The most dangerous legal document a founder will ever sign is about five pages long, has a friendly name, and closes in an afternoon. It's the SAFE. I've watched founders negotiate a priced round for six weeks — counsel on both sides, redlines flying — and then sign a $750,000 SAFE between meetings, over email, with a lawyer who was cc'd but never actually asked to read it. Because it's "just a SAFE." Just a promise about your cap table that detonates eighteen months later. So let's go through it the way I'd go through it with a founder at a coffee shop: what it is, where it came from, and the math that decides how much of your company you still own when it converts.

Where the SAFE Came From

Y Combinator published the first SAFE in 2013 — the name stands for Simple Agreement for Future Equity — and the thing it replaced was the convertible note. Understanding why notes were annoying enough to kill is the fastest way to understand what a SAFE is. A convertible note is debt: the investor loans you money, the loan accrues interest (typically 2% to 8%), and instead of you paying it back in cash, it converts into stock at your next priced round, usually at a discount to that round's price, often with a valuation cap. Fine in theory. In practice, notes have maturity dates — usually 18 to 24 months — and early-stage companies routinely hit maturity with no priced round in sight. Then the founder is technically in default on a loan to their own investors, negotiating extensions from a position of total weakness, sometimes with the noteholders legally entitled to demand repayment from a company that has no money.

The SAFE deleted the debt. No interest. No maturity date. No repayment obligation. It's simply a contract that says: we give you money now, and in exchange you promise us shares later, when some defined event happens, at economics we fix today through the cap and the discount. YC open-sourced the forms, the whole industry adopted them, and by the late 2010s the convertible note was basically extinct at pre-seed and seed. Good riddance. But — and this is the point founders miss — removing the debt features also removed the guardrails. A note's maturity date forced a conversation. A SAFE can sit quietly on your books for years, compounding in share count, while you forget it's there.

The Valuation Cap: The Whole Game

The valuation cap is the maximum company value at which the SAFE converts into shares. If your next round prices below the cap, the SAFE converts at the round price. If the round prices above the cap, the SAFE converts as if the company were worth only the cap. The cap is the reward the SAFE investor gets for betting on you early, and it's also the number that determines how much of your company you sold.

Do the arithmetic with me, because this is where founders get hurt. You raise on a SAFE with an $8M cap. Eighteen months later you raise a priced seed round at a $12M pre-money valuation. The new investors are paying $12M divided by the share count for their stock — call it $1.20 a share if there are 10M shares. Your SAFE investor, though, converts at $8M over 10M shares, or $0.80 a share. That's $12M divided by $8M — a 1.5x discount on the share price. For every $100,000 they put in, they get the shares a priced-round investor would pay $150,000 for. They took the early risk, they got the discount, that's the deal working as designed.

Now the founder-side version. Raise $1.5M total across SAFEs at that $8M cap, and those SAFEs convert into $1.5M divided by $0.80 — 1,875,000 shares — on a cap table that had 10M. That's roughly 15.8% of the post-conversion company, handed out before the new round's dilution even starts. Then the seed investor wants 20%, and wants a 10% option pool created pre-money — out of your hide, per the pool shuffle we cover in the term sheet guide. Add it up and the founders who owned 100% on the day they signed their first SAFE own somewhere around 55% to 60% after one seed round. Nobody took anything from them. They signed it away, $250,000 at a time, in documents they thought were informal.

Caps Versus Discounts (and the Post-2020 Reality)

The discount is the other conversion lever: instead of (or in addition to) a cap, the SAFE converts at a percentage off the round price — classically 20%, so a $1.00 share price converts the SAFE at $0.80. On the original 2013 forms, caps and discounts were both common, and some SAFEs had both, with the investor getting whichever was better. Here's what's actually happened in the market since: the plain discount has nearly vanished. Post-2020, the cap-only SAFE is the overwhelming standard, because a cap is a floor on the investor's effective ownership and a discount alone is not. A 20% discount with no cap means that if your seed round prices at $40M, the SAFE holder converts at a $32M effective valuation — great for them in dollars, but they own a sliver, and early investors want the sliver protected.

So when an investor today asks for "cap and discount," understand what they're asking for: the better of two conversion prices, at their election. $8M cap with a 20% discount means that if the round prices under $10M, the discount wins for them ($10M minus 20% equals $8M); above $10M the cap wins. They are protected in both directions. That's not crazy as risk pricing, but it's a stronger claim than the cap alone, and you should know you're granting it. If you must give both, push the discount to 15% or raise the cap — one lever should move when the other appears.

MFN: The Clause That Follows You Around

MFN — most favored nation — says: if you later sign a SAFE with better terms for another investor, this investor gets those better terms too. It shows up when an investor accepts weaker economics now and wants insurance against you upgrading the next person. Sounds harmless. It isn't, for two reasons.

