safes feel free because nothing happens at signing. no priced round, no new shares issued, no obvious dilution event in the cap table. you just take the money and keep going.
but safes don't remove dilution. they defer it. stack four of them and you've quietly sold a third of your company before a priced round ever sets a real number.
founders model the money coming in. almost nobody models the ownership going out. by series b, most own far less than they assumed, and a cap table has no undo button.
the four-safe trap
a safe — simple agreement for future equity — is a promise. you give me money now; i give you shares at the next priced round, on terms set today. those terms, the valuation cap and the discount, determine how much of the company you've actually sold. at signing, no shares are issued. the cap table doesn't change. that's the trap.
a realistic seed journey, all on safes, no priced round yet.
- pre-product: $500k safe at $5M post cap
- early traction: $750k safe at $8M post cap
- bridge to seed extension: $500k safe at $10M post cap
- pre-series-a: $1M safe at $15M post cap
total raised: $2.75M. feels reasonable. no real "round" yet. founders feel like they still own most of the company.
then the series a hits — say $5M at a $25M post-money. on conversion, all four safes settle. each one bites a known piece of the cap table, and the bites don't politely take turns.
add the series a itself — $5M ÷ $25M = 20% — and the option pool refresh (usually 10% pre-money, costing another ~9% off the founders). total dilution at series a runs to roughly 51%.
founders entered the round thinking they owned ~85% and would give up 20%. they exit owning ~36% combined. the gap between what they assumed and what actually happened is almost half the company.
why the model fails
three reasons, all human.
one. each safe in isolation looks small. "$500k at $5M post — that's only 10%, fine." it is fine. it's the fifth safe that breaks the model, and you don't see it coming because the previous four didn't trigger anything visible.
two. post-money safes — the yc 2018 standard — dilute only the existing shareholders, not the new safe investors. each new safe makes the cap table worse for everyone who came before, including you.
three. the cap on each safe sets the floor of the post-money it'll convert at. if the series a comes in lower than your safe caps, the safes convert at the cap, which is now disproportionate equity. founders model the conversion as if it'll happen at the series a price. it usually doesn't. it happens at the cap.
what to do instead
model every safe before you sign it. Carta has a calculator. so does Pulley. or build a spreadsheet. plug in the existing cap table, all outstanding safes, the safe you're about to sign, and three hypothetical series a scenarios. look at your post-conversion ownership. if the number scares you, negotiate the cap or take less money.
combine safes into a priced round earlier than feels comfortable. the longer you stack, the more compounded the dilution. a priced seed at $8M is often less dilutive than four safes that average to a $9M cap, because the priced round forces everyone into the same terms instead of layering them.
negotiate the option pool timing. pool refreshes traditionally happen pre-money in a priced round, which means founders absorb the entire dilution. push to make the refresh post-money, or smaller. it sounds technical. it's worth several percentage points of ownership.
the math nobody runs
most founders run cash projections monthly. almost no one runs a dilution projection. the equivalent question for ownership is: if i take this safe at this cap, and one more in six months at the next probable cap, and the series a closes 18 months from now — what do i own at series b in four years?
if the answer is below 15% for the founder team, you're on the path where founders lose voting control, get pushed out at series c, or walk away from a company that's no longer theirs in any meaningful sense.
$2.75M raised across four safes. ~51% diluted at series a. the math doesn't care what you thought you were giving up.
how zift handles this
zift doesn't model cap tables — that's Carta's job. zift handles the rest: cash, burn, runway, mrr, the briefing on monday morning. the financial visibility that lets you decide whether to take the safe at all, because the only reason most founders take dilutive money on bad terms is they don't know how close they are to running out of cash.
if you're a finance lead managing this across multiple portfolio companies or a multi-entity startup, zift handles that too.
a cap table has no undo button. a runway projection that lies to you is how you end up needing one.