Your SAFE cap is a dilution decision before it is a valuation.

Because most capped SAFEs today are post-money. Carta reports that 91% of SAFEs on its platform in Q2 2026 used a post-money cap, and Wilson Sonsini reports 89% of capped SAFEs in its first half 2026 deals were post-money. On that structure the cap is the company value at which the SAFE converts, counted after all the SAFE money is in, so when you sign you can see the share the investor will convert into if the cap applies.

Two things change that picture later, and both are worth knowing before you sign. If the priced round values the company below the cap, the SAFE converts at the new investors' price instead, and the SAFE holder gets more shares than the cap math said (Y Combinator, Post-Money Safe User Guide, Q3, https://www.ycombinator.com/documents). And the priced round itself, with any option pool it adds, dilutes everyone after the SAFEs convert.

One more property of the post-money SAFE: each new SAFE you sign comes out of the founders' share, not out of earlier SAFE holders', because the SAFEs are not diluted by each other (same guide). Three SAFEs that each sell 5% are 15% gone before the priced round.

Work backwards from what the money has to buy. The share is not a number to choose in the abstract; it falls out of four answers:

1. The milestone. What must be true for the next round to happen: a number of paying customers, a revenue level, a product shipped.
2. The money to get there. Months to the milestone times what you spend each month, plus a buffer.
3. What you are willing to give up for it. Count the dilution still to come in the priced round, and decide how much of the company you want to hold after it.
4. What investors will actually accept. Your math sets a floor; demand decides how far above it you can go.

One decision, start to finish. An illustration, not a real company. A founder needs 750k for 15 months to reach ten paying customers, the milestone the seed investors in the conversation have said they would price. Working back from the ownership the founders want to keep after that seed, they decide the SAFEs can take 7% to 9% of the company. 750k divided by those shares gives a cap between about 8.3m and 10.7m.

Carta's median for rounds of 500k to 999k is 12.5m, so that range sits below the median and leaves room. Two angels say they would sign at 10m. The founder sets the cap at 10m, which sells 7.5%, inside the range, and does not stretch toward the median, because the money is enough and the investors are in. Had five investors wanted in, a higher cap would have been worth testing; had they hesitated at 10m, the next step would be to ask why.

Two free public datasets give a reference point. Neither is the market; each covers the deals its publisher can see.

Carta, Q2 2026, U.S. post-money SAFEs on the Carta platform, median cap by round size:

Source: Carta, State of Pre-Seed: Q2 2026, published 2026-08-13, data as of 2026-08-10, https://carta.com/data/state-of-pre-seed-q2-2026-full-report/. Carta reports that caps rose year over year in every band.

Wilson Sonsini, first half of 2026, SAFEs in deals the U.S. law firm worked on: the median cap was 15m, down from 20m in 2025; 85% of SAFEs carried a cap, down from 93%; the median discount held at 20%. Source: Wilson Sonsini, The Entrepreneurs Report Q2 2026, published 2026-09-03, https://www.wsgr.com/en/insights/the-entrepreneurs-report-q2-2026.html

The two seem to disagree: Carta's caps rose, Wilson Sonsini's median fell. They count different deals over different periods, so neither report settles why. One possible explanation is mix: Wilson Sonsini's median SAFE fell to 240k in Q2 2026, and the firm ties its lower caps to those smaller financings, so an all-deals median can fall while each size band rises. Either way, compare your cap with rounds your own size.

A median describes what other founders signed, not what you can get.

• Population. Carta counts companies that use Carta; Wilson Sonsini counts its own clients. Neither is a random sample of startups: Carta's figures are U.S. data, and Wilson Sonsini is a U.S. law firm. If your company would not appear in either, the reference may not fit you.
• Sector. Money is concentrating. On Carta in Q1 2026, AI companies captured more than 60% of venture funding (Carta, State of Private Markets Q1 2026, published 2026-05-29, https://carta.com/data/state-of-private-markets-q1-2026/). A sector investors are crowding into prices differently from one they are avoiding.
• Your evidence. Revenue, a repeat founder or a competitive round changes the conversation. A median carries none of that.

The unknown that matters most is what your target investors signed at your stage in the last two quarters. No public report has that.

1. Write down the milestone, the money it needs and the share you are willing to sell across all SAFEs before the priced round.
2. Divide to get your cap range, and find your band in the Carta table.
3. If you are far above the band, have a reason an investor will accept. If you are far below, ask why you are discounting yourself.
4. Ask your target investors what cap they last signed at your stage. If they object to yours, ask what drives it: the price, your traction, the ownership they need, or whether you fit what they invest in. Each answer points to a different move.
5. Have a startup lawyer read the actual SAFE before you sign. This page is education, not legal advice.