A cap table (capitalization table) is the record of who owns what percentage of a company, on a fully diluted basis. It lists every share, option, warrant and convertible instrument, and shows what each holder owns today and what they will own after conversion. It is the single document that answers "how much of this company do I get for my money?"
Most founders can produce one. Fewer can defend one. The gap shows up in diligence, and it is usually not the spreadsheet that is wrong — it is the founder's understanding of what they already sold.
This piece covers what belongs on a cap table, the 2026 ownership benchmarks, the arithmetic of post-money SAFE stacking, and the specific errors that slow a round down.
What a cap table contains
| Line | What it represents | Counts toward fully diluted? |
|---|---|---|
| Common stock | Founders, early employees who exercised | Yes |
| Preferred stock | Priced-round investors, by series | Yes |
| Option pool — granted | Options issued to employees | Yes |
| Option pool — unallocated | Reserved but not yet granted | Yes |
| SAFEs and convertible notes | Not yet shares; convert at a priced round | On an as-converted basis |
| Warrants | Usually lenders or advisors | Yes |
The distinction that causes most confusion: issued versus fully diluted. Issued shares are what exists today. Fully diluted includes everything that could become a share — the unallocated option pool, every SAFE, every warrant.
Investors negotiate on fully diluted. Founders often think in issued. That single mismatch explains a large share of the surprise in first-round diligence.
2026 ownership benchmarks
Carta's Founder Ownership Report 2026, covering rounds raised 2021–2025, gives the median trajectory:
| Stage | Median founder ownership (fully diluted) |
|---|---|
| Post-seed | 56% |
| Post-Series A | 36% |
| Post-Series B | 27.3% (AI) · 21.8% (non-AI) |
| Post-Series C | 16.1% |
At Series C, median founder ownership falls below the employee equity pool for the first time.
Sector matters more than most founders expect. At Series A, founders in digital industries retain 37.5% against 30.5% in physical industries — a seven-point gap for the same stage.
Per-round dilution, from CRV's 2026 equity structure guide:
| Round | Median dilution |
|---|---|
| Seed | ~19% |
| Series A | ~17.9% (down from 20.9% a year earlier) |
| Series B | ~13% (2025) |
Two figures from that same guide are worth holding onto. The seed-stage option pool typically runs 10–15% of fully diluted shares. And over 70% of equity financings include an option pool top-up — meaning the pool is refilled at the founder's expense as part of the round, not the investor's.
For context on what is being priced: median post-money seed valuation reached $24M in Q4 2025, and median pre-money Series A approached $50M by late 2025.
How post-money SAFEs stack
This is where cap tables break, and the mechanism is not obvious.
Y Combinator publishes the post-money SAFE, released in 2018. Three variants exist for US companies: valuation cap with no discount, discount with no cap, and uncapped MFN. Companies in Canada, the Cayman Islands and Singapore get the valuation-cap version only.
"Post-money" means the investor's percentage is fixed as a share of the company after all SAFE money, before the new priced round. YC's stated advantage is precision: founders and investors "calculate immediately and precisely how much ownership of the company has been sold."
That precision is real. The trap is what it implies.
Under a post-money SAFE, each new SAFE dilutes the founders — not the earlier SAFE holders. The first investor's percentage is locked. The second investor's percentage is locked. Every lock comes out of the founders' column.
The worked example
Three SAFEs, none of which felt large at the time:
| Instrument | Amount | Post-money cap | Ownership sold |
|---|---|---|---|
| SAFE 1 — pre-seed | $500,000 | $8,000,000 | 6.250% |
| SAFE 2 — bridge | $750,000 | $12,000,000 | 6.250% |
| SAFE 3 — extension | $1,000,000 | $15,000,000 | 6.667% |
| Total | $2,250,000 | 19.167% |
The largest single instrument sold 6.67%. The founder raised $2.25M and gave up 19.17% before a priced round had been discussed.
