Three SAFEs, three caps, eighteen months apart. Each one felt like a small, sensible decision. Together they quietly handed away 40% of the company before a single priced round closed. If you're stacking SAFEs one raise at a time without modeling them together, you're probably closer to that number than you think.
Why one SAFE at a time never feels dangerous
A single SAFE is easy to reason about. You raise $1M on a $5M post-money valuation cap, so that investor owns 20% once it converts. The math is clean, the percentage is fixed at signing, and it's easy to move on to building the product.
The danger isn't any one SAFE. It's that founders keep evaluating each new SAFE against the current cap table instead of against the cumulative stack. A $1M SAFE at a $5M cap looks like 20%. A second $1M SAFE at a $6M cap looks like 16.7%. Neither number, on its own, sounds alarming. Add them together with a bridge note on top and you can cross 40% before your Series A investors have even seen a deck.
The math nobody runs until it's forced
Post-money SAFEs fix each investor's ownership percentage at the moment they sign, not at conversion. That's the feature that makes them fast to close. It's also what makes stacking so easy to misjudge, because nothing forces you to look at the combined effect until a lawyer builds the actual cap table ahead of a priced round.
Run the numbers on a realistic early-stage stack: a $1M SAFE at a $5M cap (20%), a $1M SAFE at a $6M cap roughly fourteen months later (16.7%), and a $500K bridge at a $4M cap when the runway got tight (12.5%). Layered naively, that's north of 45% gone before a single primary dollar from a Series A has been priced. Add a typical 20-25% Series A round on top and the founding team can be looking at owning less than a third of the company they started.
This is the exact blind spot that shows up in founder cap-table teardown after teardown: the SAFE calculator answers what does this one SAFE cost me, not what does my whole SAFE stack cost me. Kruze Consulting's breakdown of SAFE dilution walks through the same compounding pattern, and Y Combinator's own founder library is explicit that SAFEs were designed for speed of closing, not for making cumulative dilution visible.
What a 40%+ pre-Series-A stack actually does to the round
Series A investors have a target ownership number in mind, usually 20-25%, and they price the round to hit it regardless of what already happened on the cap table. If the founders are already down to 55-60% before that round prices, the new money doesn't split the difference. It comes out of whoever is left with equity to give, which in practice means the founders and the option pool.
The knock-on effect is worse than the raw percentage. A founder holding 30% post-Series-A, instead of the 45-50% a clean stack would have preserved, has less room left for the option pool refresh the next round will demand, less negotiating leverage on liquidation preference stacking, and a materially different outcome in every future dilution event, because every subsequent round dilutes the smaller number, not the bigger one.
The three-question test I run before signing the next SAFE
- What's my fully-diluted ownership after this SAFE, assuming the worst-case conversion price, not the best case? Model the cap, not just the round size.
- What's my cumulative SAFE overhang as a percentage of the company, including every SAFE still outstanding, not just this one? A single running total catches what a deal-by-deal mental model misses.
- Will my Series A investor still see enough founder ownership left to feel like they're backing a motivated team? If the honest answer is barely, that's the moment to renegotiate terms or delay the round, not after the term sheet arrives.
The 30-day move
Before you sign the next SAFE, build one spreadsheet: every outstanding SAFE, its cap, its discount, and its worst-case conversion percentage, summed into a single cumulative dilution number. Update it every time a new SAFE closes, not just when your lawyer builds the real cap table for a priced round. That one habit is the difference between finding out you're at 40% from a spreadsheet you control, or from a term sheet you don't.
If the cap table already needs a hard reset before your next raise, a structured cleanup pass three to six months out catches most of this before investors do it for you. And if you're about to sign a SAFE on top of ones you already have outstanding, it's worth renegotiating the cap explicitly rather than accepting the first number offered.
Frequently asked questions
How much dilution is normal from SAFE notes before a Series A?
There's no single normal number, but a stack of two to three SAFEs commonly totals 25-40% combined ownership once every cap converts at worst case, especially when caps step down as the company matures.
Do SAFE notes convert before or after the new money in a priced round?
SAFEs convert into equity immediately before the new Series A money is priced in, which is exactly why their cumulative percentage, not each individual SAFE's percentage, determines how much of the round comes out of the founders.
What's considered a safe cumulative SAFE overhang percentage?
Many experienced fundraising advisors treat 20% as a caution line and 30%+ as a signal to slow down and model the next round carefully before signing anything else.
Should I cap total SAFE dilution before raising a priced round?
Yes. Setting an internal ceiling, for example refusing to sign a SAFE that would push cumulative overhang past a set percentage, forces the conversation with investors before it becomes a forced conversation with a term sheet.
Can I renegotiate an outstanding SAFE's cap before a priced round?
It's uncommon but not impossible, particularly with an existing investor who wants the company to raise successfully. It's a harder conversation than getting the terms right up front.
I found out my real number from a lawyer's spreadsheet three weeks before a term sheet, not from my own model. Build the model first. It's the only way stacking SAFEs stays a strategy instead of a surprise.