will require. If that math shows the founder below comfortable levels even before the first priced round, the problem wasn't the check, it was the order, the cap, and the MFN that nobody added up in time.