First, it kills your flexibility to price later SAFEs properly. Say you sign a $250,000 MFN SAFE in March with no cap because you're pre-everything and grateful. In June, with more traction, you sign $500,000 at a $10M cap — a fair price by then. Under MFN, the March investor's uncapped SAFE now converts at the $10M cap too. Your improving traction just retroactively repriced old paper, and the dilution math you ran in June is wrong. Second, MFN makes every future negotiation heavier, because each new term doesn't cost you once — it costs you across every MFN SAFE outstanding. If an early investor wants MFN, the trade I'd accept is MFN with a defined sunset — it expires when you sign a capped SAFE, or in 12 months, whichever comes first. Open-ended MFN on your earliest, most desperate check is a landmine with a very long fuse.

Pro-Rata Rights and the Side Letter Problem

A pro-rata right lets the SAFE investor invest in your next round to maintain their ownership percentage. On the YC forms it's not in the SAFE itself — it lives in a separate side letter — and that's a design decision worth understanding. The SAFE converts into shadow preferred or standard preferred depending on the form, and the heavy rights (board, vetoes, information) wait for the priced round. Pro-rata is the one future right investors routinely ask for early, because it protects their ability to double down on winners.

Should you give it? Mostly yes, to real investors who add value — pro-rata for a $50,000 angel costs you nothing and buys loyalty. But watch the aggregate. Every pro-rata letter you sign is a claim on your next round's allocation. If the round is $2M and your SAFE holders' pro-rata claims total $1.2M, your new lead — the fund writing the big check, the one whose name gets you hired engineers — is fighting for $800K of room. I've seen a seed round nearly collapse because the founder had handed out pro-rata to eleven angels and the lead's model required a bigger position than the leftover space allowed. Track pro-rata grants in a spreadsheet the day you sign them. The future you, mid-round, will be grateful.

The 2018 Revision: Post-Money SAFEs and Who Gets Diluted

In 2018, YC rewrote the SAFE, and the rewrite changed the most important number on the document: the cap became a post-money cap. Under the old pre-money SAFE, the cap fixed the company's value before the new round's money arrived, which meant SAFE holders diluted each other and everyone diluted together when the round priced — and nobody, including the founder, could say exactly what percentage a SAFE would convert into until the round closed. Under the post-money SAFE, the math is brutally simple: invested amount divided by the post-money cap equals the ownership percentage, locked at signing. $500,000 on an $8M post-money cap is exactly 6.25% of the company. Immediately knowable. Which is genuinely better for founders in one way — no more surprise dilution math — and quietly worse in another, and the "another" is the part to understand.

Here's the trick. Under the post-money form, the SAFE's percentage is carved out before the new money and before the option pool increase, and SAFEs no longer dilute each other. Who absorbs all of the dilution from the new round and the pool? The founders. The SAFE holders' percentage is locked; the new investor's percentage is locked by the term sheet; the pool is sized by negotiation; and common — you — is the residual that takes the entire hit. The 2018 revision made SAFE dilution transparent and, in the same stroke, made it entirely yours. That's not a criticism of YC, exactly — transparency is real value — but it means the day you sign a post-money SAFE is the day you sold that percentage, not the day it converts. Treat signing as selling and your behavior changes: you raise less, at higher caps, less often.

And stacked SAFEs multiply. Three post-money SAFEs at different caps — $250K at $5M (5%), $500K at $8M (6.25%), $750K at $12M (6.25%) — are 17.5% of the company sold across three afternoons. Each felt small. The chart below shows what this does to founder ownership across a typical sequence: founding, a SAFE round, a priced seed, and a Series A. Watch the green shrink.

That fourth bar is the one to stare at. Founders under 50% after Series A is common, and it's survivable — but founders under 50% after seed is how you end up as an employee at your own company by Series B. Every percentage point you give a SAFE at a low cap is a point that never comes back, because the next round dilutes what's left, not what was.

Conversion Mechanics: When the Paper Turns Into Shares

A SAFE sits as a contract until a trigger event converts it. The main trigger is an equity financing — a priced round of preferred stock above a threshold, conventionally $1M or more of new money. Conversion is automatic: no consent needed, no negotiation, the SAFE holder gets shares of the new preferred class (or a shadow series with identical economics) at the cap-or-discount price, whichever favors them. Your lawyers compute the share counts, the cap table updates, and the SAFE is done. This is the moment all the math above stops being theoretical.

Two other triggers matter because they're the downside cases. A liquidity event — the company sells before any priced round — gives the SAFE holder a choice: take back the greater of their invested amount or their as-converted share of the proceeds based on the cap. On a good sale they convert and take their percentage; on a fire sale they take their money back ahead of common. A dissolution — the company dies — works similarly: the SAFE holder gets their purchase amount back before common sees anything, to the extent assets exist. Note what this means. In failure, SAFE holders stand in line ahead of founders and employees. The instrument that "isn't debt" still has a liquidation preference of 1x built into its payout structure. Friendly name, creditor-ish behavior.