Now the Series A arrives: 20% for the new investor, option pool topped up to 12% post-round.
| Holder | Post-Series A |
|---|---|
| Series A investor | 20.00% |
| Option pool | 12.00% |
| SAFE holders | 13.03% |
| Founders | 54.97% |
Note what happened to the SAFE holders: 19.17% became 13.03%. They were diluted by the priced round like everyone else. The founders went from 80.83% to 54.97% — losing 25.86 points to a round that "only" sold 20%.
And this is the clean case. Carta's median founder ownership post-Series A is 36%, not 55%. The example above assumes one priced round, no pre-seed equity, no prior pool, and no bridge before the A. Real cap tables have more events, and each one compounds.
The misconception worth correcting
Splitting the same money across more instruments does not reduce dilution. $2.25M raised as five SAFEs at an $11.67M cap sells the same 19.29% as three SAFEs at the same cap. What drives dilution is the ratio of money raised to valuation cap — not the number of instruments.
Raising more on the same cap costs ownership. Raising the same amount on a higher cap costs less. The instrument count is noise.
What an investor actually checks
Having sat on the other side of this, the diligence pass is narrower and more mechanical than founders expect. Four things get checked, in this order.
Does the fully diluted total reconcile to 100%? Sounds trivial. It fails more often than you would think, usually because the unallocated pool is tracked in a different file from the granted options.
Do the SAFEs have a single source of truth? Three SAFEs signed at three different moments, stored in three different places, with caps that may or may not have MFN clauses pointing at each other. If the founder cannot produce one schedule listing every instrument, its amount, cap, discount and date, the round slows down while a lawyer builds one.
Does the option pool match the hiring plan? A 12% pool with no plan attached is a number. A 12% pool with a headcount plan showing the grants it funds is an argument. Investors ask which roles the pool is for, and the answer reveals whether the founder has thought past the round.
Has a 409A valuation been done, and when? Option grants priced off a stale 409A create a tax problem for the employees who took them. It is not the investor's problem, but it signals how the company is run.
None of these are hard. All of them take time you do not have when a term sheet is live.
The two errors that cost the most
Negotiating the pool without negotiating when it is created. An option pool established pre-money dilutes only the existing holders — the founders. Established post-money, it dilutes everyone including the new investor. The percentage is what founders argue about. The timing is what determines who pays for it, and it is barely discussed.
Treating the cap table as an artifact rather than a model. A cap table that only shows today is a record. A cap table that models the next round — with the pool top-up, the SAFE conversions and a range of new-money percentages — is a decision tool. The second version tells you what a term sheet actually costs before you sign it. The first tells you afterwards.
That distinction is the same one that separates a financial model built to raise from one built to run the business.
Frequently asked questions
Clear answers to the questions founders ask most often.
A shareholder register is the legal record of issued shares. A cap table is a management view that includes issued shares plus everything convertible — options, SAFEs, warrants. The register is what exists; the cap table is what will exist.
At incorporation. The cost of building one retroactively, after three SAFEs and two advisor grants, is measured in legal hours.
They should, on an as-converted basis. A cap table that omits outstanding SAFEs overstates founder ownership by exactly the amount that will surprise the founder later — 19.17% in the example above.
Roughly 19% at seed, 17.9% at Series A, 13% at Series B on 2025–2026 medians. Ranges are wide, and the option pool top-up sits on top of these figures in more than 70% of financings.
Not the valuation itself, but it determines price per share, and it is what a business valuation has to reconcile to. A discounted cash flow gives you enterprise value; the cap table is what turns that into a number per share — and the discount rate you use is a separate argument entirely.
Every share that would exist if every option were exercised, every SAFE converted and every warrant called. It is the denominator investors use, and the one founders should use.
The short version
A cap table is not a compliance document. It is the arithmetic of every financing decision you have already made, and the only honest preview of the next one.
Three SAFEs that each felt like 6% can be 19% together. A 20% Series A can cost a founder 26 points. An option pool created pre-money is paid for entirely by the people who already own the company.
None of this is hidden. It is all visible in the arithmetic, in advance, to anyone who models it before signing rather than after.
Sources: Carta, Founder Ownership Report 2026 (rounds raised 2021–2025). CRV, Startup Equity Structure 2026 Guide. Y Combinator, SAFE documents. Worked example calculated 23 September 2026.