Why "Not Debt" Cuts Both Ways

Founders love repeating that a SAFE isn't debt. No interest, no maturity, no default, nothing on the balance sheet as a liability — all true, and all good. But the comparison usually stops there, and it shouldn't. A convertible note's maturity date, for all its terror, forced a reckoning: extend, convert, or repay, on a date certain. A SAFE never forces anything. It can ride your books through three years and two more SAFE stacks, and because nothing is ever due, nobody ever has the conversation about how much of the company is spoken for. The absence of a deadline is exactly why founders under-track them.

Also, not being debt doesn't mean not being senior. As we just covered, the liquidity and dissolution payouts put SAFE money ahead of common. And at conversion, SAFE holders get preferred stock with the round's preference stack in many structures — meaning your earliest $100,000 check can end up with the same downside protections as the Series A lead. None of this is a reason not to use SAFEs; they're the right tool for most pre-seed raises. It's a reason to stop describing them as "basically free money with paperwork later." They're equity sales with the price deferred. The deferral is the convenience and the danger in the same package.

The Mistakes I See Repeatedly

Same errors, different logos. Number one: forgetting that the cap table is fully diluted. Founders model their SAFE conversion against issued-and-outstanding shares — founder stock only — and forget the option pool, the promised-but-unissued advisor grants, and the other SAFEs. The cap on the form divides against the fully diluted count, which is always bigger than the number in the founder's head, which means the SAFE's percentage is always real and the founder's residual is always smaller than the back-of-envelope said. Model fully diluted or don't model at all.

Number two: the uncapped SAFE with no discount. An investor — usually a friend, an advisor, a first-believer angel — puts in $25,000 or $50,000 on a SAFE with no cap and no discount, "to keep it simple." What's that investor buying? Conversion at whatever the next round prices, with zero reward for being earliest. If your seed prices at $2M post — unlikely, but say it — they did fine. If it prices at $20M, they own a rounding error and resent it, and you have a bitter early supporter at exactly the moment you need references. Worse, if your seed never prices and the company sells small, they take their money back ahead of you. Uncapped-and-undiscounted is the worst of both worlds for everyone. Give early believers a real cap, even a low one — $2M to $4M at the true friends-and-family stage is honest — or give them a discount. "Simple" that misprices the relationship isn't simple.

Number three, and it's the meta-mistake: signing SAFEs without a running conversion model. Every SAFE you sign should update one spreadsheet — amount, cap, discount, post-money percentage, pro-rata granted, MFN status. If you can't state, right now, the total percentage of your company promised to SAFE holders, you are exactly the founder who'll discover the answer in a term sheet negotiation, out loud, in front of a fund that just recalculated your leverage.

SAFE Versus Convertible Note: The Honest Comparison

You'll still meet investors — usually East Coast, usually older funds, sometimes international — who prefer notes. Know the trade. A note pays them interest (2% to 8%, accruing into the conversion amount, so a $100,000 note at 5% for two years converts $110,250 worth of stock), has a maturity date that creates a forcing event, and sits on your balance sheet as debt, which matters if you ever want a bank line or a venture debt facility — lenders look at outstanding notes and see claims ahead of theirs. A SAFE has none of that. Cheaper to paper, faster to close, nothing accruing, nothing due.

For the founder, the SAFE wins on almost every axis: no interest accruing against you, no maturity cliff, no default scenario. The note's only genuine founder advantage is perverse — the maturity date forces both sides to confront the cap table on a schedule, and some founders need the discipline forced on them. If an investor insists on a note, it's not a red flag by itself, but price the difference: the interest and the maturity are real costs to you, so the cap should be higher or the check bigger than the equivalent SAFE. And if they insist on a note because of the maturity leverage — because they want the option to squeeze you at month 20 — that's not a financing preference, that's a control strategy, and now you know who you're dealing with.

What I'd Tell You Before You Sign One

Five rules, coffee-shop version. One: every SAFE gets a cap, and the cap should be your honest guess at the next round's price, not a number chosen to make the investor feel warm — a cap set at half the real value of the company is a 50%-off sale of your equity. Two: add up the stack before each signature; post-money caps make this trivial arithmetic, so there's no excuse. Three: treat pro-rata and MFN as real concessions with real future costs, grant them deliberately, and write them down. Four: use the actual YC forms — the current ones, from YC's site — and be suspicious of any "custom SAFE" an investor's counsel produces; the customizations are never in your favor and the whole point of the form is that nobody has to negotiate. Five: the day you close the SAFE, email your lawyer and your cap table; the instruments only stay simple if your records do.

The SAFE earned its name honestly — simple, fast, and genuinely founder-friendlier than the notes it buried. But "simple agreement" describes the paperwork, not the consequences. The consequences are percentages of your company, sold years before you feel the sale. Sign them like that's what they are. When you're ready for the priced round these things convert into, the term sheet guide picks up exactly where this one leaves off — and the due diligence checklist covers the records the new investor will demand about every SAFE you ever signed.